AI AT SCALE STAGE

Joining Jason on stage for our next panel discussion is the Head of Product at Glean, Amrajan Doan, the Chief Product Officer from Rubrik, Anika Gupta, the Chief Product Officer from Webflow, Rachel Wholen, and the Chief Product Officer from Harvey, Aneek Drumwright.

All right. Can you guys hear us okay? I know folks are streaming in. It is funny at events like this where 9 is early. Um, we'll go through the whole panel and we'll find out how, how early they let their teams get in, um, in the era of 996. So anyhow, this is an exciting exciting session. We banned panels at SaaStr for a year because they're boring and people talk in a line and it's nothing. But if folks are friends, it's pretty interesting if they know each other. So that's the only exception we make. So I think we have, we have a special panel, I think, of friends. And we have what I think is the single most stressful job today in B2B is being a chief product officer. For, I think for 10 years it was a great job. You'd bring your mug to the office and you'd tell your team, what are we going to ship this year? And you'd kind of— you'd do some Kanban and you'd spread what you're going to build. We'll push this one to next quarter. Jason wants this. Let's just tell him it's coming, but it's never coming. That was like the job through 2024. Okay. Honestly, you think I'm kidding? But now we have this. I think is the hardest job because everyone here knows we're under the gun to ship Agentic features and ship AI. And it can't just be for fun. It, if you'll, and we have leaders here. We have Rubrik, one of the most successful IPOs in security the last 12 to 18 months. We've got Webflow gone through a massive transition, like the, the, every designer loved building on Webflow. Um, now what does that mean? We have Glean, um, which is one of the early breakout leaders in AI for managing data inside your company. And we've got Harvey, which, you know, comes outta nowhere to 9, how much did the valuation go up today?

Uh, no, but it went up.

It's still just at $9 billion. Okay. I think it's at $11. What was a small sleepy market and disrupting it. And so it's great, but it's so stressful because like if you're, if you're even slightly older, you need to get an agent that, that, that will, where you'll own the market and not be disrupted. Even if you come out of nowhere like Harvey, Anthropic's got new features that the amount of innovation in legal is like 50 times what it was 3 years ago, I would say. So, This went from a great cush job with a big team. I think, I'm being facetious, would you agree, maybe the most stressful job because everyone's like, what are we shipping now? I need, and I don't just need AI, I need something I can monetize. I can monetize. And if you look at the public stocks, there's a group, and we'll talk about it in the next session, that have gotten a big boost in the market. Atlassian just launched, they finally were able to really monetize their AI agent and it's crushed. Stock's up 30%. Others are still figuring that out in the journey. So that's my rambling preamble to from the cushiest to the most stressful job. I mean, a little facetious, but I genuinely believe it. So we're going to do something. We're going to, we're going to, because we've got a great group, we're going to, we're going to first talk about what everybody's building today as an agent. We're going to talk a little bit more about what agents really mean. If you were here when I was with Amjad, the CEO of Replit, yesterday, even he said it's about the loosest defined term there is. We're gonna talk about that. Then we're gonna do some quick demos. And then we'll go through a few other things, and if we have time, we'll take questions for folks. But, let's talk. I think I may be first, but what do you, what's the latest? And the demo, I'm gonna skip ahead, but the demo for Rubrik's super interesting because in security, it's not like I can hack anything I want, it's faster. We can spin up a parking pass agent like we did for this event 3 days ago, And it like, it's not that much sensitive data. It's okay. You know, Rubrik, it's not okay. It's not, it's not okay for your agent to accidentally leak, leak some sensitive data or to create a, create a little bit of a hole in your application. So what's the agent you're allowed to build and that you're building?

So when we look at our product, so Rubrik is in this space of cybersecurity, but we really focus around cyber recovery. So how do we help organizations recover their data applications identity after they've already been hacked? And we occupy this place where our customers are relying on us to provide this core service. So nothing we can do can take the core service down. And when we think about delivering agentic outcomes, we're focused on, well, how do we take the amount of time that someone is spending managing our services and our product down to close to zero, where they just say, hey, let me define my outcome and we'll, or let me define my intent, will deliver the outcome. But the challenge with it is how do we do it in a way that is super, super high accuracy and where we're preserving that trust with the customer? Because if we give them the wrong recommendation or we take an action on their behalf that impacts their production applications, then they're no longer gonna wanna be a customer of Rubrik. So that's been the focus. It's been actually like a much more challenging problem to build agents within our product than I think I thought it was going to be a year ago to really deliver those outcomes. Outcomes.

Yeah.

All right, then Rachel, so Webflow's gone through, I mean, Webflow pre-AI was the designer's choice. It was the cool kids' tool, right? Um, and, uh, maybe it still is today. I may be dating myself, but the whole world's changed how we create web products, websites, web everything, right? It's, it, it, I mean, we vibe-coded millions of new websites probably in the last year. Um, I know more databases have been built by agents than humans now, right? It's probably true one degree away for websites, right? More agentic products have built websites than humans even today. So, um, how has it changed? And what, what's, uh, what's, what's the latest agent or agentic product you're working on?

Yeah.

So Webflow is an agentic web marketing platform. As Jason mentioned, we started off as a website builder. It was very humble beginnings.

Yeah.

And what we've found is that it's very hard in the enterprise to actually build agents to keep up with what is happening to your website, which is, you know, Perplexity, Claude, ChatGPT, you know, whatever DuckDuckGo is trying to do. And there's lots of complexity in trying to get discovered. So we've really focused on how do we help your brand get discovered and then how do we optimize for your MQL. And that is a whole other set of technology that we've built. And then, you know, we assume that websites are gonna get built through LLMs. We have, you know, thousands and thousands of agencies that use our platform with our MCP server, with our APIs, and we've made ourselves completely headless and visual. So you— an LLM wants to be able to both understand the context as well as an LLM can see now, right? And so we've really focused on how do we become as accessible to LLMs as possible? And then how do we help you, a brand, build a long-running agent that honestly just reduces the complexity and shares your brand context, your voice, your tone, your design system with an agent?

Got it. Um, so, so lo— and just, just to help, um, so in Webflow, what does a long-running agent mean? What does that mean?

Yeah, so today what I'm gonna show you is our new answer engine optimization optimization agents. Yeah, that helps you automatically detect when you need to upgrade your technical AEO, for example, like your schema markup. That helps you look across your site and understand what competitors are doing, and then recommends different content that you should be producing, and then actually goes and drafts it for you. And then all you're doing as a human in the loop is approve, approve, approve, disprove.

I see. And so that context window basically needs to remain open indefinitely. Is that why it's definitely—

it's You, you want agents that are just constantly going and, you know, basically protecting your brand and helping you be surfaced inside of answers, inside of answer engines.

Got it. Okay. So, so Glean is, is interesting, right? When we, when we look back on the old days before Claude 4 or 5, remember back then? It was a while back. I mean, Glean was one of these ones that broke out when AI started to break out, right? It's all— I mean, I'm not a total expert. But it solved a big problem. How do I access internal information and others elegantly, right? Whether maybe today we call it a prompt. I think even when Glean took off, we weren't even calling it prompts. It was just search, right? And how do I— I have this trove of data. I can never find it. But man, I can only imagine what customers expect of you in 2026 is an order of magnitude more than even like Q4 last year, right? So is that right? And what does that mean for your very latest agent?

Yeah, love it. So Glean started about 7 years ago. So when retrieving information, retrieving context was the key jobs to be done, like it was the most visceral to become a better knowledge worker. That's what you needed.

Yes.

So like fast forward to today, retrieving information, retrieving knowledge is part of the job, but the whole world is moving from like getting informed. To actually performing, like getting AI to perform for you, etc. So what is not— a lot is changing there, but what is not changing is the superior context that you have, all the tested knowledge, idiosyncrasies, specific unwritten, like, understood information that you have in the company. The better you establish that context, then you can get the perform, perform piece done by AI rather than you as All jokes. So, and it's not really sold as much on the context, you know, I talk to like I use a variety of AI front doors including Glean Cloud, etc. But most folks I talk to, if they are spending, let's say, 5 hours per day in front of an AI assistant, they are spending half of that time in quote unquote building context. Hey, I am feeding documents, I am feeding my decisions, my patterns, etc. So the thinking here is for AI to operate at its max, then you need— like, you shouldn't be actually spending as much time building context, building skills, updating your memories, etc. So that's where Glean comes in. On one part, you can use Glean as the AI assistant, as the front door, but also Glean actually serves the system of intelligence as a single MCP server. If you want to use your own favorite assistant like Cloud Code, Cursor, Codex, where instead of tapping into individual MCPs, like these harnesses would tap into Glean to get the full and superior context to get the job done there. So Glean is now operating on these two angles. You can use Glean in the UI or you can use Glean as a system of intelligence.

What percent of folks are using Glean in like cloud today instead of, instead of using the app?

It's a— I mean, if I bundle all of the cloud Cursor, Codex this model, it is one of the fastest growing parts and definitely faster growing than the, the UI of Gleam, which is still by a large margin a very high daily engagement driver. But the interesting thing is the more that usage happens in Clouds of the World, it actually has spillover effect back into Gleam. And then we'll get into some examples of it, like how do you build agents that are not just for yourself but for your entire team, for your entire department, etc. So this kind of an unexpected spillover effect that brings users back into other Glean UIs or assets or agents to use them in different parts of their work life.

Well, that's— I didn't notice that in there, but let's come back to that before the session because that's— and is, so in a way also if using Glean in Claude and others is blown up. It's almost like a super memory too, isn't it? I mean, there's only so much data you can access inside of the LLM, right? It's not just MCPing into it.

Exactly.

It's taking, it's taking the LLM and giving me memory across everything across my organization that's relevant to me. Right.

And the key thing is MCP, as much, as much independence as it brings, it's a very runtime fetch of information. You are bound by either latency or you are bound by the search APIs under the hood provided by these teams. So it's a very different paradigm when you rely on high compute offline processed information of connecting the dots between different employees, teams, projects, subsidiaries, organizations, companies you have, you might have acquired over the last decades, etc.

Yeah.

And processing them offline to actually develop a superior context and understanding of the real-time pulse of the organization. They are very different paradigms and they can be actually very much supportive of each other. How do you use real-time information in runtime with superior offline processed knowledge or intelligence?

All right. And so, Nick, are you— so are lawyers asking for agents?

Yes.

Your average lawyer that's not at the bleeding edge, that's not here today.

Average lawyer. Look, I think the legal practice— so where does Harvey serves in like the mid-market enterprise. We're in 60+% of AmLaw 100. We're in Fortune 500. You're basically talking about like elite athletes, or like the lawyers at these law firms.

Yes.

And like, you know, the internal teams working at these large corporations. Like, I— it's been a really fun like customer to have because they are super intelligent, incredibly curious. They there's like a real competitive spirit to what they do just inherently, right? Because they're, you know, selected for their judgment, their strategy, how rigorous they are, how right they are.

Yeah.

And so I think because of that, you just have a very curious base. And I would also say that it's— so when I joined, I probably talked to like— I did a listening tour with like 60 customers and they were like—

Before you started?

No, my first 6 weeks. I was like pretty militant.

I was going to give you the best diligence process.

I know.

Taking a role ever. We're going to make you the hero. The poster child.

Proving out that we're hard workers.

Okay.

But at least you did it right after you started, right? That's my number one thing for a head of product. If you're not out talking to customers, show them the door, right?

Yeah.

But one of the things that was really interesting is it was like, you know, when we first started partnering with Harvey, it was because I wanted to have of, you know, the lawyers on my team using AI safely, using the right knowledge sources, using LexisNexis, you know, being able to have rich citations, all of these things. And now, like, fast forward a year, and yes, like, I'm talking to litigators that, like, want to talk to me about our Agentic capabilities and how should they be leveraging it, how should they be using with, like, their team of associates, etc. And I'm not saying that 100% of our customers and end users are there. But do I think that there is a real curiosity and demand for agentic work to support them? Yes. I would also say there's a real demand for how do I leverage agentic work really safely? So what is the verification process? Can I actually verify someone else's agentic plan?

Right?

So if I have a more junior lawyer supporting me, Can I verify their work?

Right.

Let's walk through that because that's interesting. Right. So you've got, you've got, you've got a partner reviewing an associate's work or what's the workflow there or what's the agenda?

That's an— that's a simple workflow.

Yeah.

Right. And so if I'm right, even right, right now, Harvey is partnering with schools of education, right, with law schools. My cousin is in law school at USC. It is incredibly top of mind on how do you grow as a lawyer, how do you become dominant in the field and learn the practice while also leveraging AI. Yeah. If I am an equity partner at a firm, how do I coach up-and-coming associates on how to leverage AI, how to leverage it safely, how to verify? But specifically with the workflow we're talking about, it's okay, so I'm getting the work output from my associate. Do I have the opportunity to not only look at what the prompt was from the associate, but can I also look at the plan? And so it's a really interesting problem.

A plan for like a legal brief or for a document? Like, because there's a lot of different legal matters, right?

When I say a plan, I mean an agent plan, right? An agent is just the way that they are in coding models, right? They're receiving a legal prompt, they're writing a plan, that plan is, you know, being approved or customized. And do I agree with that process that ultimately created the output, and how from a product perspective can we make that loop very efficient? How can we make it collaborative? And like, I wouldn't say we're doing all of it perfectly today, but is that the direction we're going in terms of agentic work? Yes. So get going back to your question, do I think lawyers are, are asking to leverage agents? Do I think there's a real appetite for this?

Absolutely.

I actually would say the conversations are one step further, which is let's assume we're using it. What does it mean to collaborate with it? What does it mean to coach with it, performance manage with it? What does it mean to verify? And it's—

yeah.

So anyways, that's my take on it.

So just, just two things. I don't want to do— I want to move on to the next topic, but you've got senior partners at law firms, one of the last two bastions of the ThinkPad and the red nub, orange, whatever color it is. This is an accountant's They've got— they're stuck in this world. They're sort of— they're very smart and very intellectually curious, but not necessarily at the bleeding edge of technology. They are today creating prompts for their— having their associates draft prompts that they then review and then edit and commit to an Agentic workflow. I'm just— that is such— it may partially explain the massive disruption in the legal industry, but that is such a different change from the dictating phone, right?

Firm to firm, practice to practice. Like, these are unbelievably prestigious firms, right? And the legal practice is so respected, so important. And so I would say that like everyone is finding their own path as to what this looks like. But is there appetite for this capability? Yeah, so when something comes up around diligence or discovery, and AI is actually really good at finding, you know, those needles in the haystacks, reviewing a wide volume of documents in a very consistent manner, flagging that for then human-in-the-loop review. Yeah, do I think there's appetite for that and an acknowledgement that the tools to be really— continue to be excellent at your job have evolved? The speed at which you can now deliver those outcomes has probably become much faster because of AI? Yes, I do. Again, I think we're— personally, I think we're oversimplifying the persona. I think it is, you know, these partners at law firms are running like huge businesses, right? And how they're choosing to leverage AI is partially at their discretion. And what I see my role as is, okay, can, can we listen to what the needs are and can we create the tools that make you under— make it easily understandable? Where when AI is a really great tool, when an agent's a really great tool, when you then need to verify what an agent has done, is that accessible? Is it easy to do? Are we investing in citations? Because that is uniquely important to legal. So yeah, that's my take.

Yeah, yeah. All right, well let's, I wanna do our demos and then let's talk about some other issues, especially like around guardrails and how to manage these agents. I just think, I think the legal challenge is even, I think what Harvey's done is even more impressive than others because forget about that it's legal, security's harder, even harder to sell into legal, I would argue. But partnerships pay outta their own pocket. Let's not, when you sell to partnerships, or private equity or venture capital for— like, they're reaching out. It's one thing if I'm at Rippling and I got to get a few bucks from Parker. I mean, maybe he's, he's cheap, okay?

But like, have you tried to do that with Parker?

I know him a little bit over the years. I'm being— I mean, I know you have to, you have to argue for it, right? But ultimately, you're not taking money out of his personal pocket. When you go to a law firm and— what, I don't know what the biggest Harvey deal is. Let's imagine it's $20 million, $30 million a year, okay? That's got to come either from another substitute tool, like I'm taking that out of my other budget, right? Or the partners have got to say, this is why it's so impressive. There's so much ROI here that I'm going to pay $50 grand, $100 grand out of my own pocket when I kind of want to use it on the house in the Cape. I was kind of looking at that Roma Ferrari, and I know I'm not supposed to drive it to the office, but maybe in the Hamptons I could. I mean, I'm being facetious, but this is traditionally one of the many reasons technology has been sold to law firms is there's so many things in theory you could buy But I gotta, you know, you wanna pay for yourself. So I just think that it's more impressive the rate that it's scaled than a budgeted technology budget where it's just a line item for the CIO or the CFO.

I also think realistically there's also customer demand, right? If I, you know, let's take for example, I wanted someone to review an employment contract for me. I would probably expect them to partially leverage AI at this point. Point. And so I think it's also like there's just this massive shift in society that no one is protected from, right? Because us as consumers are now expecting. And when I say consumers, it could be a company or it could be like I as an individual user.

Okay. All right, good. All right, let's see this. We'll have to knock on wood on this, otherwise we'll hack it. But if you guys are there, can we try to roll the rubric demo? The rubric one, I think Emila cut out 3 minutes, but it is interesting because it's got to work, right? It's got to be safe and secure. So do you guys have it? If I'm facing east, let's see.

Okay. Yeah, great. So we're going to pull up this, what we call Ruby, which is our AI agent for operating our platform.

I love Ruby's little motion.

Oh, thank you. Yeah. Thanks to our design team. So when we think about the future of Ruby and where it's been is like we started this system as essentially like a simple RAG application where we fed in all of our documents on how to set up the product, how to troubleshoot, and people could ask questions and get, I would say, like, you know, okay answers based on whatever we had in our documentation. What we have worked on over the past year to fundamentally change is truly to make it agentic. And what that's meant is like, you know, I'm not gonna go into all the details of this, specific prompt here, but for, for example, for in this case, we're saying, okay, our, our customers, they are growing their data with, with Rubrik and how much we're data we're protecting for them, and they need to do forward-looking capacity planning. This is a task that would typically take them at least a day, if not longer, to piece together all the information to actually do this capacity plan. And what we've built is agentic workflows that take all of the insights that we have from their systems, marry that with our understanding and domain expertise of this space, and be able to provide them high-quality answers and eventually take actions on their behalf. So there are more examples here that we're, that are being kind of going through quickly of custom reporting, other, other things that people need to get done through, through the application. and our vision here is that eventually you can operate all of Rubrik through this interface, that you do not need the UI and dashboards at all, that it will be a fully agentic experience. And the way that we're, we're going about this is we had to build a platform that allowed us to scale to multiple use cases. We didn't wanna say, okay, we're gonna custom build all of these different workflows and anticipate the workflows that our customers are gonna have. We are going to build a platform such that it's generative and it can work in any kind of workflow that our customer wants, from onboarding to troubleshooting to planning, etc. And what we're envisioning here is a truly like an experience that is similar to Claude Co-Work, right? Where you can essentially say, hey, this is my intent. We come back with a plan and you can edit that plan and then you execute and then we'll execute that on your behalf. And what we're doing from a UI perspective right now, it still looks very much like chat, but where we're going is we're actually gonna pull some of the UI elements from the dashboard. So not everything is best represented in text. You want graphics, you want tables, you want other kinds of elements. So we're looking, we're building a framework where we can actually pull in those elements as needed to solve for the workflows. And then also looking at ways to actually actually represent these workflows so that a customer can come in and understand what has happened and go back and say, "Yes, this is a repeatable thing that I wanna do again," or, "I wanna tweak this for the future and then automate this." And trying to bring together both determinism, which is important because in a recovery scenario, you don't wanna be guessing and you don't wanna be using a probabilistic mechanism recovery. So instead what we're using is we're using like all the, the LLMs to help come up with the plans. But the plans themselves of how you're going to do a recovery in a cyber scenario are deterministic and then are explainable to the customer. So that is how we're building our agents today. And it's been a very interesting journey of figuring out what is the right architecture to make this possible. How are we building the evals around this, especially because we have, we want to basically everything that's possible that's done through the UI to be done in an agentic way. So how do we do that? How do we democratize this so that we don't have a central team that is building all of the use cases and optimizing all of the use cases, but every PM and every engineering team within our company is thinking from an agent a security-first mindset in, in Agentic workflows. So it's a work in progress, but it's been a really exciting time because for security teams, security teams are very worried right now, right? We've seen the rise of Metasploit and just a lot of concern around how many vulnerabilities are now being like brought to light and can people truly patch and patch the vulnerabilities fast enough? Fast enough, which the answer is no. So they're on high alert. They're very interested in basically any solutions that can run at machine speed because the attackers are running at machine speed now. And so it's a huge challenge. And they're very worried about, hey, we have only a certain amount of budget. We have a certain amount of people. We can't do everything with people anymore. We have to figure out how to leverage agents. So we're seeing a lot of pull towards agentic solutions. But also caution because they are not willing to kind of cede control completely to agents for their, their core infrastructure and core security systems.

So let me ask you just one question on this. We can look at the Webflow dome. So you're moving to an interface where I can chat and talk to everything in Rubrik, which I love. I just, I can't learn another UX. Yeah, I'm just done. I can't do it. So I love that you're pulling all the data so that it, so that I don't have to have an analyst to figure this out. I can get access to any data. You're just beginning making that actionable through an agent. I could ask a lot of questions, but one is, as you spin up these agents to take core actions, right? To address load balancing or load planning or other pieces. Yeah. Will the customer do it all? Will you have forward deployed engineers help them? Will partners help them? Because they want that agent that's gotta work, right? Yes. And so what is the human agent interface or plan to roll these out?

So yes, there has to be some human in the loop to, ideally where we want it to be is that the human in the loop is approving the plan. So they're approving saying, hey, this is what, like we say, hey, this is what we're gonna do. Do you agree that this is the right thing to do to achieve your outcome? We're also working with partners. We work with a lot of like GSIs and other kinds of partners who are delivering a larger program, especially to like Fortune 2000 companies and companies. And in that case too, we're saying, well, the partners also want to be able to automate things. So can we deliver this experience to the partners?

Yeah.

And they're actually operating this on behalf of the customer, but they don't have to take— they don't have to do a lot of the guesswork. They don't have to call in.

Okay. And then we could spend the whole session on this relationship. So the GSIs could help plan the agents. Who's responsible for the agents? When it's just a human taking the action in Rubrik, the human's respon— I mean, if Rubrik has a bug, that's on you, right? But if the human makes a mistake, it's on the human. When you have these agents out there, this is not Saastr spooling up a parking pass application. Like, who's— is it the GSI?

I mean, at the end of the day, we're still responsible. Even when a customer makes a mistake and something isn't working, they blame us, and rightfully so, like they should. It's our responsibility. Responsibility to make it obvious what the right choices are. And today people do make mistakes because there's so much complexity in setting up our product. And even though we have constantly sold on a message of simplicity and we have beautiful design experience and all this stuff, um, it's still— people can make mistakes. And people are going to make mistakes more than agents if we build our agents with the right guardrails around it. And so our belief is that actually we'll make fewer mistakes in an agentic world, we'll tell people what the right solution is. We'll keep— take more of the guesswork out of the equation, which there's a lot of guesswork, a lot of different ways you can configure things, give them best practices, um, and that will actually yield better outcomes for them.

I got it. I just think the, um, let, let's get ready for the Webflow demo next. I just think the, um, your job's gonna get more fun because I think the complexity's gonna go up an order of magnitude.

Yes.

When these agents are out, I mean, We literally had the CEO of Replit with me yesterday. Even him and his team, we build things on— we're in the top 0.1%. We build things they didn't quite know were buildable on Replit, and it's exciting. But that means also folks are going to do things with Rubrik's agents you didn't even know they were planning to do, and you didn't— in some case, you might not even know were totally possible. And you're responsible not just when the core button doesn't work, but when millions of agents are— I mean, it's, uh, it's We just got to get over the hump of having those agents effective. But their level of responsibility, I think, is something we haven't— at a product level is going to be stressful. Yeah, right. It's going to— you're going to be responsible for everything and you're not going to predict what your customers are going to do with it.

You're not going to know, especially when you have both headless experiences as well as like—

you don't know what they're going to do with headless, right?

Yeah, you don't know, but you have to offer that because it's super powerful.

All right. If we have the Webflow demo, can we, can we roll this one?

Great.

So Webflow is an agentic web marketing platform. We have about 18% of the top 2,000 websites on Webflow. And the reason why people choose Webflow is we help you span from SEO, the 10 blue links on Google, all the way to AEO, which is, you know, being ranked for answers. In ChatGPT, Perplexity, Claude, etc. And the reason we know this works is we've already shipped some of the kind of like human automation of this. And we saw customers have a 75% bump in their organic traffic when they started using our technical AI AEO automation. And so we took that and said, hey, what would this look like as a fully autonomous loop? And so what I'm going to show you today is how Webflow helps you measure AEO. We autonomously make recommendations and then act on it with a human in the loop. So what we're gonna show you is first, uh, we, we actually have a lot of, uh, data that only Webflow has access to because we have access to server logs. We can look at what bots are showing up on your site. We have, we categorize all LLMs as well as all categories, knowing if they're coming for inference, they're— it's an agent coming to your site, if it's a, uh, if they're, they're doing a training. And so the key is that you can actually take that data and uncover nuggets and be like, oh, this blog post actually showed up in training more. This blog post is getting shown up more for citations. And then we actually surface the actual prompts that, uh, you're showing up for depending on what type of, what type of bot was coming. And so that's kind of the key to really understand how to fine-tune and kind of hone your brand and how it's actually showing up with answer engines. Now, when we started this, you know, like a year ago, maybe 10% of answer engines— of visitors were showing from answer engines. Now it's more like 50%. So if you are not showing up to answer engines, your, your brand is completely invisible to your potential buyers. And so what we've also thought about a lot is having a human in the loop.

Mm-hmm.

So again, Webflow, you know, has a wide variety of customers, a lot of non-technical users who want to see what changes are going to happen. So in this case, for example, the actions being recommended by the agents are to change meta descriptions, to change schema markup, to change alt text. These are all kind of like technical AEO recommendations. And then what you'll see is you can actually go in, review those changes, and auto-accept them. This is running across your site autonomously, 24/7, uh, and then you actually get to go and accept those changes. Now the, the key here is that this is constantly updating and something where we can continually add more and more agents in. So we've also built out a full agent builder. It's completely customizable. And it's something that, you know, this is kind of the future of what Vibe Marketing is, being able to completely access your agent headlessly, build it headlessly, but also be able to see visually WTF is happening. And so I think that's the key here is you want actually the human to be able to visually see a change, but also the agent to be able to act autonomously.

So the agent is updating the AEO daily or constantly. You can accept or unaccept. Just one question I think a lot about, and then we can go to Glean. So AEO is very interesting. How can you make sure when someone goes into really just Claude and ChatGPT, but others, maybe Gemini and others, what should I buy? What should I use? Super important today. What I think, is gonna be even more important is its cousin, and I don't know what it's called, is agent recommended applications. We only buy now for ourselves because we're 21 agents and 3 people, we only buy the products that our agents, 10K, our AIVP marketing, our AI, we won't buy, I don't care what Claude says anymore because if my agent, we switched email to Resend, okay? I don't know what Claude would say, I don't care, my agent said use Resend for email marketing, that's what we use. Jason said use ElevenLabs. And so it's not even coming up first in Claude, it's being the default choice of whether it's the public agents, whether if, if, if Harvey or Gleaner Rubrik exposed this, like let's say I'm in the Rubrik agent and, and the agent probably can answer this question if you don't have the guardrails, what should I use? Like you wanna win that race. So, which is even more than AEO. AEO is still human to, to LLM interaction, right?

It, it is and it isn't. I mean, I think, I think you really wanna show up in what I'm calling agent-led growth.

Yeah.

Right?

And so I think that, you know, you're gonna have an AX, an agent experience.

Yeah.

You're going to have agent-led growth and you're still also gonna have humans that are in that loop in some way deciding which agents, you know, I don't always say yes, I wanna use Resend or whatever is the, the default. A lot of times, you know, you really need to make sure that you're your product is showing up in the right way for the consideration set that you want to show up for. And so that if you, if you are not really focused here, and again, we, we really think about our agent builder as being an agent interface to LLMs, right? And so that can show up in many, many different ways, whether it's cloud code, whether it's in cloud, what, you know, lots of different ways in terms of how training and inference are happening.

Yeah, I'll just give one fun example. I want to go on. So like, we're leaving I can't believe we're still on Marketo, the world's oldest piece of mark— but I can tell you why we're still on it. We were because we were prisoners. We couldn't get— before LLMs, we couldn't get our data out. We had 200 segments of a list of 450,000 people and there was no way to get it to leave. Now it's pretty easy. But we asked Claude what we should do and Claude was like, well, Marketo is pretty old, but it's good for enterprises and older companies. We asked 10K or AV marketing. It gave it an F. It's like I hit API limits in an hour. Whatever you do, do immediately leave. Like, 10K was so aggressive that Claude was— gave it— it still came up as a mediocre opportunity, but the agent was like, unacceptable. It literally said, F 10K, F does not work. I hit my API limits this morning. You cannot— slightly paraphrasing them, but it gave it— I literally never seen it. F, right?

Well, and that's like the DX, right? That's the developer experience.

Yeah.

Also matters. So. We're kind of in this role where you have to optimize UX, DX, and AX.

But they're blending.

Yeah, and they are blending in a lot of ways.

OK, let's see what Gleam demo we've got. Sorry, I should have given you a quick heads up.

It was lovely while it lasted.

We have a few extra minutes. OK, let's see what we're seeing.

So I think the video started, so I'll narrate a bit. So what we are looking at here is not the user, the consumer experience. We are actually looking at— the builder view here. So, as an agent builder on Glean, to the top left, the title would say, hey, there are two sales systems, like systems of record. One is Gong, where for, for Glean sales team, in any given day, there might be somewhere between 300 to 800 Gong calls being recorded with AEs doing like net new sales, renewal conversations, et cetera, right? Then there is amazing insight, net new information coming in. So then the notion is, hey, how do we get all this key information into another system of record like Salesforce with the right details and then write, like maybe get the agent to write into Salesforce, fill in or update the necessary fields, etc. So now what we started doing here is an agent instruction is in the middle column. We might get back into it, but on the right is the builder. Now I am actually testing the agent with just one call. I dropped in a Gong call recording. Mind you, when I deploy this agent, I might not actually use it as a human to come and say, hey, here is one call, do it. I might actually run it as a departmental agent, put it on autopilot to say every time there is a Gong call, run this agent and then let the AE know that agent is recommending to update this. I might even go bolder to say, take humans completely out of the loop and then run it on full autonomy. That every time there is a Gong call, then do the analysis with this full context and then update Salesforce autonomously, right? So going back to these notions of agent identity, what is the cost of doing something wrong, et cetera, we'll get to it. But right now we are looking at one trial run with one call Agent is applying its instruction set, its policies against it. And now agent is saying, hey user, I have what we need. Here is what I recommend to update into another system of record. Mind you, these updates could also trigger a wide variety of downstream workflows that are living in Salesforce, right? You might have built some Agentforce agents there, some workflows, et cetera. But now when I got this result, everything looked good, maybe except the technology stack notes. Like, as the AE, I thought, hey, the conversation mentioned more items than the 4 that were extracted, right? So I am asking agent to double-check. There are 2 parts to it. One, the agent might change its answer, but second, there is also some persistence with this memory. Now the agent knows Hey, I did something, but it didn't land very well with the user. So how do I persist this so that the next run or the next 100 runs, I will actually adjust the agent behavior. So then you actually get it to a level where, hey, this is not an individual productivity agent for me as the AE. Yeah, this is actually a revenue operations agent that runs on behalf of the entire team. But then if an agent is writing into Salesforce, that who is that agent identity? How do we keep a very rich trace at scale so that when you need to investigate, like, what, what can be improved for this agent? Who goes and evaluates that trace? Whether Glean AI can recommend, like, self-healing improvements, etc. So these are very exciting, like, product management challenges or opportunities that we are looking at. Not at the individual productivity level, This is well beyond it. But how do you step change the entire operating rhythm of a department?

So what do you— so from this example, so you've got 800 calls a day happening to Glean. Yeah. Gong can push data into Salesforce natively, right? So what is the agent abstracting from that for the organization that you wouldn't get out of just querying Gong or just looking at a Salesforce record? What are we getting out of it?

If we talk about that instruction set, the middle column, which was relatively, let's say, light. It is very much about, hey, given a call, there are gazillion ways to assess that call. The worst thing that you can do is to get a skill from the internet to say, hey, give me a call analysis skill and then apply it to all these things. So basically, if you think of a spectrum where on one end you get a vanilla generic skill to assess a call, On the other hand, imagine that the CRO is looking at every call and evaluating it by herself or himself, giving it all the mastery, expertise, context. Then the argument here is, what is the higher level of performance or expertise you can apply by bringing in more and more context, more and more expertise to assess that goal as if your company's CRO or founder would do it if she had the time, right? So this is—

and Gleam is extracting those, those learnings.

So when you are saying use our sales framework, you don't need to teach Gleam what the sales framework is, what that context is. You don't also need to say, hey, here is that document. Like Gleam would say, hey, I know the source of truth on what the sales framework is, But I also know like all these conversations happening around the sales framework, where it fails, where it needs to be updated. So I'm going to apply all these esoteric knowledge around how that Salesforce, how that sales framework comes into the picture for edge extreme error cases depending on the segment, depending on the geography of the customer, etc.

I got it. If you guys come, I think today Kyle Norton, who's CRO of a company called Oner that I'm on the board of, he's got a couple hundred sales reps. They use Momentum, which is kind of like a next-generation version of Gong. Salesforce bought them. It's probably not a coincidence. They go through, their CTO built this cool agent, I should have had him present here, where they go through every single call. This is SMB, so they're doing 40 or 50 calls a day, and every single day they get a real-time update on win-loss analysis and why.

Competitive loss, exactly why.

They can scroll it back through the month, through the day. But every single day, if you've got 200 folks doing 20 calls, 40 calls a day, it can synthesize all and say, okay, here's the trends. We're starting to lose deals based on this. And it's, it's pretty cool, right? That, I mean, it's too much data for a human to process, right? So, Anik, I know you don't have a, you don't have a QuickTime you want to show us, right? Do you want to paint a picture for us, or what do you want to do with your virtual, virtual agent demo?

I mean, I, I think the hard part is Harveet is not like a single— it's not like an agent feature that we have. The entire platform is agentic. And so the places that— so like our chat interface is agentic, right? So when people prompt questions, you will get multiple documents spun out. You will get a PowerPoint deck built with your template based upon your analysis. So I think there is like the end user portion of our Agentic platform. And then the second component is really the agent builder, which is kind of two types. You have a natural language agent builder that is, you know, based upon skills and datasets that you are selecting. And then you also have a capability that is much more deterministic in the sense of when this scenario comes up, please follow these exact 10 rules. Like 10 steps in this manner. And I really think of this as this agent builder as serving kind of two roles in the platform. I think at the law firm in-house kind of legal operations level, it is how does, how do the agents reflect my best practices as an organization? How do I ensure that the agentic work being done is based upon my bar as a firm? Firm as a GC. And then the other way that an agent builder really serves the lawyer is when I have repetitive tasks that I'm consistently doing, how do I create agents that just know me and know exactly how I like to do things and I'm able to repeat it ongoing?

Maybe that will— we're— I've got another half an hour of questions, we'll run out of time, but across a firm, how do you— how is Is it guardrails? Is it taste? What are you doing to have that firm-wide perspective in these agentic actions? How is it reflecting what you just described? How do you enforce, is it firm-wide? Is it individual attorney-wide? Is it group-wide? How do you actually make that happen?

So I don't think it is like a one-size-fits-all. Most major law firms have what they call knowledge teams or like the office, the innovation office that really is focused on best practices and the AI transformation. And so that is like a team of lawyers that are special.

Do they train Harvey?

And yeah, that's right. So they are like building agents that their whole law firm can use.

Yes.

Based upon Harvey, we have, you know, I think 180 legal engineers who are working with these law firms in-house teams on building bespoke agents based upon the firm's best practice, the playbooks, their precedent. And these legal engineers have themselves been lawyers for like 8 to 10 years.

Most of them have been lawyers.

Yes.

Your version of forward-deployed engineers were lawyers, right?

They are all lawyers. And so I think it is both like, how do you build self-serve customization capabilities?

Yeah.

And then also as, AI is like the shift of our lifetime. How do you have subject matter experts that can work with these organizations on their— to your point, it's not about like an individual's evolution, it's, it's about an organization's evolution.

All right, last question, then we do have to break, even though I've got about 4 questions for each of our panelists. Your— I know everyone has their own terms and chronology. Your forward-deployed engineers, so to speak, most of them are lawyers with deep legal expertise that get some help deploying the workflows for the law firms?

For us, yeah, we have both, um, forward deploy engineers that are, you know, you have an agent PM, you probably have one or two actual lawyers within that FDE unit, and then you have software engineers and they're working bespoke with specific—

you got a— you basically got a pod, you have a full pod, right?

And then we also have legal engineers which work ongoing with customers leveraging the agent builder resources that we have out of the box. But they're doing the next level of customization, really like unlocking the power of what the platform can do.

But just, I think it's just a good insight for everybody. If you can pull this off, when I go to buy Harvey, and it's not cheap, right? This is mid-market enterprise enough. I not only get an FTE helping me configure these workflows, these agent workflows and others, but I get a lawyer as part of that pot. I get, I get the best of both worlds, right? And the lawyers, you bring in their decade of expertise for how this should work. I mean, the FDEs are going to get pretty good after 6 or 9 months, but you get both when I, when I buy, right?

FDEs are only with specific customers that need like truly a—

Oh, not everybody needs an FDE.

Not everyone needs an FDE, but legal engineers are part of our deployment always because it is is so valuable, right, in having a subject matter expert help you understand how to fully leverage the Harvie platform.

I just think it's less common out there in general. It's not the most common agentic process where everyone gets a subject matter expert more often than they get an— I mean, we all need FTEs too. Many of us need FTEs. So it's a great learning. All right, let's thank the panel. This was great. We're out of time unfortunately, but thank you very much. At least I think so.

Thanks guys.

Thank you. All right, thanks you guys. The world has changed. AI is the accelerant. The amount of disruption that the software industry is going through now is unprecedented, and it's accelerating. Innovation's everywhere today. SaaStr is all about being positive. It's all about helping. It's about doing it together. The reality is we have to move faster. So what I'm telling every CEO I meet, you got to refound your company. If we could start over, what would we build in our space? What would we build today? How can we do it twice as fast? And just go find a friggin' way to build it.

Once again, please welcome the founder and CEO of SaaStr, Jason M. Lemkin.

All right, everybody. Again, thanks for being here. We have— I always kind of do, I guess, a late intro or summary. I know most of you were here yesterday and we're pretty deep into it this year. And I think if I had to summarize everything, I'll get to the next slide. You know, if you were here last year, it's nothing like last year. I mean, the rate of change is so high and I, I kind of wanted to summarize if it's helpful. There's so much on social media. Anthropic just is projecting, I think, that they're going to be at $50 billion in revenue by the end of June. That was out today. But we have a SaaSpocalypse. With many leaders down 70% in the stock market this year, what's going on? And I thought I would kind of try to summarize at least what I see, what we see. Tired versus wired. Having a little fun with an old trope, but tired versus wired, I think. And you, if you were here yesterday when I was with Amjad, the CEO and founder of Replit, the idea that we're going to vibe code our CRM, it's just dead, guys. It's just dead. Let's not talk about it anymore. If you think it's possible, use, use, use your Replit or Vercel credits from here. Try it. Go to one of these apps which have gotten so good today, Vue Zero and Replit, and say, build me my own Salesforce. See what you get. You're not going to get it. It don't work. Even if it did work, you could build your own lightweight CRM. It wouldn't be collaborative. It wouldn't work with the team. It wouldn't have the security. It wouldn't integrate with the 100 apps that are here. No one's going to vibe. You're going to vibe code a lot. We vibe coded almost every single application running this event. But you're not gonna vibe code anything you can build that's good, that's actually good, that's agentic. The opposite, it's quite the opposite. That was what people were saying last year. The world's gonna end, you're gonna vibe code everything, all our apps are dead. Now if you talk to the folks here, if you talk to Monaco or Artisan or Lightfield or Oracel or others, the goal is that the CRM brings in deals itself. Who cares if you can vibe code a crummy clone of Salesforce from 2006? What if the CRM puts deals on your calendar? That's what we're building today. This is what you should be building a version of if you're not yet for your own companies. Not tracking deals for humans, putting deals on the calendar. If you— some of these vendors here, guys, are doing millions per week and didn't even exist last year or barely existed last year. OK. They're doing millions per week in new revenue, not because the leads contacts and opportunity tabs are prettier or have purple gradients, but because they're putting friggin' deals on your calendar. That's worth $50,000, $100,000 even to the smallest company, whereas paying $200 a month feels like a lot of money for software. Okay, so that's your goal. Who cares if you can build your own CRM? I want deals on my calendar. The other thing, and hopefully most folks in this room are well past it, but the world still isn't. I don't trust AI last year. It hallucinates too much. Everything's wrong. Guys, look at the apps that we've talked about. Look at the CPO panel we just have. The biggest issue we're going to have going forward is our agents are going to get so good they sit idle all day long. They don't have enough work. 10K, we've talked, we're going to talk all through event about our own agents, 10K that we built, our VP of Marketing, VP of Customer. 10K is just sitting here all day waiting for something to do. He's already done all the marketing for this event. 10K is already— I didn't get to the slide before. 10K is already nudging us and annoying us to get going on SaaStr 2027. I'm not kidding. It sends us 3 things we should be doing this morning for SaaStr 2027. I'm like, 10K, we gotta get through this event. Leave us alone and we gotta take a break. Okay? Our agents sit idle. Some of the vendors here that we've used for our AI SDRs that we've talked about a lot, they sit idle all day because they're so good. They've already hit our entire base of 450,000 people. So once you actually get an agent into production and it's valuable for your customers for real, and they'll pay a lot of money for it, we're gonna have to start thinking, what do we do the 23 hours when they just sit there waiting for something else to do? This is something we should be thinking about, not hallucinations. You will— any tool you buy, if you don't train it at all, if you do nothing, it will hallucinate. But the guardrails are good. You train it for 30 days hard, you keep updating every week. We've worked through a lot of these issues. Third one I wanted to highlight is that last year if you came here, and in some ways it got worse at the start of the year, the theme was SaaS is dead. It's dead. It's all AI. It's dead. Everyone's dying. Every company's terminal. In many cases the stock markets still believe it. I'm gonna show you a slide in a minute, but if you read SaaS you've seen it. Not everything is dying. We've seen folks reaccelerate. Atlassian, Twilio, These are old companies. These are decade-old companies that have seen reacceleration, but not everybody has. And we started this year where the stock market said, oh my God, I just don't believe a lot of these public companies are going to be worth anything. They plummeted. The SaaS-pocalypse was the public market said, we don't believe in NRR anymore. We don't believe that 110% NRR is trustworthy. We believe all of the budget in the CIO's office is going to AI spend. We believe your 3-year contracts are hiding the fact that nobody's going to buy your products in 3 years. The public markets gave up. They sent leaders, epic products like Monday, HubSpot, Atlassian down 60 to 70% in a couple of months. It's not— what we have to do and what all founders have to do is let that go. It doesn't matter. We are actually in a B2B Cambrian explosion. We are seeing millions of new applications and species. Look at folks out here that are doing tens, hundreds of millions of revenue that weren't even that barely existed a year ago. If you were just on our product panel, Harvey just raised at $9 billion. They didn't really exist years ago. Replit, Lovable, they didn't even— Lovable didn't exist 2 years ago. Replit did a total tilt, which should inspire you. It was founded in 2016 as a pre-AI, and now is doing— coming up on a billion in revenue. So because we can all build apps, because we can vibe code apps, because we can build more, we're seeing an explosion. Now it's stressful because competition is exploding. If any of you are in categories where you have no new competitors because of AI, congratulations, you're lucky, right? But I think everyone has some— who are between 2 and 100 times more competitors per category. So that's stressful. But step back for a minute. That is a blossoming of B2B and AI apps. It is a great time to be in B2B. It may not be a great time selling a 2023 app that hasn't changed and evolved. It may be pretty sad and terrible, but abstract away. If you're struggling, if your growth is still decelerating, if you have not found your AI tailwinds, okay, great. Have, have a cocktail tonight again. Okay. This is your last time to feel sorry for yourself is tonight. Okay. Pick yourself off and realize this is the greatest time ever in B2B. It's just the worst time to be selling your grandpa's software. Okay. No one wants it, but it is everything. Budget for AI is It's everywhere, and the apps are so much better. I'm behind this. Even SaaStru, even our name, I talk a lot about B2B and AI. There is some psychological toll on old school SaaS, but man, this is the best time ever for B2B and AI. The other one I just wanted to highlight, maybe two more, things I'm just bored of talking about 'cause they don't matter, layoffs. Now why do I say this? Every time you read up X and Twitter, I mean, this was just this last week was brutal. Cloudflare beat the quarter, raised their guidance, and laid off 20% of their company. Okay. Everyone that is at scale that wants to reboot their company is layoffs. Layoffs last year were about getting more efficient, getting more profitable. But you know what? The public markets didn't care. Like, it's not enough to be profitable anymore because if the public markets think your revenue is going to zero, profits are not sufficient. So the layoffs of last year were about profitability. The layoffs of last week were about rebooting my team so they know how to build agents and tap into AI budget. But I don't think it matters because, I mean, listen, I think layoffs are tough. I hope nobody has to do layoffs. I actually think layoffs themselves accomplish nothing. But what's much more interesting is exponential productivity and what you do with it. Exponential productivity. Productivity. And I think the tough learning— there's a tough learning and a great learning. And even Amjad was thoughtful about this at the end when I pushed him, because, you know, our AI VP of Marketing customer success costs $257 a month. Whether that's productivity or deflation, it's a lot to reflect on, right? We replaced $500,000 worth of employees with $257 worth of agents, but the productivity is higher. It has made Amelia and me me, and to some extent David, much more productive. Much more productive. And this is what exciting. How can you use AI both internally and with your customers to make them radically more productive? Radically more productive. Internally, to be direct, I think by the end of this year we're going to give up on reskilling. I think folks are going to stop pretending they can reskill folks that are not going to be able to make the jump to the agentic age. You're going to give them time. Every CEO of a public company has to get up on stage and say, I'm going to try to reskill my team. But at some point you have to give up because it's too competitive. The market's moving too fast. So it's going to get worse. There's going to be more layoffs that are going to sort of compete with reskilling. But we're just going to move on. We're going to have the teams we have and we're going to start hiring more and we're going to have more agents. But who can be exponentially productive? It is so— it is exhausting for me and Amelia that we manage 21 agents. I didn't imagine this being my job, but the fact that we could manage this entire thing with, I mean, we have a lot of folks helping us, but with 2.5 full-time people with agents, we are so much more productive. Find your superpowers for your team and your customers, you'll make a lot of money. The last thing, just to synthesize, and I know for some of you this will be obvious, but there's two things happening in the public markets and in M&A. Look, we're all— almost everyone here is a builder, but we would like to make some money in our stock. It would be nice if our shares were worth anything. Okay. It really does help with the journey if someday— doesn't have to be today, doesn't have to be next week, it doesn't have to be this year, but it'd be nice if we can make, make a few thousand, million, $10 million, $100 million from our equity. This is both like, like it's been a Debbie Downer 12 months. Basically, there are almost zero B2B IPOs from the end of 2021 until 2025, and basically almost all the IPOs in 2025 were a dud. Okay, great companies like Navant, dud. Figma rocketed, now it's way off its highs. Crappy time. Nobody's bought— PE, private equity, the Vistas, Tama Bravos, at least they would buy you even 24 months ago. If you came to Saastr 2023, we have a QA at the very end tomorrow. Come if you have time, I'll do QA until I pass out. And until 2023, founders would come up on stage and they'd say, we'd have some great founders. I'm at $20 million, Jason M. Lemkin. I'm at $50 million. I'm at $80 million growing 30%. What do I do? I'd say, well, are you profitable? They're like, yeah, I'm cash flow positive. Well, have you gotten any offers from PE firms to buy you? Like, yeah, I got 5 offers. Maybe they were at 3 times revenue, 5 times. They didn't love it. But you got to $20, $30 million in revenue with even 30% growth and cash flow positive, you're going to get offers from private equity firms. They're buying nobody. If you listen to what the best, Thoma Bravo is the biggest that bought all these B2B companies, they're saying, "Yeah, we're doing some purchases, but only if we see real lift from AI." That sucks. IPO window mostly closed, M&A down, public B2B stocks, public software stocks for the first time ever are trading at a discount to the overall S&P 500. That means the whole point of software was it's so efficient, write once, buy anywhere, read anywhere. It always traded at a premium. It was such a wonderful business model going back to Microsoft Basic, Office, Intuit QuickBooks, the most wonderful model in the world. For the first time ever, the public market said this is worse than almost any other business I can find. But it's all about to change. I know some of this is obvious. It's all about to change. We are about to see the 3 biggest IPOs in tech forever that are gonna create more value than all the other IPOs in our tech lifetimes. SpaceX, Anthropic, OpenAI. We'll add in Databricks. We'll have Arslan, one of the, uh, co-founders here in about half an hour after this. That's gonna create more market value, more value, and more wealth than every other IPO since 2000. It's gonna be good, and it's gonna lift boats up, and it's gonna lift tides up. And M&A, we're gonna lose, we've lost, and I don't think we're ever gonna get back where any B2B company that got to $20 million with decent growth got bought. But if you hit it, I mean, Cursor just is getting, I mean, it should close. Cursor's getting purchased for $60 billion. When Wiz got bought for $30 billion, that was a jaw-dropper. It had never happened. Now Cursor at $60 billion. $60 billion. If you listen to our podcast with Rory and Harry, I bet there'll be a $100 billion startup acquired in the next 12 months. It only makes sense with where the markets are. So, um, bad news is you got to do better than ever, right? You've got to step it up. You've got to build a great agenda product. But if it is, there's more opportunity, more IPOs, more wealth creation, more acquisition dollars than ever before. So you got to decide which camp are you going to be in. Are you going to be in the tired camp? Are you going to be in the wired camp? But Everything is getting bifurcated. We are not— when we did the first SaaStr, guys, 2015, it's a little while ago, we all felt the same in the room, and I hope we all feel like a community today, but as this next 12 months going on, it's gonna feel more and more bifurcated. It's already feeling bifurcated, but the ones benefiting from AI aren't just growing more quickly, aren't just raising venture capital earlier, they're just gonna explode, and the rest are gonna just continue to decline. Decline. So you gotta find your way to be in that category. But it's gonna be awesome if you can. A quick review on us, and I know most of you know this, so I wanna hit a few other points, but if you've watched any of our workshops, you'll see this. Literally just for the rate of change, and I think what we've done going from 20 humans to 21 agents and 3 humans is any of you could do it. But even when Amjad from Breplet said we're living in the future. So I just wanted to, if you haven't seen this, this is our journey. 2024, we were 20-something humans to do less than we do with 3 humans and 20 agents, literally less in 2024. If you came last year, we had one agent, this Delphi chatbot that we all, if you came to the 101, the pre-opener yesterday, we all built one. You can build one for free, go to delphi.ai. You can do it now. You can build a free chatbot, train it on your knowledge, and have an agentic chatbot clone of you running before you leave the session. Just go to delphi.ai. That was all we had last year. The only hack we had is I put everything in it. I put the prospectus for our sponsors. I put every— I put the Wi-Fi password for this event. I put every detail. And we started to see that just with one agent, it started to do things we didn't know it could do. It started to try to close one sponsorship deal for $60,000 on its own. It started to do more and more and we realized, wow, that's pretty powerful. So, so Amelia and I came out of SaaStr AI Annual last year and said, okay, we're all in on this stuff. The first one we added was Artisan that's out there to do outbound for us. We trained it, it got good. Now we have 21 of these agents and it's pretty good and it may sound crazy. And as Amjad and Francis here said, we live in the future doing this. But I can tell you like we don't even have an engineer on our team, like Yeah, I co-founded a SaaS company and built all the wireframes and specs that went from zero to a couple hundred million in revenue, but I'm only so good. Anything that Amelia and I have done, anyone in this room can do. And you can do your own version of it. Amelia gave you the recipe for your own AIVP of marketing. You may not do it exactly as we do, but nothing we have done is particularly clever. We just push the envelope a year earlier than 99.9% of the folks. It's really not that hard. You can do it if you want. And one thing, I think most of the room knows this, but I just wanted to make this point to hammer it home. Hopefully most folks here today are not AI skeptics. They're not the Debbie Downers that I see on Twitter or LinkedIn sometimes. It doesn't work. AI SDRs don't work. I tried them 18 months ago. Chatbots suck. They don't work. Listen, if you were here with the Amjad session, you'll realize that our AI VP of customer success was the frontline manager of the 100-something sponsors we have here, and overall they loved it much more than humans. This wasn't a chatbot. They reached out to all these folks, found out their issues, got their complaints, did it autonomously on its own by email and by chat on our website, and then summarized all the issues. Okay, everything is so much better than last year or really than December, right? And just to highlight 3, and you guys may know this, if you're skeptical of AI SDRs and AI sales tools, stop being skeptical. Instead, invest a month in picking a platform and training it. They're pretty damn good today. Listen, you may have the world's best SDR and outbound team out there, but our SDRs have done better than any single outbound person has ever worked at Saastr. Ever. They are better, but they weren't good in 2024. Okay. I touched on this with Amjad from Muppet yesterday. These vibe coding platforms, Replit, Lovable, v0, even vertically focused ones. They're so much better than a couple months ago. If you're skeptical, just fire it up. There's like free Replit credits out there at the little vibe coding tent. Try it. Try to build something you wanted to build. If you tried last year and it didn't work, I get it. I had a little bit of a flameout on my first Vibecoded app, if anybody saw it. I don't really like to talk about it. It was like a year ago. Now we build anything. We build— it all works. And it— some of these platforms got better last fall for the harnessing. And then everything got better in December when Claude 4.5 shipped, Sonus and Opus. Everything got better in December. And then over the last months, they figured out how to take advantage of the LLMs. LLMs got better, everything's better. So like you're benefiting, these platforms are so much better than they were. And then your product should be better. Like if your product Claude 4.7 and Phi-5 on OpenAI, they're so much better than before that if your product, if your own B2B products are not much better than 3 months ago, you get an F because you should just get a boost from the fact that these LLMs are so much better. Right? I mean, just for fun, if you follow this Aster thing, you know, just right before the event, I know I told the story, but think about it for a minute. We— Amelia spinned up a parking permit app. I know I talked about it, but let's pause for a second, okay? What this app did is— I love these guys at the event center. They're great. But one thing they do is they give us a 5,000-page PDF with custom parking passes for everyone that needs one here. What do you do with a 5,000-page PDF? Well, we used to have a human, that would find each page customized to the right day and the right parking lot. And a human would manually send them and make a lot of mistakes and complain and often quit and not want to do this trivial task. So Amelia, like a week and a half ago, was like, could our, could our, could our AI agent do it? Could 10K do it? We actually had Annie do it, which was a new agent, and it was marvelous. And she spent about an hour or two in the Replit agent building this. And it's automated all the parking passes. And it sounds like a small example, but that wasn't even probably possible until January or February of this year. You should have done a lot better of this with your own apps. The LLMs are so good with some work. You should be doing magic in your app that you couldn't even do at the end of last year. So try again. This is my meta point from all this. If you're skeptical, if you don't think it works, try again because it probably doesn't work. Even for us, we started early. It's more interesting, the agents we're using, the ones from third parties like here and the ones we built internally. On Saastr.ai, we have own apps that I've built. I started pretty early before everything got better. We built a cool app to help you value your startup. It's free. Just go to Saastr.ai, AIVC. We built an app that judges your startup pitch deck objectively. 4,500 times. So the other day we passed a million times these apps have been used. It's just incredible scale. You can do this too, right? The apps that I built with no help and no engineer have been used 1.2 million times since last fall. 1.2 million times. And when last year this was our agentic stack, nothing was connected. We had one chatbot on Delphi, digital clone of me, it was cool. It could tell you the Wi-Fi password and where the happy hours were. It integrated with nothing. It was a siloed app on Delphi. We had Salesforce connected to nothing. We had one person using HubSpot connected to nothing. We had the world's oldest marketing automation platform, Marketo, connected to nothing. And this, this rogue digital Jason, nothing. Now we've got 21 agents all orchestrated and working together. You can do it too. Okay, I wanna breeze through the next points to make, so I can talk about the markets and where we're going. But just a couple points, if you follow SaaStr you know, if you don't I wanna highlight it here. I've said it before, I'm probably gonna say it again this afternoon, and I'm definitely gonna say it if you come to our morning kickoff tomorrow where Amelia and I go through our entire agent stack, soup to nuts. If you have the energy and you care about this, we'll go through every single agent, We'll show you the backends. We'll show you how all they work until we drop, I think starting at 9:00 or 9:00, check the agenda. But still, for the love of everything, buy, don't build. Even though we build every day, buy, don't build. If any of these products work for you, stay with our products, buy them. It is not worth building. The stuff that we do, which is badass, you can't buy a parking pass application out here, I don't think. If there's one, let me know. I'll run out and I'll put my credit card down. We had to build— who's going to build that app, right? Not only is no vendor here going to build it, no dev shop is going to be patient enough to build it. Like, no, no one's going to build it. So that's when you build, right? If we— I believe sometime this year we may be able to buy an AI VP of marketing that is better than QB, and I'll be sad and I'm not going to turn QB off. We're going to talk. But yeah, we will sunset QB when we can buy it. Just doesn't exist today. But for the love of everything, buy it if you can. And then one thing, and I did, I touched on this with Amjad if you were there. We've been saying this for a while and I'll show you, I know this is gonna be hard to read, we'll share these slides, but a lot of folks say chatbot sucks. I went to a website and I talked to their chatbot and it sucked. I literally went to a public SaaS company the other day that does, has a marketing automation solution, and I asked them what the pricing was from SaaS, it didn't know. It's not okay to not train your own agent on what your pricing is. If that was your last experience, I get you, but there's no excuse for that. So what we did, QB, our AI VP of customer success, we had QB go back, let's see if I can read this to you. We had QB go back and talk to all these 100 and something sponsors who are just customers and handle it itself. Was it perfect? No, QB was honest. But short answer, 120 folks were mostly satisfied with me talking to them. They were mostly happy. Okay. Pylon, which does support, the writer, go talk to them. They said we screwed up with the wrong logo, but that doesn't mean Kube did a bad job, right? It just means we screwed up. But at least he serviced the issue helpfully. But all these folks from Crest Data to Revo, they said they had a great solution. Gamma helpfully. So this isn't just a chatbot. This is a customer success agent on its own, proactively reaching out to folks finding out their issues and where it can solving them on its own, this is better than 99, 95% of humans are gonna do. 95% of humans, when we had human customer success people at SaaStr, like we had 2 people one year, they didn't send out any emails until after the event. You can laugh, right? But that like wasn't the great, they said they were still learning. They were still learning SaaStr after 2 months. They didn't reach out to a single sponsor, okay? Qubi don't have those problems and Qubi is trained on every piece of data we have. QB is trained on everything. QB knows if you submitted your booth graphics on time. QB knows everything you were supposed to do. QB knows the humans never cared and did anything. So it's an example. You can read this. If you really train an agent or just a chatbot really, really well, I think it's going to outperform 95% of humans. And the meta-learning is at the top. Customers will prefer, prefer a great AI to a mediocre human. They'll prefer a great AI that solves their problem today, that knows the answer, to someone that has to get an engineer and an appointment to you next week, that doesn't know, that says, "I'm just new to the account because your old account manager quit last night and I never told you," or last month or last year. They prefer a great AI, so build a great AI. And sorry, is anyone from Team Saastr here? Do you know how much time I really have? Anybody here? Nobody knows? A little bit of humor. What's that? Isn't there a session at 10:30? No? I have a little longer? Okay, sorry. It says 16 minutes. Okay, thank you. We're all on this together. Most things are so much better this year. This is one thing that's just a little bit worse. Okay, so I do want to summarize one thing and then I want to move on to where the world is and where the public markets are and if a few other learnings. Where we are today, right? So much different than last year. You can be, you can live in the future like us if you want to do. But let me kind of summarize some things that maybe everybody knows, but I think a lot of folks don't know. If you build an agent that truly works for real, not an AI feature that like, like that, that chatbot that didn't even know pricing. If you build an agent that, that as we talked about earlier puts creates customer appointments on your calendar for you, whatever it is, if it's Harvey, like with Aneek that was in the last session, and truly lets you process 100,000 legal documents for real and get back to your clients at a rate you never could before, if you deliver near instant ROI, demand is out the door. And you look at, okay, Anthropic's revenue is like nothing we've ever seen, okay, it'll be at $50 billion in, in less than 4 years, or 4 years plus or minus. But if you're not in the loop, if you're not one of the, some of the folks here, you might be wondering, how the hell did they get to $100 million so quickly? How the hell did they book $3 million last week? How the hell did, like I've never seen this before in my poor older SaaS company. It's so hard to close 2 deals, okay? Well, it's not, I tell you in a lot of cases, it's not 'cause they have the world's most amazing sales teams. Okay? It's not because they have the best PowerPoint slides to the pitch. It's that there is incredible demand because the ROI is so high. Right? And just to give an example, like you could go out and buy an AI SDR tool here for $50,000. It's a lot of money, $50,000. I mean, I don't know how it is for you guys, but when I was a SaaS CEO, a $50K deal was a big deal back then. Okay? You got a little bit of a celebration. I mean, you know, it wasn't the biggest deal of the month, but in the early days, that was— we'd go out to lunch. You know, it was hard. But that's because we were selling medium-term ROI and ROI calculators and really great software, but ultimately somewhat incremental improvement to your life. If you can get me more revenue tomorrow, $50 grand is nothing. If you can get me a half million of revenue, $50 grand is nothing. And so for truly great agents, People will pay so much more than software, and that is your job. So your job is to build the number one agent in your space that will in many cases either deliver instant ROI or really replace humans. And it is so jaw-droppingly valuable that you can charge 2 to 10 times more than your software. So just like if you look at, if you look at companies like why did Shopify and Toast succeed pre-AI? Their SaaS businesses actually aren't that big. Toast and Shopify is— if they were just software, they're pretty mediocre businesses, but they figured out how to add merchant services and payments and payrolls, and all the vertical SaaS companies are doing that. But the point of the story is those value-added services are worth geometrically more than just the software at Shopify. Shopify was just a software for a website for e-commerce. I think that's only $2 billion or so of its, of its $14 or $15 billion in revenue. It had to add more value. The same is happening with agencies. Agents, an agent that will get you revenue, that will manage your front office or your back office, that will automate your AP and AR, that will automate these things.

More—

most likely it may replace a few humans. It's a social issue, but most likely it can be worth 10— 2 to 10 times what you're charging for your software. Your job is to find and build those agents. Um, okay, maybe just two other points. These next two, I— and I want to highlight, if If you feel behind, if you don't— not totally sure what to build, I can tell you all the products we use, we— their ROI is so high. They're great. We pitch them. We talk about them on The Agents, our new podcast, every week. We demo them. But I gotta say, overall, a lot of Agentek products are very primitive. They're still— if you play with them, and we'll demo our whole stack again tomorrow morning. Open it up if you haven't. Try Delphi, for an example. You're gonna see front end and back end that look a lot like software from 3 years ago. It's just they've integrated LLMs and they've integrated workflow on top of it. It's early days. It's early days. AI SDRs didn't work 18 months ago. They began to work when we did it last year. Now folks like Monica that we've used can get appointments on your calendar for you. That wasn't even possible a few months ago. So the software is still early. So if you're behind in your category, my My point is, if you don't have the right team, I don't have an answer for you. Okay, maybe you're back to layoffs for tired versus wired. But it is not, I can't think of a single category where it's too late in B2B. It is still early. If you have the right team, it's true, you can clone your competitors in months, and if you have the best team, you can pull ahead. Okay, it's still very early days. They just work really well if they're trained. The third point to summarize, then I'll spend a few minutes more meta on where everything is going. Emilie and I have said this so many times, but if I had to summarize it, I couldn't say it too many times. If you look at what we're doing and others like us are doing, why does— A lot of folks are like, "Oh, well, these agents only work for SaaS because you have a brand." Well, you know, a lot of people have brands. You have to earn one, but you really don't need to go that far, it turns out, for people to have heard about you and your industry. That's not why. Why these agents all work for us is because we work with them every single day. We make them better every single day. We train them, we iterate them, we talk to them, and tools matter, but consistency beats everything else. There is no set and forget today in agents, but the key is every single day. The outputs in many cases are just pretty good, but consistency beats it. So that's the cheat code, and there's no reason you can't copy it for yourself or for your customers. Okay, a few minutes on higher-level stuff, the markets, the world, to challenge you and to inspire you. I put this up on SaaStr, but Gartner itself is having struggles in the age of AI. Gartner's actually public, most folks don't know it, you can track it. Their events business is actually doing pretty well. Everything else is sort of being destroyed because We're using LLMs to decide what vendors to buy. It's benefiting G2 to some extent that's here, but Gartner is in trouble. But I like Gartner data because they're still doing the work. You can say, oh, Gartner changes their mind to their bias, but Gartner is the only source that I know that is actually talking to hundreds and hundreds of CIOs weekly and publishing data. So I do love the data. And so look what it says for software. Even though there's a SaaSpocalypse, even though folks, Many folks are seeing new AI competitors steal their customer base. On balance, overall software spend is radically reaccelerating from $12.8 billion spent on software last— sorry, no, 12.8% growth, $1.2 billion— trillion, $1.2 trillion. Yeah, sorry, $1.2 trillion to $1.4. That's going from 12.8% growth in all software spend last year, which was high, to 15% growth this year. It is accelerating. It is accelerating. Now, it's stressful because it's not that simple. A lot of the, like, about, depending on the data you look, about half of what CIOs and enterprises are spending is net new for AI. A lot of that is going to Anthropic and friends, but they're also cutting a lot of older vendors and consolidate to free up revenue for new agentic tools. So it may not feel to you Like spend is accelerating. You have to tap into AI budget. You have to tap into AI spend. But if you do, these are the best of times. Software spend overall is accelerating like we have not seen in a decade. So you got to go grab it. And so we'll see what happens over the rest of the year. What's happened now is in some cases, in some cases, reacceleration has started. So I just put together this chart the other day on SaaS-ter. The craziest one of all, if you don't study it, is Palantir. You can say Palantir in some ways is benefiting from defense spend. Of course it is. But Palantir was kind of a dog a couple years ago. Growth had fallen after the IPO into the teens. Folks were leaving. I was talking with the CEO of a $10 billion company the other day, and he's like, "Yeah, a couple years ago, I could steal anybody from Palantir." He's like, "Not now." It was a dog, but, but boy, did anybody tap into AI budget more than Palantir, going from 27% growth just 2.5 years ago to 85% growth today. 85%. Alex Karp said in their last earnings call he's projecting over 100% growth at their scale. Look at that reacceleration from the budget. So that's the outlier, but it doesn't mean you can't be an outlier. There's sort of a middle group that is tapping into AI budget and accelerating. So I picked Datadog, Cloudflare, and Shopify. Shopify. Datadog blew out the last quarter because it's the same old Datadog that we all love. A little too expensive, but everybody loves Datadog. But they just nailed getting all the hyperscalers to buy them, right? Basically all the AI guys run on Datadog. So you want to be that too. Datadog was not founded last year, okay? Datadog was the, was the new New Relic, if you've been doing it for a while, that we all loved. And it used to be cheap and simple and it was great. But they got every AI hyperscaler to use them. Figure out a way that everybody that is blowing up in AI is also buying your product. That's the Datadog secret to reacceleration. Twilio and Atlassian are just super interesting, and then I'll contrast it with who isn't. Twilio was dead less than 18 months ago. It was dead. It was dead. Growth had fallen to the single digits. They pushed out Jeff Lawson, one of my favorite CEOs. The shareholder activists had pushed him out. Wanted to just run it for cash. And then boom, what do the agents use for communication? What do ElevenLabs use, which is already doing $500 million in like a year and a half? They're using Twilio, okay? They're using Twilio's APIs, and boom, growth goes from 4% or 6% to 20% in 18 months. That's an epic turnaround for a public company at this scale. Not everyone is selling, are selling at the API level. Not everyone can do that, but if you can, if you can be an API that is key to agent spend, AI spend, you can blow up like Twilio from the dead to accelerating and predicting further growth. Atlassian is the latest example if you study it. Atlassian, it's not a perfect story, but Atlassian was at pretty good growth at, Boy, I should know, $6 billion in revenue or so, but plateauing growth. And then they really did get their Rovio agent. We had the chief AI officer yesterday presenting, if you came. They didn't, they started off with an AI agent that sort of helped, but it wasn't monetizable. It wasn't good enough. Then they built a robo agent that actually had enough value to the customer base that people would pay for it, and boom, look at growth up to 32%. So they went from a 60% solution like we talk about at SaaStr. "Fine, but not something I can pay for." To something that was truly disruptive for their base, it worked. The cautionary tale with Atlassian is that their customers love it, but it hasn't yet increased the rate they're bringing in new customers. That's sort of flat. Your real job is to build an agent that people love, like Atlassian did, sell it to your base, but also attract new customers. Atlassian still has more to prove for the latter, but boy, it's really accelerated. We're waiting on folks, we're waiting on— So, but folks that are a little bit earlier, a lot of the classic B2B players— ZoomInfo, HubSpot, Monday— maybe they may have been unfairly punished by the public markets, but they're waiting for that reacceleration. They're waiting to tap into that agentic revenue. So it's this bifurcated market. But I don't think it's too late for anybody. If HubSpot's just crossing 300,000 customers, its net new ads are pretty impressive. Atlassian's challenges Its customers love its AI, but it's not attracting new customers. HubSpot's challenge is its AI tools are very early and no one's really paying for them. But people are still buying more HubSpot. They're crossing 300,000 customers. So even though, you know, maybe HubSpot's behind, if they can get out great agents in the next year, they got 300,000 customers to sell it to and they're still growing their customer count double digits. So it's not really too late for anybody. It's just super stressful this year. So my challenge to everybody It's just who are you? Be honest. I'm in too many conversations with founders, even in board meetings, guys, even in board meetings, where founders are not 100% honest. Founders that are growing 500% are pretty honest. Okay. But folks who have seen growth slow but not to zero, they're not honest. They're not honest about what's happening. They're not honest about what's happening in the market. And so I kind of leveraged a segmentation that Rory did on the 20VC with me and Harry. Which of these 4 categories are you on? If you're new AI native, you have no legacy customers, no legacy stuff like, like, like, you know, like Harvey on stage or Artisan or Monaco. You're golden because it's all up, right? Strong demand. No, no scar tissue. You don't have a bunch of people on your team that don't really want to build these products or work harder. You're category 1, everything's great because there's nothing, there's no pre-AI life, right? Category 2 is the best of all worlds. You're pre-AI, but AI is driving deals up and new customers up. That's Twilio, that's public companies like WorkOS. I think Michael might have been here, which built an OAuth system and others years ago. It was great, but the company struggled. It went through layoffs a couple years ago, and then boom, like All the Agentic tools use WorkOS, so it's exploded. RevenueCat, where I was the first investor, they do mobile subscription management for 50% of mobile paid apps. It's a great company, but boom, then like every Vibe-coded app, like mobile just exploded and they're in every single app. If AI is driving deals up and new customers up, that's really what most of you want. That's the best place. 3 is AI is just expansion. That's the Atlassian example. Still better than most. And then Category 4 guys, again, after your cocktail this evening, you got to get— you got to be honest. If you're in Category 4, the recovery is not coming. If you're not getting the AI budget, it ain't coming. It ain't coming. There is no additional budget for pre-AI workflow tools, pre-AI Kanban tools. It's just not coming. So stop pretending. Stop pretending it's just been a rough year. Stop pretending we're going back to 2024. I know nobody in this room is literally thinking we're going back in time, but I got to tell you, at a lot of companies that were growing 50% or 100% that are growing 10% or 15% or 20%, there's a lot of performative work being done. There's a lot of folks running the same playbooks as a year and a half ago. There's a lot of folks talking about AI in board meetings with slides. But I don't see— I don't see the products coming to market that people want to buy. There's a lot of folks running not just the same playbooks, but acting as if something's going to happen and growth is going to reaccelerate. If it hasn't reaccelerated by now, you have to make it reaccelerate. The markets are not magically coming back to you. Maybe just a couple other points and I'll wrap. Even we had to do it. Even we had to do it at Saastr. If you've been here before, you might notice that there's a lot of different sponsors here. So I asked Claude to pull together a map. Actually, I asked TenK, our AI VP marketing, take it back, to pull together a map so you could see it. I'm a little colorblind, but even I think I can see these colors. This is the number. This is like how all these sponsors are. Is it orange or red? I can't tell. Sorry, someone help me out. Orange. The orange are AI-native sponsors that are here. The other ones are the pre-AI ones that came back. So like almost If you consider our sponsors customers, and we do, it's good for learning. The vast majority turned over this year. If we didn't change, we would have canceled SaaStr. There wouldn't have been enough revenue. We don't make enough money on the tickets to this event to put this event on. We could talk, but we're running out of time. These are our friends out here. Without them, I mean, it's fun, we learn from them. Without them, we can't turn the lights on. If we didn't change, we would have sent you an apology note and done a digital event. Because look at this turnover. It's epic. So look, we found a way. You can find a way, but you can't pretend. If you pretend, we would have been dead. And so maybe my last point, or I'll run over 1 minute. I'm a little confused on the timing. So what does this mean? I would say 3 things to founders, 3 things that are actionable. And maybe I'll wrap up on this. As Amjad said yesterday, we're early. Too many people on Replit running semi-autonomous AI VPs of marketing customers. We are early, okay? But learn from us. Learn from the vendors here that are crushing it, that are doing millions, millions and millions of new revenue, sometimes per week. The agents are coming. We are all gonna be in a gentec world. Take it seriously. Take what you can do autonomously for your customers, and in many cases replace human labor. I genuinely believe if you can build the number one agent for your customers, you will win. We're gonna have one fun session Tomorrow, here's Adam Guild. I'm on the board of a company called Owner. It just crossed $1 billion in sales for restaurants, for SMBs, for its customers. Talking about the agents they build, their AI CMO. It's actually pretty new. He'll tell you what they built for the AI CMO. He'll show it to you, why it worked, why it doesn't. This is your job to build the best AI. For him, it's an AI CMO for small restaurants that don't even have a single person in marketing, whatever it is. If you build the best agent, there is— maybe we will have agent parity in 36 months. Maybe even in 24 months. Today there is not agentic parity. There may be parity. You know, a lot of categories we end up where it's a little hard to tell. 4 years ago, sales loft versus outreach. What's the difference? You know, who the hell knows, right? It's almost like, do you like chocolate or vanilla? Was sales loft versus outreach back in the day. The agent performance is so disparate that the best agents are just gonna win over the next 12 months. So that's your job. The second one, which is probably most folks in the room are cool with, but a lot of the bigger companies are struggling with, should you be open to agents?

Agents.

This is one, you know, we congratulate Salesforce a lot at Saastr because it is super open to all these agents. All of the, all these sales CRMs are able to plug into Salesforce and extract the data as well as push it in. Okay, we run Salesforce headless, we run 10K and QB on Salesforce, and folks like, uh, Workday and others are struggling because they're not sure how open to be your platforms. But if you're not open to agents, if you're not agent friendly, they're gonna bypass you. Again, I told this story earlier, but we're leaving Marketo, our, our most dated piece of software, for a new marketing automation platform. We're leaving for the most agent-friendly marketing automation platform we can find. You've got to be friendly to the agents. And so the last point I'll make here, and then I think I'll leave it on this, when you go back from SaaStr, you'll, you'll learn a lot, you'll have a lot of ideas, it's great. But I honestly think, even though it's a little bit of the future, the The single most, the single best thing you can do is make your API, most of you have an API, make it agent-friendly. Don't do what Marketo does, which is run out of API calls in 1 hour and make you wait a day or a month to get more API calls. Figure out what the agent wants, because if the agent picks you, you're gonna come up in the LLA recommendations. Folks that are building agentic products are gonna pick you because you're the most agent-friendly, right? We moved our emails to resend because it was the most agentic-friendly email platform, okay? And we even built on Saastr. You can go to Saastr AI and find it in the navs. We have an AI agent report card, and it grades all the leading APIs, and you can even put your API in. It says there, submit your API, and we'll scrape your API, tell you how AI agent friendly it is. Do this when you get a chance. Right now you have to submit it, but if you don't get a B+ or better, just tell your team to build a more agentic friendly API, and it will act— you can scroll through lots of grades from Stripe got the only A+, right, to a few Cs and Ds. It will tell you and agents exactly why their API sucks for today, why agents won't wanna use your API. If you get an A- on this, then at least folks that are at the edge that are wanting to build on agents are gonna pick you. There may be categories where it doesn't matter, but this is actually not that hard to change your API, okay, to allow an agent to get more data from it, to allow it to pull your API more often, to structure the output the way it wants to do. This could be a couple days of work to turn whatever version of you that can be headless into the winner in your market. Even if you don't have an agent, make your API agent-friendly. Because guys, as tough as it is, and I will wrap on this, everything's about to get even more competitive. Everything got so much more competitive the last year because AI just created 1,000 flowers. As I said at the beginning, old SaaS may be dying, but new SaaS, new B2B is exploding like never before. And you can feel it in the competition. And but it's not that simple. The competition is so much better than everyone was. Not only are the vendors better, but everyone knows what's great now. Claude and ChatGPT are so great. You can't get away with crappy agents and crappy software. We all know. And this The fact that we can replace humans or make humans more, much more effective means that even niche markets are much larger. And the one example I'll give, and then I'll give two more and then I will wrap. Um, if the craziest one are some of these CX support products, and I've done a lot of investments, worked a lot, but these Sierra Decagons and also legal, Harvey Law Grad, these were niche markets before agents. These were niche markets that are like 10 to 100 times larger because the agents add so much value. So it also means in a lot of markets there's a lot more competition because you can monetize them much better than you could with agents. And then maybe two last things to think about and then we will wrap. I copied this from, or a version of this from Owen from Intercom the other day, but another thing over the next couple months is, we use 21 agents at Saastr, But everyone really, your customers just want to use one. They want to use one interface, maybe not one agent, but one interface. And there's a big battle that's coming, even though the number of apps is exploding. When customers come to you, they want to use one agent, one interface, and this is rebooting. This means sales, marketing, and support are all blending because I don't want to talk to three different people or three different agents. I want one agent that knows support and sales. And this is coming everywhere, and exactly where to predict it comes, I don't know. but it's creating mass overlap. And you used to have 4 to 5 years before people would compete with you. Now people may even be accidentally competing with you next week as these pains overlap. And the final point of tough love, and then I'll break, is— and I know this is tough for so many folks that come from other countries, from out of town. This is the tough truth, guys. And as I said in the beginning, it's not about layoffs, it's about having superpowers on the team. A lot of teams don't want this change. A lot of folks do not, you know, do you really, how many folks really want to work even harder because there's more competition and because, how many folks really want to learn how to build an agent that isn't just a feature or a widget but that people will pay $10,000 or $20,000 a year for? Some of your folks on your team are pumped up and excited and love this stuff and they're hacking and they're screwing around each night. Empower them. Maybe even promote them next week. Because a lot of your team doesn't want to do the work. And I know it sounds harsh, and I want to tell you that we can retrain them and reskill them, and that having team lunches and retreats is going to solve this, but this is so much change, guys, and not everyone's going to be with it, and you just are going to have to one way or another brute force your way, or the rest of the world's going to bypass you. So thanks everybody, really appreciate it. The best of times, but different, and we'll keep going from here.

In SMB, it's like everything's just cranked to 11. Every day matters. They're always a little more under-resourced. They're always trying to operate a little bigger than they are. My team is ready across the globe to lean in with every small business, driving a ton of success. We're seeing companies go fast and redefine redefine what's possible with technology. This is the platform that's going to augment your human workers to do incredible things. You're going to see SMBs with 50, 100 employees with multi-billion dollar revenue streams. There's never been a better time for our platform. I still feel the same excitement. There's so much we can automate and it's only going to get better.

Please welcome back to the stage Amelia Ru, along with Adam Alfano from Salesforce and Eitan Saban from PayPal.

All right, go!

We are in.

All right, hello!

Little delay on the way in there, but we got it.

They were like, do we come now? Uh, I'm just going to get my laptop up if you guys don't mind.

All right.

Demo.

How's everybody doing? All right.

Woo-hoo!

Straight into the demo.

Well, maybe not straight in.

What a professional.

We're going to demo for sure. So the name of the game today is we're going to be talking about how Salesforce is using AI to turn SMB into one of its fastest growing segments. I'm so excited for this session. Because I also kind of have now personally gotten a relationship with both Adam and Eitan, which is super fun. So Adam, who works at Salesforce, I've gotten to know just from like the Salesforce team and doing more with our agents. And then Eitan and Jason actually used to work together, for those who don't know. So this will be fun. We're going to talk about our agents. I'm going to, I'm going to try and do a little bit of some demos. Yeah, here we go. So that you guys can also see what we're doing here today. So let me go full screen on this guy. Perfecto.

OK, cool.

So I'm going to jump around a bit. So just forgive that it's still in Google Presenter mode, because I want to actually pop into my Salesforce in and out. So the other thing I'll tell you guys real quick is, while we're up here, I thought it'd be fun if you guys want to submit Q&A. I'll try and save some time as we have it at the end. So if you guys want to submit Q&A as we're chatting, you could just scan the QR code or go to sasterannual.com/q&a. .com/qa. I just added a little feature with my Vibe-coded agent this morning and said, hey, can you add some audience Q&A? It'd be really fun if people can submit questions as we're chatting. So as you think of them, you can submit them. You'll literally— you'll see me bring them up on my computer later, and we can take some fun questions to the end.

Yeah.

Awesome.

All right. Should we do it?

Let's go.

Okay, let's rock and roll. All right. I want to talk about the numbers. So let's start with this. Salesforce, and Agent4 has closed thousands of deals, particularly in the SMB and the startup segment. So walk us through that, because like a year ago, you know, you had more humans, but now you have humans and agents selling together.

Yeah.

How has that fueled the growth that you're seeing in your—

I mean, it's a lot of fun. It's— you're right, it's thousands of customers and it's every industry. There's some incredible use cases. Like we have a heavy equipment company, Equipter, that's got an outbound agent that's delivering emails about 2 times more effectively. We've got a healthcare company like Mimit Health that's doing patient onboarding with agents and surgery prep. And, you know, we have use cases like awesome AI companies like Windsurf and ElevenLabs using Momentum to do revenue orchestration.

Anthropic was here yesterday, right?

And Anthropic was talking about— and look, it runs the gamut. But I think what our charter has been for SMBs is to really make these agents accessible make them safe and secure and easy to execute in every edition. So if you buy our like starter edition now, you have AgentForce in there, it'll write emails, it'll analyze data all the way to like really sophisticated use cases around help agents and SDR agents like you're using. And I would say, you know, when we started, it was this very broad aperture, this kind of horizontal agent that could do just about anything. It still can. And we'd go in and talk use cases. What we're finding with SMBs is like really targeting the use cases around either SDR, web agents with qualified—

So go-to-market.

Or outbound agents, or we use Slackbot every day. I work in Slackbot, I don't really log into Salesforce, and that's how we analyze and manage my team of thousands. So really the goal is to make it really instructive. You go into Slack, your agents are there, you can execute in a headless environment without having to go log in and root through tabs and fields, and it's really fun. I mean, you're using it every day, so you know better than everyone.

We really are. Yeah, I'm gonna show people part of our headless Salesforce in a second, 'cause I feel like it's something that once you wrap your head around it, you're like, "Yeah, why do I need to log in actually?" Yeah. Well, one thing I wanted to ask both of you, 'cause I think it ties into PayPal very well as well. So somebody came up to me yesterday, I was showing folks how to vibe code like an AI VP of marketing, just something light that they could plug in with their, I showed them how to build like a little Salesforce connector, so that they could get it all working. But—

This was Raplet, or what were you—

It was Raplet, yeah. So a really great question came up, though, from the audience Q&A, which was, can I do this if I have messy data? And so I want to get both of your takes on that, because from PayPal, you have so many reps. And then I want to get it from a Salesforce perspective, too.

Yeah, do you want to start?

Yes, absolutely. So raise your hands if you used PayPal in the last month. Venmo, thank you for your— Venmo. Thank you for your business.

I think you're going to give money away. Is that where this is going?

No?

I can't commit to that now. But here's the thing. Think about an activation use case for our SMBs. We talk about north of 100,000 merchants that are onboarding on our platform on a monthly basis. And we need to get those entrepreneurs, those SMBs, to be up and running with our business. One of the things that we realized that many of those merchants are not able to process or actually were not processing or stopped processing during this journey. There is no human power or actually there is a massive human power that, or workforce that will need to handle all of those leads. I'm talking about 8,000 of them a month. For us, it was about how we can bring agent force for this case with the great help from the Salesforce team to help us to get in reach to each and every one of those merchants. And what we were able to see is incredible conversion that humans were not able to do. But the cool story here is the interface between human and the agent. We get the agent, We get the agent to do the heavy lifting, setting up the meeting, and then for the sellers to spare their time on really what matters. Have this conversation with those, with those SMBs, with those entrepreneurs that really need the help in the right time.

And for context, how many reps do you have now at PayPal on the sales team?

Within my organization, 200 reps. 200 reps.

Okay.

So think about the activation use case and the FDRI use case that we deployed 14 weeks ago.

Okay.

Fully in production. We stopped talking about sandboxes or talking about AI. It's about the execution for our use cases to drive the outcome. So in a short period of time, you were able to see full production, obviously with a ramp up, Similar to humans, right? We'll be able to bring humans, you know, there is no seller in Salesforce that will be able to close a deal in day one, right? Same for agents. And think about the meetings, that the amount of meetings and the conversion rate, it's about 50% higher than humans to get those meetings right there in the right time.

So I want to dig into that a little bit. So for those who missed it, so Eitan's got 200 folks, just to recap on his sales on his team. His meeting conversion rate went up 50% already with agents. It's only been 14 weeks. I want to— like, how did you get your— how did you get that conversion rate? Like, did you already have good data? Like, did you have to force your reps to put their data in? Like, how did you get them all to conform to working with the agents?

And that's the beauty about Salesforce. Look, we run on Salesforce. I built my career on top of Salesforce. I think many of us here ran on Salesforce. The data is in the CRM. We know what caused a lower conversion rate. We connect to other systems between Seismic that hosts all of our contact data. The agent force really helped all of those data points. We take all of our conversations from Gong and the transcription of all of those conversations and we know exactly why we were not able to drive a higher conversion rate. So now the moment the agent learn from all of those data point, I just gave you two, two for example, it's much more diligent, disciplined, and thorough understanding of the whole process to allow us to drive the conversion rate higher.

You asked the question about can you deploy agents on Messenger?

Data, right?

I mean, my take would be, if you're out there and you're trying to figure out your agent strategy, I wouldn't wait to like solve for world hunger on data while you get your agents set up. There's a bunch you can do with agents as you figure out how to get your data together. Now, if you take Saastr as an example, I think you have like 70% deliverability with our SDR agent, which is better than your humans are doing. And you and Jason always say one of the reasons is because it has an omniscient understanding of a data environment that's really well curated, right? So that's a, that's an optimistic, idyllic place to be. But the truth is, we go into a lot of companies where, you know, they're still building their data strategy, they're kind of consolidating, they're thinking through how they're going to align data, but they can set up a web agent that can understand their website and their FAQs and their product data and other data and still have a really great, safe, accurate conversation and bring in rich data and start to accumulate better customer intel while they figure out the rest of it. So I think you gotta just, you know, there's so many great companies in here with different agents you can use. You gotta roll up your sleeves. When you think about the tooling the agent's gonna use, that's where I think it gets interesting. Um, if anybody's followed along with Salesforce a little bit lately, we've been talking about headless a lot. You know, Salesforce has traditionally been a good place, in my opinion, for humans to do work in Slack and Salesforce and otherwise. We think it's great tooling for agents, right? Because you're building these awesome tools, whether it's with CLI, whether it's with some kind of coding tool, or maybe it's our agents, but the accuracy is good. Accuracy and outcomes are gonna depend a little bit on the tooling. And so we have a lot of customers that have built apps in Claude or maybe in Replit, and now they're faceting them to a, to a headless Salesforce, activating them in Slack, and they know they're gonna get 100% determinism on accuracy. They're gonna get better data integrity. They're gonna get better execution, better deliverability like you've seen. So I guess my answer would be you have to do both.

Yep.

In our case, my team's like, I've got about 3,300 account executives in SMB and their motion, their stack for agents looks like we have qualified.

Yeah.

On the front end is like the web agent running the SDR execution. We have an outbound agent that gets to leads that we don't have the person power to nobody was gonna call these leads back, and now an agent's doing it, and we'll do tens of millions of dollars on that channel, and it was honestly business that no one was gonna address. Slackbot is like the hub for where they go to get all their intel, they can prep a meeting, they can write emails, we run our cadences through there, everything that we need to do to execute, and it really unlocks the entirety of the intelligence for our company, and then the help agent is like our CSM for scale, and it's allowing us to get to, A bunch of renewals that again, it's tens of thousands of customers, we can't get to all of them with the kind of precision we want. Now there's agents that can do it, and that'll evolve. Ideally, we get to even broader human out-of-the-loop processes, and the sales reps and GTM engineers and everyone are out in front of customers and really having enriched discussions, and the agents are cranking on large-sum processes all the way from prospecting to close, and that's the goal.

Yeah.

Go ahead.

One point about the messy data. The messy data, I believe, will continue to be there.

Oh, forever.

And it's always relevant. Exactly. And it's also relevant for us.

Because agents write messy data sometimes.

Exactly. And the point I'm trying to make is think about it's also relevant for humans. Humans always jump between one app to another to figure out what is the right answer. But when you bring an AI agent to help to figure this out themselves, the outcome for a seller to really sit in front of the merchant—

It's more consistent.

Rather than figuring out the data and get the right outcome from that. That's the key. And that's where it kind of transformed us from moving away from experimentation to bring that to life in a full scale of 200 sellers.

Yeah, and there's a couple vectors on the data I think you have to think about. Like the structuring of the data in terms of like the codified processes that the agents are drawing inference from and then writing back to becomes important. And then I think the other thing we're learning, and you know, we saw a lot of this with Gong and we have Momentum, is that conversational data, both human to human, agent to human, it's gonna be agent to agent, that really informs back the dataset as well. And a lot of the orchestration is coming off of the transcript and conversational intelligence. And so, I think you start to corral all of that, you're just gonna have an agent that has, significantly more contacts that can do a way better job solving problems. That's, and you know, you're certainly one of the best examples of that.

Yeah, we've said, yeah, we've kind of said now twice, we actually are trying to hire a human to report to our 10K agent because he has a lot of, I don't know if you guys saw the session yesterday, he has a lot of ideas in go-to-market. And I'm like, I can't keep up with you, dude. Like you're working 24/7 and I still need a little bit of sleep sometimes. So one thing I wanted to run folks through, 'cause I feel like this is a really good grounding exercise, and I know folks took a lot of notes yesterday, is I'm just going to show you some of our agent force, both on the native side here, but then also how we use a headless. Because we didn't realize till the other day that Jason doesn't even have an account on Salesforce, but he uses it completely every day.

Yeah, by the way, Lindsay's over there if anything goes wrong.

So for those who don't know, my SDE is over there if I need any help. But, um, so yeah, we built this one and deployed it back in 2025. It's funny, rereading it, I was like, man, We could actually beef this up a lot more now, 'cause it's gotten a lot better. But just to show folks, like, so I'm just in like the basic backend here of Agentforce. You can describe what you want your agent to do. So this was the very first one we built, and famously—

How long did this take again?

Well, Lindsay helped me, but—

Yeah, but it was like, it was pretty quick, wasn't it?

It was quick, like within a day, I would say. Yeah, within a day.

Nice.

And it, like, one of the things that made it quicker, I would say, is like famously Mark was like, well, you know, we're deploying it internally at Salesforce, for us to follow up with leads that, to your point, nobody wanted to follow up with. And I was like, well, if it's good enough for Mark, it's good enough for us. So we copied the same use case, and I said, okay, all right, embarrassingly, let me go look and see how many leads, even at SaaStr, we hadn't followed up with. And embarrassingly, I was like, well, this one spreadsheet has literally 1,000 people on it, um, which I was very embarrassed about. Um, but I was like, okay, I'm gonna give it to—

some of them are probably here, thanks.

I know, you I know, thank God, thank God the agent re-engaged them. And it was one of those things where like, you know, we routed it to humans, we try and do the follow-up, but you know, sometimes again, humans don't always check all the boxes in the way an agent would to follow up with every single lead. They, you know, AEs like to pick and choose which ones they want to go after.

Right, so in our process, actually there are 10 nudges.

Okay.

That we work with, obviously reaching out to those merchants, try to figure out—

To get them to—

to get into the meeting. At the end of the day, it's about the meeting for humans to work with those merchants. In the world which is human only, those nudges will fail. They often do. It doesn't really matter which type of cadence tool you're using. We're humans. We get tired. We don't work on the weekend. And so many other reasons that now for an agent to run religiously, almost, obviously in a level of discipline and rigor that cannot match to what humans will do. However, for the humans to take that meeting, to show up there in their best version of themselves and get to the meeting, convert that, a sale into an action, active live merchant, that's the win.

Yeah.

So for those nudges, I don't think there is any human that would be able to replicate that.

I mean, they don't sleep, they work 24 hours a day. But the other thing, and we talked about a little bit on the upfront, is it has a completely omniscient understanding of your data environment down to your business.

And every type of customer.

Exactly. And so what ends up happening with that is the deliverability, the execution, the context, is, from my experience, better than, like, an average human.

For sure.

Not as good as the best.

Yeah.

But better than an average. And that makes it really powerful. But it also doesn't really make mistakes. Like, partly that's the determinism. Like, a lot of the tooling that folks are building with AI, it's really clever. It does interesting things. But it's taking advantage of, like, a probabilistic construct where it's probably going to do the right thing, but it also very likely may not, or possibly may not. And what are, especially when you get to like regulated industries and stuff like that and they're using Agentforce and why they pick it is they still have the determinism in there to make sure that it's gonna like sort of govern and drive the compliance they need to, but still be able to, you know, really get into the depth of the system to understand the nuance of data to send the right message at the right time.

It's funny when you think about the training time too for agents, like yes, like we've talked about, okay, your data should be in a, to your point, like I would say 80 to 90% in a good place. Like what I told the folks yesterday that asked, I was like, you know, I spent a day cleaning up our Salesforce and then I realized a bunch of stuff wasn't in there. And I was like, you know what, I'm gonna give myself a day. I'm just gonna clean it up over the weekend. And then when I'm done at the end of the day, that's where it's at. And that was fine for AgentForce, right? Like it's never gonna be perfect. It's never gonna have everything. And then I was like, I'm not even gonna sweat the contact data because if it's from a year or even 2 months ago, those people might not even be there anymore. They might not even be relevant for the agent.

I think you got to have an action bias right now. And, and, and I think actually you start using the agents and that becomes an instructive way to figure out what data and construct you need around them too. I think sometimes it works in reverse. Like, I don't know, it's a pretty boring project to sit around and try to stitch every— like, I hope everyone uses Data360 and other tools, but we are— the, the thing we're seeing now, which is kind of interesting, is smaller companies that have built their aperture on a Claude or a Replit And now they're kind of like retrofitting in headless to bound it to processes that are more reliable.

They may even just buy one API.

So they used it. They built something cool. And they realized after, like, oh boy, this is cool, but it's not accurate.

It's not—

Yeah, exactly.

I don't want to say that there is an endless amount of leads within PayPal. But I mean, obviously, my number will grow up, for example. For, uh, for next year. I don't want that. I actually, I'm okay with it. But here's the thing, humans will pick and choose the right opportunities for them to work.

Of course, they're gonna pick the ones they think are the best or the ones where they can get the most commission.

Exactly. And what they, what AgentForce allows us to do is to work exactly against each and every, every opportunity matter.

Yeah.

And that's become, uh, the execution discipline to make that we're touching each and every one. There is no— we're not— think about the close-loss ratio while we bring AgentForce to life. It's a completely different ballgame from where we've been prior to AgentForce, but also bringing the humans, our sellers, our most precious asset, to work on what really matters. We get to the meeting with the human.

Well, that's right. And we're starting way further down the funnel because they're getting a high-quality lead that's already been qualified. It's like, you know, it's booked in context. Now they have all the context from Slackbot. They can go in and have a more high-impact discussion. Conversion rates go way up.

And think about one of the things we witnessed again in this short period of time is the fact that we know exactly what to talk about with a merchant. Yeah, you set up the meeting, The agent also helped us to get prepared for this meeting. It's not that we're showing up to a meeting that, who am I talking here with? Think about all of the metadata around the industry that we talked about earlier, about the merchant that we— the details about the merchants that we talk about, their history with the company. It's something that humans that generate the meetings for us will not be able to capture. So it's a much wider nest of information and valuable information to, uh, to make this meeting valuable. Valuable.

Yeah, yeah.

It's—

we've talked about it before too. Like, this is something where, you know, David and I used to spend a lot of time qualifying leads that came in, and then when we had more folks on the sales team, you spend— you end up spending a lot of time either picking and choosing or qualifying the way that you think you should. But the thing is, with an agent, like, we've had QualifiedUp for a while now, and like The qualified inbound agent is really good at like asking it, asking the prospect certain questions on your website, but based on the actions they already took. So it's like, all right, I already saw what they were doing on the website. I already saw what pages they were bouncing around to. I saw what they were reading on Saastr.com. I see what they're doing now on Saastr Annual. Like, let me make some educated guesses and then have me prompt them to try and get a little bit more information so that by the time it gets from the agent to Amelia or David, We have a lot more information that we're working with of, okay, one, this person already booked a meeting automatically with Qualified versus us trying to say, okay, hey, you reached out to the website, we qualified you, and then we'll get back to you, and it's already been a day. It's instant now, right? So we wake up with meetings booked, but also that context of here's why they booked it, here's what they were looking at, here's probably the talk track you you should give them based on what you've closed already.

100%. Speed to lead in the market space I'm in.

In the AI era, dude, it's—

100%. And these SDRs are now focused on way more high-quality things. Their conversion rate goes way up. They're actually doing a part of the job that's much more enjoyable than mining through low-quality leads that aren't ready to convert. So our team, we got a couple hundred SDRs globally that are doing that job, and they're working hand in hand with agents all day. It's incredible.

Yeah.

One thing I wanted to show folks, so this is still our agent that's running. I'm sure I'm gonna build more post-annual, but this is still our original outbound agent that's running. So you can see we built it again fairly quickly, but I think to your guys' point too, like this is something where this context that we gave this agent on following up on old leads, giving them the context of, you know, some, hey, some of these leads are a little bit stale, like be probably more personable with them than maybe a person would when you're doing outbound, because they might be a little grouchy that we ghosted them, and we never followed up with them.

Ask about the weather, you know?

Yeah, it's such a great point, and I call it the maturity of an AI. Yeah, the maturity of the agent. The agent that we use in week 1 against those 200 leads that we work with is a completely different agent than the agent that we work with 8,000. Oh yeah, in a month we'll work with 80,000 leads on a weekly basis. So that's where, you know, this maturity journey is such an important to learn and teach, troubleshoot. Another thing that's really unique for obviously a larger organization is PayPal, highly regulated, is to bring the teams together.

Yeah.

You know, in a PayPal environment, you will never be able to kick off an agent without work with, with the marketing team to approve the agent, without legal, without compliance. Bringing the team together to make sure that we all bought into the outcome of this agent is critical. That would be one of the key advices of what the learning from the last 14 weeks. You cannot run it yourself. You have to bring everyone across the organization to help support that.

Yeah, so I think for folks that are in the room, when you're in the middle of the spectrum maybe, right? So we're at the smaller end of the spectrum. We've got like 3 humans and a dog. So for us, we're like, "Well, Amelia and Jason, we'll just vibe it ourselves and hook up our agents ourselves." So we have all the people in the room, to your point, but at a bigger org like PayPal, you've got to get everybody bought in very tactically to get your agents working so that you could roll it out and scale it out to the rest of your team.

I think that's a really, like, and that's a great example, but if we were sitting here a year ago, we probably would've been talking about like agent use cases and how you build an agent and that would be intriguing. And now setting up Agentforce or all some of the great agents on the planet. Yeah, it takes 2 seconds. It's not the hard part. But I will say like, as somebody with hybrid teams where the agents are doing pretty large sum human out of the loop processes for a lot of things, it definitely takes some orchestration of the team and it takes a lot of orchestration of the agent to. And I think that's the part that's— this term has been used a lot now, but that, you know, this AI is magical, but it's not magic. Yeah, it's pretty darn close. But if you don't care for these agents and curate them and onboard them the right way and tune them and continue to evolve them, then they'll kind of run into the same flaws almost like a human employee would if they weren't properly trained and provided the right tooling. And so, you know, our SDR leader Vanessa has just done a phenomenal job you know, thinking through the intersections of these teams and trying different use cases where the agent does a little better than the human team and vice versa. And I think that's the thing that you and Jason are doing such a great job of educating people on. Like, you've burned the boats. It's like you're all in on these agents, right? That's the teammates.

That's the team.

And it's really instructive to how we need to start to augment our teams to work with them, because a lot of companies have a great agent that works fine, right? And then stumble a little bit on like how they integrate it into their real workflow. Process and bigger teams like yours, it's, it's tricky.

And your example about Vanessa within, within Salesforce helped to distinguish between AI as a feature to a system. Yeah, I think what Vanessa were able to do at Salesforce and something that in many ways I'm, you know, following the same playbook is to build an AI institution that bring together all of those use cases and help to orchestrate together. That's a function.

Yeah.

That we have to relate to that not as a feature, it's a platform that connect the organizational, um, changes, but also to bring the outcomes together.

Yeah.

And even more than that, to have a number against it. In the end of the day, there is an outcome that we will—

yeah.

—need to drive for the organization. So an agent have a quota to to carry. Yep.

I was talking to Dan Darcy. I don't think he'd mind me sharing about his, like, customer success team and how they're onboarding agents. And I don't remember exactly how he said it, but it's like, you know, these agents need to be cared for. They need like a virtual mom to, like, get them on the right track and steer them in the right direction and give them the right aperture. And I think that, that hybrid team structure is, is something we really want to help our customers with. And some of the folks sitting down here thinking about every day, but it's— it definitely takes some iteration and some testing and tuning. It's a new way of working for them.

I'm constantly— well, I mean, Jason, it's funny. Somebody asked me the other day, like, how do you know when you're building something in Replit and then Jason's building something in Replit? And I'm like, well, he's not— he's not on right now because he's probably back there. But like, if I'm building something in Replit, you'll see their little bubble pop up. So like, we can see who's building in the agent and talking to the agent at the same time, which I think is really nice. But it's pretty funny because I was like, there's There's times where we're like trying to tell the agent different things and I'm like, oh no, I'll come back. Like, I'll delete, I'll come back. But one of the things I wanted to show you guys on screen is I asked our TenK today, who's our AI VP of marketing. So if you missed the session yesterday, I had a whole hour on how to recreate something like this using something like Salesforce to get, you know, interestingly, a dashboard, but also a lot of other data stored into something like this is I asked TenK directly, the agent, and I said, hey, can you recap me how you're using and deploying Salesforce Headless? Which I didn't even explain it. You can say it right here. I did it this morning. Hey, 10K, recap how we're using Salesforce Headless.

It says— Very trite question. You really—

Which I was like, is he going to know what Headless even means? Like, that's such a human thing we all said of, like, you know, basically not by logging into Salesforce, you're using it headless via the API. And so he said, here's my recap, meaning the app can talk to Salesforce via the API and the authentication that's in here. So he's explaining a little bit that he's authenticated to Replit. So Replit has a native Salesforce integration. But I explained yesterday, we also hooked up a Salesforce Connected app because it kept expiring and I didn't want to expire. So he's got both integrations. And then he's describing here what he actually does for us headless, which is He's pulling in all the sponsorship pipeline data. He's got all the media pipeline data that we have, our full deal list, full historical sponsorship revenue cards and data. You can also chat with the Salesforce data. So I think that's the biggest thing here is like, not only does Jason not have a seat in Salesforce, we have an API seat that's hooked up to this. He can just, I've seen him do it in the agent, like he can just talk to our data. He literally was doing it the other day, like, hey, Pull up sponsors that I, you know, that have competitors here that are bringing a bunch of people to SaaS Annual this year. It did it. And then I was like, great, now email it to that person, like automatically without— again, this is the CEO of a company, like going into the data that the go-to-market team has put in and with the help of the agent is able to execute a full campaign.

Yeah.

Yeah, I think it's really cool. And this, this aperture is exactly what we're talking about when we talk about headless. My vantage point is more Slack. I'd encourage everybody to check out Slackbot. One beautiful thing about Slackbot is it's got all the inference and it's unmetered. And so I think it really creates like you can just take the thing for a ride and do whatever you want with it. You're not going to get hung up on tokens. It's going to really have depth of understanding. But either way, wherever you come at it, it's understanding the data construct. And you did that great study on sort of the most AI-ready, agent-ready platforms. The whole platform's MCP and API enabled and you can just go have at it, which is, I think it's really exciting. All right.

Now we're out of time.

So any parting thoughts for folks deploying agents? With that parting thought. Any last thoughts for people deploying an agent?

Just do it. Seriously. You know, we talk a lot about AI. Obviously there is a lot of fear about— the question is, and I don't think the question is whether AI going to replace people or humans? The real question is which people AI will make irreplaceable. Yeah, that moment, the moment that you deploy, you iterate. It's similar to humans. It takes time to onboard a seller. It takes time to onboard any employee in a company. Same, same for those agents.

Yeah, I just want to say thanks. It's awesome to be here and thank both of you for being seeing awesome users of AgentForce and headless. And I would say an action bias is the key. We've said that a few times. To the extent you want to work with us and take on some of what Amelia's done, we have FDEs ready to go. Actually, Amelia's FDE is at our booth this week. I don't want to sell her out, but you can talk to her about how we built these things.

As long as I'm at the top of her list.

Part of this is a partnership of us just innovating together and cracking these things open and solving problems. It's a very different way of working. And we've been at this for a year, and all kinds of different repetitions on how to set these up and execute them. And we're getting really good at it. So I would say have an action bias, get working. You're gonna hit barriers, you're gonna have to reconstruct some data, you're probably gonna have some things you need to address, but you'll do those, and we wanna help and provide some talent for you with our forward deployed engineers to go do it. So let's go do it and build some of this cool stuff out. Let's do it. Thanks for coming, guys. Yeah, thanks everybody. Thanks, Amelia.

It's not a sprint, it's a marathon. 11-something years ago, we were 7 people in a room trying to figure out how do we scale a company. We previously had to convince people data and AI was important. Now we're getting, how do I move faster on this? How do you get all of that insight into their hands? Over 60% of the Fortune 500 are using Databricks. Getting a chance to be on the front lines of watching that journey, we do it because we're passionate about the company, what it is, because what it could become. How do you think about the future? No one tool will be the end-all be-all. So how do I leverage all of them? We keep adding new capabilities and we're just at the start.

Please give a warm welcome to the co-founder and senior vice president of field engineering from Databricks, Arsalan Tavakoli.

Thank you. All right, thanks everybody. I think, thanks for Arsalan for coming. I think this actually will be one of the, a perfectly timed discussion because we have had an intense journey here this year on building agents, deploying agents, what it means. But not everyone's selling to what, what is Databricks, 50% of the Fortune 500s, 110% of the Fortune 500s? 500? It's a lot.

Let's go with 110%. I like that.

You know what, for real. So Databricks is, um, I mean, they do— even though they're not public yet, they do— it's $5-point-something billion, reaccelerating at 50 or 60% growth, massive penetration in basically all leading enterprises, both pre-AI and AI. If you guys haven't followed the, the Databricks journey, um, uh, it's an incredible— and we're going to talk about it— an incredible lean in to AI tailwinds, incredible budget attached to it. And no one, I think, has better visibility into what enterprises are really doing versus what they say they want to do on Twitter. And so I want to dig in on what's, what's, what's really happening with bigger companies, what's going to be happening, and how the world's changing. And just as a side, it's just kind of fun. And then we'll dig into it. I mean, the world, you know, The last time we had any founder from Databricks coming, I certainly had never used a Databricks product. Today, ironically, you never know how the world changes. Every single action at this whole event is on Databricks in a sense because it's on Neon database and everything that Databricks brought. So every single thing we're doing, it shows how the world changes from the bottom and we'll touch into that. And it's a pretty cool story. But Arsalan, let's step back for a minute. What, what about AI everything that's happening on Twitter, especially maybe less LinkedIn, but on Twitter, what are folks exaggerating, getting a little bit wrong? What are you seeing for real from the front lines? Sure.

And first off, when you said this like a talk show host setup, I didn't know what you meant, but this is, this is an interesting one for me. Look, I think if you sit on X, it's kind of interesting. You'd get this viewpoint that everybody has figured out You know, AI, everybody's like following, like, I love Andrej Karpathy, but everybody's following him. Everybody's building their own LLM resource agent to do everything. Yeah. I think the rest of the world is definitely a little bit different. Like, you know, still how many of them actually properly adopted AI in the way of coding tools or like systems? It's pretty nascent. But every one of them, we were talking about this backstage, is realizing every CEO says if we are not adopting AI, We are behind, right? And so now you have all of them telling their employees, if you're not using AI, we are behind. You've got to figure it out. We're going to measure your performance by it. Go use tokens. And so now we have this other problem which is happening where everybody's like, okay, all my employees are token maxing. My spend on tokens is going up, but I have no idea what I'm getting out for it, right? So I think everybody realizes how important AI is. Outside of though, We're going to give them access to pick one of the models and we're going to start using coding in our engineering team. It's still early days for many of them and they're all looking for how do we redesign our processes and how do we actually think about governance and data to make AI work well, right? So I mean, that's what we at least see in many of the organizations we work with.

So let's step back at a high level. Maybe a few years ago, I imagine, and correct me if I'm wrong, imagine you're meeting with some fairly senior level buyers, CIOs on down. And the problem they want to solve is, I want a data lake, I want to do something with a massive amount of data in my enterprise. I actually want to be able to access it and do something with it. I'm oversimplifying, right? Yeah. Has that changed in the age of AI? Are they looking for something different? What— how has the urgency changed or the discussion changed with buyers?

I think urgency has changed significantly, right? Like, you know, we used to joke about this that in the, you know, Databricks has been a data and AI company now for what, like 13 years at this point. And we used to have to, we would scream the data part, we'd whisper AI, 'cause AI people were like, that means like self-driving cars and iRobot, who cares? And so everybody wanted to do it, but the answer was, look, I wanna bring in my data and it's helping me replace my data warehouse or my data engineering system to be able to get good analytics and fine. Now what's happened is everybody's like, AI is a top line imperative. I need to use AI, I wanna drive all of it. And they immediately run into the problem of you'll hear different words like, I have data silos, I need an ontology, I need a semantic layer, I need context, right? And what they've realized is what's actually really, really hard for— in order to get AI work, I have to break down all those different silos. I have to understand, make my data high quality and be able to be accessible to agents, not human, to really transform this. So the urgency of I need to kind of get my data state in order and governance around it in context has gone through the roof because people realize without that they can't actually get AI to work in any meaningful way.

So does that mean you're being— you're being— so just walk me through that. I mean, back in the day, like Marc Benioff was classically great at going into a room and saying, we will solve your problem for putting your data on the cloud. We were just going to solve the problem. Yeah. What's that Databricks pitch today? So they have this anxiety, they want to get ready for agents, they want to get their data. What, how are you, what's the pitch that the big problem you're solving for the stressed-out CIO?

Yeah, so I think that there's a— it's good you're seeing if I have my elevator pitch memorized, right? So I think that there's a couple of different points. You talk to every organization, they're like, AI is gonna transform it, I want intelligent apps. And by the way, I don't want intelligent apps just my engineering team what it's building, I want I want sales, I want finance, I want my like merchandising team, all of those being able to use it. But I struggle. I can't today. And the main reasons that they say it is like, A, there's way too many things to choose from. Like I have like, am I picking one of the different hyperscalers? Is the right answer going to be do I pick Anthropic? Do I pick OpenAI? Do I pick Gemini? Do I pick open source? So one is like, which one of those do I build on to move forward? Second, like my data spread out. Help me there? Third is, can you actually talk to me about context? And context is different than data. Think about it as you're onboarding a new employee. How do you explain everything that goes on in your organization so they can operate effectively? How do you get that? Because that's what agents are. And then lastly, how do I control this? Because agents sound great, but my favorite was everybody was like, open claws out. So what I did is I took a Mac Mini, disconnected it from the internet, and put open claw on it. And I was like, Great, congratulations. What does that do? So how do I unleash all of these but make sure that governance works well? And so the Databricks pitch is like, that's what you want to do and build applications on all of those pieces, right? We break down all the different silos, we allow you to work across any of the models or hyperscalers, it automatically builds context for you, and then there's kind of world-class kind of AI and data controls like Unity catalog and gateway across it. You put that in and it's built on open source. So if after I leave that meeting they're like, Arsalan's jokes suck, we don't like them, we want to use a different vendor, there's not some big migration. That which says you can move fast without being locked in is huge, um, for the organizations that we work with.

Let me just push on that a little bit. I want to come back to governance, security, compliance, because it's underdiscussed, right? I do want to talk about it, but I want to understand what context really means. I know what context means to me building apps, right? I know what— but how in 2026, yeah, what does that mean to a CIO? What do they— and, and maybe what that really means— I assume what that means is they want to do more with their data than they could do before. But yeah, what is it? I mean, context is super important for agents, but I want to know what, what a not Not what the most bleeding-edge tech CIO wants. What does 95% of the world mean when they want you to help them with context?

Yeah, so look, I think that there's a couple of different ways to think about it. There is one in some organizations that you have, especially if you're in a regulated industry. Like, I was meeting with a large bank yesterday, and the answer is if somebody wants to, let's call it, talk to their data and ask questions, when you say, hey, show me what is my revenue, What does my revenue look like, or what does my loss look like? Those are not just a table called revenue. They have very, very clear definitions, right? And so you want to be able to map those somewhere, which is like, these are our core terminologies of what it means to our industry and our world. I think that that's one. The second thing is though, there's a lot of different things that are kind of, let's call it shorthand, that is not captured somewhere. I'll give you a very simple answer. If I ask, you know, and I'm talking to my data and I say, Hey, can you show me in the last fiscal year kind of what basically my top spenders were on the major clouds that we support at the end of the last fiscal quarter in EMEA? What's the answer? Sounds like a simple thing. Anybody at Databricks would know what I mean. But then this like, what does he mean by clouds? Yeah, what is fiscal quarter? What is their fiscal quarter? How does it— what is inside of EMEA? How do you think about top spenders? Like, these are all ones that you'd— if If you were normally in the company, you're like, "Oh, I went and talked to them, and they told me this term means X and Y." It's like, how do you build up that glossary? Now think about that across a 10,000 or 100,000 person organization, how many of those tidbits there are. They're usually buried in your emails, your meeting transcripts, your basically other kind of calls, documents. Then on top of that, you have to layer in, "Hey, I want to build all this, but also, Jason, you have access to different things than I do." So your context, how I build it for you, that information is different than mine. And if all I want is I want to be able to ask a question and get a perfect answer, when I ask that question, it needs to understand all of these nuances. So how do you build that up? And that's a hard, hard problem because people have historically done it statically, but it goes stale fast. So inside of Databricks, we have something called Genie Enterprise Context that literally looks at all the questions you ask, all the data you pulls in, and figures it out on the fly and evolves it. Because the other hard part in the organization is, let's take SaaS. If I say context, you're like, yeah, there's a document for 2 years ago, it's outdated already, right? It's like, how do you basically pull new information in and deprecate old ones? It's a really hard problem, but that's the key to actually having agents be able to work effectively.

All right, but let me, let me, um, let me just push on this. I wish I should have put a slide of a screenshot of Genie, right? Okay. Genie is at least like your master agent or your meta agent, right? Um, maybe it's the wrong term, but Specifically, now that you have this meta agent genie where I can ask it in plainish English how to access my data, you talked about win rate, loss rates, or whatever you did with the bank. Just about what can I do now in Databricks I couldn't do 12 months ago? Like, what has changed at the product level and for the buyer? Like, what does the buyer want out of you? It doesn't have to be 12 months, it could be 18 or 6. Yeah, but what's your step function in what you can do with your agent that you could not do before at Databricks?

Yeah, look, I, I think the following is right. How did you take any organization, um, I mean, like, how do you run the business? A lot of it is you want to be data-driven in some way, right? And so you're looking at it, and let's say, uh, as an example, something as simple as you want to make a decision inside of your stores, right? Like, I'll take a company that we work with, like 7-Eleven. You want to make a decision of what do I have to staff in my inventory in my stores. Previously you would have done something like the following. Okay, what are the questions I want? I have a dashboard. Somebody went out and built me a dashboard, right, and it has a fixed set of questions. When I— and A, that pipeline is kind of running. Every time I wanted something new, I have to go ask somebody. It's a 1-week turnaround, and so I get something static that every week it kind of comes up. I look at it. If I want a new question, I'm like, I would really, really care about now what about X, but it's gonna take me a while. I'm just not gonna do it. So now inside of Databricks, yeah, with Genie, what ends up happening is that all of the shop floor owners can just say, what are the top-selling things that I had yesterday at, you know, what discount? And so what are the main ingredients that I need to restock today in the store? And it literally goes out and it gives them an answer immediately. And so that means two things, right? You're able to broaden the set of questions that you're able to ask. You weren't before. And how many times have you been in a meeting and said, oh, it would be really interesting to make this decision if I had this data point, but I don't want to wait a week for it. Whereas if you can get the answer yourself, self-serve in the meeting— so many meetings I now see, people, they start the meeting, somebody has a question, somebody fires off the question during the meeting, the answer comes back, and it changes the course of the decision of what they can do. It's a huge deal of actually being data-driven and using data to make those decisions.

And but before the Genie agent, yeah, could the average manager or whomever at 7-Eleven access this data? Could you do it in Databricks pre-Genie?

You, you—

but so non-specialized democratize access?

Yeah, the short answer is, look, we had an organization— I'll give you a simple example, right? It's a large car manufacturer that if you take a look at who are the people who understand the data and can actually write queries or code.

Last time it was so few before.

Let's say it's like 5% of an organization. And so they can— so what would happen, you could get it, but you have to go find that data analyst and say, hey, can you run this for me? Can you slice up the data and get it for me? And it comes back as a dashboard, great. And if I have another question beyond slicing and dicing, do another round of that iteration. Now the answer is in Databricks, what Jeannie says is forget the 5%, the 95%. We just had a— that car manufacturer just loaded on an extra 70,000 users, they all are going and asking their own questions of it, right? So you now mean you've cut out the latency and cut out those data resources that you need to answer every other question. They can answer it themselves in real time and they know the business, they know the domain, they know the types of questions that matter, and they're on the front lines of actually putting it in operational, like, rigor, whatever that they need. That's a big change. That's a big difference. And we've now seen that organizations, you go from those small people giving BI tools, which are— I call them dashboard graveyards— to folks where it's like these like long dashboards, maybe a couple of them somebody looks at, to now everybody can self-serve. Um, we've just seen people be able to move in a much faster way.

I want to, I want to touch on the, the death of BI just for fun, but before we get there, is— does that— are literally an order of magnitude more end users using DataBricks because of Genie than 2 years ago. The story to me would— because I'm very passionate about democratizing access to data, right? I hate BI tools. I can't— I don't have the patience for them. I never used any BI tool until I could buy my own, right, on top of Neon, right? Now if you go into 10K or AIVP marketing, you'll see 10 dashboards. I had no dashboards 6 months ago, but it democratizes access to that. So I could imagine if you say 5% of the organization could could write a query before, I think it might be 0.5 or 1. But now if it's 95, you have 10— you potentially have 10 or 50 times more people that are going to use Databricks as end users.

Yeah, I don't think I'm going to say we've exhausted the whole TAM of the world now, but I would say end user TAM. Yeah, but we— so what I would say, we've clearly seen growth. And I think that the short answer I can now say is that if I walk into an organization There is pretty much, I mean within reason, there is on all of my look at at least 95% of their employees if not more, then I'm like they could now directly use Databricks, right? There's oftentimes like all other political and organizational things, but you went now for example simple changes, a lot of people will go into Databricks, you don't need to access the underlying complexity, you just have it and we even have a mobile app now, right? People take out their mobile phone on the go, they're asking it, it's really easy to build, applications, because I also think, you know, the future of the world is not going to be just chatbots, right? It's actually applications people can interact with. People are building those like incredibly fast. We have, I think at last count, well into the tens of thousands of apps that have been built on Databricks with interfaces for kind of those end users. They have it on their mobile app they're using that they're going. We've seen it either sales reps kind of, we've seen it financial advisors, we've seen it people on the shop floor like All of them are using it and none of those people know what SQL is or know what kind of medallion architectures mean. And that's the way it should be.

I want— let's talk about building apps and Databricks next. Before is— I don't know. I know I don't think Databricks charges this way, but at a meta level, is the fact that 10 times more folks can use— humans, not agents— humans can use Databricks. Is that a counter narrative or counter argument to the seat dying? I mean, there are certainly applications where the seat is dying. We can pull them up from public markets. Yeah. But if we're democratizing access to data, even as we explode the number of agents, if 10 times more humans use it, might we see a rebirth of the, of seat economics?

Yeah, but I think that there's two things. I think that there's seats and then there's seat-based pricing, right?

So, yeah, they're not the same thing.

No, meaning, so we had, I mean, it's what happened happens when you have as many co-founders that we did. The number of debates in the early days about should we use seat-based pricing, and we always, uh, said no, we are not. We said we are a consumption platform, people should use it. Um, and so I think in that way, uh, I do think you're still getting many— and one, you're getting many more people using it. By definition, they're driving kind of a lot of consumption. In the consumption world, you don't really care if it's 1 or 1,000 people as long as they're using it. And I think that also somewhat simplifies everybody's worry about going to agents, because as long as it's whatever they consume, the agents are consuming, your pricing is kind of— it doesn't care about the number of seats. But I think the sheer amount of usage in an interactive way is going up. And I think what you're also starting to see is, whereas people used to have, I have a simple question, now the questions that people are asking are complexity, like, can you go out and run, like, do a segmentation of customers and figure out what makes sense and what's their best propensity to buy, and then run a campaign and measure it and come back. So the volume of work they're kicking off is also more significant.

That's the Jeevan paradox of queries. Yeah. Once you can actually run a query, you want to run a much more complicated one, then you want to run 50 times more. You want to be in it all day long asking the agent what you can do. Yeah, exactly. Are you seeing folks in Genie or otherwise much more often than they used to be?

Yes, we are seeing that. Look, I think that the problem happened is almost everybody I know previously said, I have a question, I get in a meeting, we talk about it, action items, we'll come back in a week. Now the problem is that that same question you ask it and the answer comes back in like 30 seconds or a minute. And so you're like, okay. And it's like, you can never— what was that? You can never just have one, right? You just need to have a follow-up question. Yeah. So it's like, now I have a follow-up and then I'm going. And it's funny, I talk to people, which which is a side note, that they say that they're more stressed out in the era of AI because their utilization is higher, right? Because now you feel, if I can get an answer, I'm keep pushing it. What's the next answer? How do I spin up 5 different streams? So you're seeing the level of engagement and intensity definitely pick up from what it was before.

Yeah, um, maybe a smallish point, but, um, I think all this BI stuff's dead. I think it's Chegged. What I mean is, we did nothing wrong with Chegg for running textbook. We just don't need it more, right? What do we— do we need 95% of these standalone BI tools when we can just talk to our data?

Yeah, now you're just gonna get me in trouble with all the partners.

No, no, I always found BI— I've never really understood it because I've always found the tools very frustrating. For me, too hard to use, right? Just too hard. I don't want to wait for a week for a report. Yeah, I don't have time, right?

Yeah, so I look— I think that there is a couple of different things because we spend a lot of time debating on this. I think, one, what is the interface for you to kind of ask questions about? I think most people naturally believe that that interface is going to go to voice, right? Like, it's text and it's voice. It's just the most natural one. And who knows, if you carry far enough with Musk, it's Neuralink. I just think it. I don't even talk about it. Um, but look, I think it's, it's voice. And so clicking and dragging and dropping, like BI tools, like I clicked here, I did that— I think that part has gone away. But on the flip side, what I do believe is the way that our brain kind of interprets data, yeah, is still very much more purpose-built, not for text but for visuals, what kind of dashboards and what— yeah. So I think what you're ending up seeing is that there's tools of where like you interact, you ask questions, and results come back in terms of kind of visuals. But then you can double-click on it again without clicking and driving. Like, I'll give you a simple example. I'll ask I ask, you know, show, tell me what's going on, on kind of like what are sales, what does usage look for the last month, and I'll get a pretty dashboard because we can think through time series graphs, we can think through ups and downs and so, and then I see that there's a spike, but instead of double-clicking, I'm like, hey, there looks to be a spike here, can you basically help me understand, you know, why that happened and go dig, and it gives me something else. So I think that that's natural. I think the hard part for the standalone BI tools was, they historically really had no semantic understanding of the data, right? You did an extract of a data warehouse, they stood there, and when they tried to add talk to data, it was mainly just, let me convert text to SQL, which doesn't work. What you really need to know is, if I ask it, how do you interpret what I'm asking and connect it to what the data actually means? What's the best way to go ask it? And that's why BI tools that are connect— like in Databricks, we have these kind of business semantics that then all of the external BI tools can connect to. It makes it much more intelligent what you can visualize. And inside of Databricks, Genie, which you talked about, is integrated with the BI layer so you can ask questions. I think that's where the world is most likely to go.

I just think every, every knowledge worker, every human that's connected to the internet should have their whatever BI they want with the agent. Yeah, like it almost should disappear into dashboards and answers. Like we should have dashboards and answers, and BI is almost a term, it's almost just an input. Like, I shouldn't have to even understand BI anymore. It should be just part of accessing Databricks or others for everyone. Every single person should get real-time visibility into every bit and byte in their organization.

Yeah, you should. And the big important part is, which we noticed on the BI tools, which I think is what customers are frustrated on, is many of them, if you looked at it, whether you were using, you know, Power BI or Tableau or Looker, they all over time said, hey, we pull it out and we kind of built an element of a semantic model inside of the BI tool. But now you've locked it away in this proprietary silo. And so what we see everybody saying is like, guys, I want to build that semantic layer like just once, that context. So any tool, if I use this BI tool or if I use this other app, yeah, all of them are shared context. Or what it shouldn't be like, know-how is captured in one. And so I think that's a big one where everybody wants to— again, why they say, hey, in Unity catalog. Let me build that semantic layer. So if I want to use Power BI or Tableau or AI/BI and Databricks or something else, they're all giving me the same consistent answers. I think that is one trend we're absolutely seeing.

Okay, I want to— I do want to do compliance, GRC, and all that next, but I want to talk about— I want to insert a topic I know you're passionate about and how it— what social media says versus reality. Token maxing. Yeah, you thought about— you have your own models too we want to talk about. and building application Databricks. But I think just today, CIO or someone at ServiceNow— the other, the other day, I think the CEO of Uber said they exhausted their entire token budget. I think it was, I think it was ServiceNow today. Yeah. ServiceNow said they'd ran out of their token budget. Uh, I, I wonder how good they are at planning, but, but nevertheless, putting that aside, you've thought about this managing cost. What, what, what reality are you seeing? For token maxing?

Uh, yeah, so look, I, I think that there's a couple of things. What— let's first go through kind of different phases and then I'll answer that question. So when people ask me what does it look like for actually adopting AI, there is like 4, let's call it 5 phases that organizations are in. Phase number 1 is that they're not really using it, and I don't count your people asking questions of ChatGPT or Perplexity or Claude as like, that's That's, that's like using Google search, right? Like, okay, you're doing that. Phase number 2 is what I call the AI sprawl era, where everybody's like, I've been told my job depends on AI, I have to use it. Um, and they're measuring it and we have dashboards that we can go see. So I'm just gonna start building random things and go out there. And like, I have 25, everybody's doing greedy local optimization. My problem, I'm trying to solve it with AI and you have 25. Number 3rd is, hey, I actually figured out what are some of the problems we have and I've started to kind of use tools to automate it, to draft my emails for me or to draft my meeting prep. And you'll get a lift there. Let's call it like 10, 20, maybe 30%. But it's— you're still doing things in the same way. The real error is when people say, how do I do process redesign? If I were to design this from the ground up, I mean, you guys kind of did that faster. Say if I were to think from the ground up, how would I build in the era of AI? And it looks fundamentally different. How do I do that? And then that's how you get real productive. Productivity lift? And the last one is, what are the new things I can now do? So that's kind of the spectrum, right? Almost everybody you talk to is somewhere between 1 and 2, with some in 3. Very few have made it to 4, you know, or 5 in that category. And so what happens is everybody is using AI. And the problem is you give all these people the tool, you almost never show them what does good look like, how you should think about it. You just tell them, if you're not using it, you might be fired. And so what is everybody happening? Everybody wants to use more? And everybody's answer, because they don't understand how AI works, is for everything I need Opus 4.7 and I need GPT-5.5 for everything, right? And these things are really, really expensive and they have no way of predicting how many tokens people will use. And so then now they're getting into this token maxing that the budget is being burst. So instead, what you're now seeing people try to figure out how to do is it's hard to change users' behavior, but how do I put a layer on and how do I think about things like how do I do caching, how do I do automatic rerouting that look for 95% of the stuff that you're asking, you don't need Opus 4.7, you can use an open source model or something simpler. And so we've now seen a lot of people and in Databricks, that's what, you know, the Unity AI Gateway is for. It's like, how do you get usage to go up but you bend the curve of spend and bring it down, right? And so I think that everybody is now starting to see our costs are going up. We can't predict what it is. And when they ask the question, what's the ROI I'm getting out of it, you get kind of a hand-wavy answer on it as well. And so, like, nearly every conversation I have with enterprises, security and how do we measure cost is one of the top pieces for them.

Yeah, but, and specifically, and I should be more of an expert, but in Genie Code, when I'm using DataDogs Genie Code, yeah, that's built on open source models. It, it, it sort of, it picks models for different use cases or applications. And how do you blend that to manage costs and models?

It's a variety, right? Because like, so first off, the short answer is in most cases, yeah, you don't want people actually picking models. Now let's put aside the fact that like you have the US banks who want to regulate it and they can't use Chinese models and the like. Let's put that to aside for a second. But by and large, the average person is not actually qualified to decide Should I be using, you know, kind of Opus, you know, 4.7, or should I be using GPT or Grok's model? Like, you want to figure out how do I get the cheapest thing that gives me the quality that I need from it. I think that that's one. Second, I think that the secret is like, for many of these things like Genie Code, a lot of what you do, the value of it is actually not tied to the model. There's like, how are you intelligent about what the best way to build things are? How do you think about the best model to use, how to use caching. Like, the other thing that I would say is many people who use, like, let's say you use Claude, Coda, like, they all work. It's kind of the brute force method, which is like, if you were to say you have a blank check, you don't care, I will brute force it, use everything. There's much more intelligent ways if you want to save costs and do it. And things like Genie Code is super intelligent. What model do I use? What's the size of the model? What do I need to fit in? How do I think about chunking? So there's a lot of work in just the harness and kind of the intelligence around it that both give you better quality and much lower cost, plus the security stuff we talked about.

And I would think, and I'm not a total expert on chunking and all of that, only have done a little bit, but I would think at a meta level, if I'm running Genie Code or any query on Databricks, a lot of the value is in the Databricks data. Yep. Not in the LLM. We're using too much LLM and not enough data potentially. Like, we're over— we could be overusing Opus or something. Yeah. Um, when what I really just want to figure out is how can I get that data structured to answer my question? And I, I don't need to over-LLM-ify every single query, do I?

No, this is 100% true, right? Yeah. I think LLMs have a value, but I think that there's other things. Like, here's, here's the funniest part for me. Um, I don't think that there is anybody in the world who can tell you with 100% certainty that they understand exactly every— even the people who built it— exactly what every model does and why and what it makes that choice. And I'll tell you why that matters. There's certain cases of things that you do that you need it to be deterministic, and it is very hard when you give something and just depend on the LLM to get a deterministic output. It may not be big, and you have— and everybody builds evals and slight checks and all of those and anchor cases, but there's these slight variations. So you start figuring out where to use an LLM to build code that executes things deterministically. When do you use the LLM? When do you use reuse? There's a lot of logic that goes into that. I mean, we have a large research team that's just looking at how do I balance quality and cost and security at any given time for what customers are asking. Got it.

Okay. One, one higher level thing I, I, I, and then I will, I do wanna talk about GRC and all that, but that we talked about backstage that's so important. Is, and I talked about this a lot in the opener, this bifurcation between budget dying for pre-AI, B2B and SaaS, and exploding for AI. Databricks is clearly a beneficiary overall of this trend, like unquestionably. But your point backstage is there's a murky middle where, where it's not clear of all of your workflows and application which one you really are and how important it is to be on the AI side. So What's your learning advice there in general in working with customers?

Look, I— it's, uh, it's funny, right? I, I think I, I say that the only question I get asked more than when are you going to IPO now is like, what do you think of the SaaS-pocalypse? Like, it's literally what I get.

I said, I said just before this, we're over that. Yeah, it's time to move on. But it's a, it's a good—

I think you may be over it, but I can tell you every investor I talk to, they're not over it. I still get asked that question.

Well, they're just wondering which one you're in, right? Are you an AI company or are you a pastpocalypse?

I mean, I'm biased, but we're clearly in one category, right? But I think many organizations are looking at this and saying, look, frankly, you're either— AI is a wave that's going to crush you or you're riding the wave, right? And people want to spend a lot of money on the latter and they don't want to spend a bunch of money on the former. I think describing exactly why that's the case for folks is a little bit harder, but I do think a lot of folks are realizing I've got to go on the AI journey and I'm struggling to figure out how. And like the notion that I'm going to vibe code my own CRM and I'm going to vibe code all my own applications, I think that that is just not a reality, right? Even if, even if you could build it, maintaining it, evolving it, liability. I think what's going to happen is, which is an opportunity for many people here, I think any industry that is highly valuable that has a monopoly today will not have a monopoly 12 to 24 months from now because—

Slow that down. Any, any business, technology or otherwise, technology that has a monopoly will lose its monopoly?

I think that because the cost of building software has gone down, the ease of building it and figuring out, like, there will be more competitors in it because here's what's going to happen.

For sure more competitors.

You will walk into one monopoly, gives you pricing power, right? And you told me you wanted spicy takes, so this is my hot take that I'll get in trouble. It's helpful, not spicy. But like, so look, if I come— what happens right now today is if somebody were to walk in previously and say, here's your application, I'm a monopoly, here's what you're going to pay, you'll grudge, you'll moan, you'll threaten to sue them, but you don't have a choice. Yeah, inevitably you're like, well, I could hire engineers to kind of build that for cheaper, right? But you won't. But the reality is that there will be many startups who will, and because they can say We can help replace systems of record. We can look at existing code and workflows and map them, and we can put things. So you're going to see many things pop up. And that's what I think of the future is going to be, is that there's going to be a lot of lower cost alternatives. And so like the incumbents in that space are going to be pushed to say either they're going to have to reinvent themselves and innovate with true AI and figure out how to come with more places because there's going to be pricing pressures, or they're going to hold on to their lines and they're going to say, seat from the bottom up, a lot of their basically power base get eroded. So I think that's what's happening where people are eager of how many companies are coming out. They're more modern, they're basically more tailored to modern workflows. They understand that agents are going to be big users of software and building for that. I think that's a big part of the future. And enterprises are asking— everybody wants to know what will it look like in 12 or 24 months? And I want to make a bet that I won't regret in 24 months, right? So are they forward-looking as well?

Yeah, I think I, I'd be curious. I mean, I think I see two, two things to be curious about. One, I see a Cambrian explosion of applications. You can see them here. Yeah, it didn't even exist, that are great. AI is creating the SaaS. SaaS may be dead, or SaaS fuckups, but B2B and AI is the greatest thing ever in terms of app explosion, right? Every single number. I think people get that, although I think sometimes they miss how much competition that means. Yeah, right. The second thing that happened that is not— that took me a while, it took me a beat to get it— is the low-end competitors are so much better. It used to be the low-end thing was crappy. At least it worked. Yeah, but it only did two things. Yeah, I get one dashboard that sort of worked and that was it. One workflow. Now because of AI, like, I see a lot of low-end competitors in B2B where the agent makes them great, especially if it's also leveraging third-party APIs. If you can pull in Salesforce or Shopify or Databricks data on top of your low-end application, all of a sudden instead of being a crappy one workflow thing, sometimes they're great.

They're great, this small data. Well, but think, you are correct, but think about the following, right? Like what you're also, couple of things are happening I think. A, whereas previously if everything was a monolithic stack, the amount of effort it took to go build a monolithic stack and convince any type of enterprise to use you is really hard. But now if you look at it and say, you know what, I've got an organization who's already using, who has Databricks, and I know that they are already ingesting data from all of their SaaS applications and into it, right? So that data already exists. The— they're trusted from a security principle. We have like Databricks Apps, which gives me a container to build it. You know what, I can very quickly kind of walk into many of those organizations. Now it's a question of how well do I understand the domain and how well can I build, is one. And then second, I've got nothing to lose. So I go into one of these big domains where you have a traditional incumbent who's charging God knows how much. I literally just set my price point at What do I need to build? 30% of their price, 40% of their price. And I go in and to your point, everybody always thought you get what you pay for. But in this world, no, there's modern people who it's a new market. Everything is greenfield for them and they're like, this price helps them get into there. The other thing that's changing is imagine in the old world, if you spent a lot of time building around SAP or something of those sorts, to migrate from SAP to SAP is like a decade-long journey. right? The other thing that AI— really a decade— the other thing, and that's being, uh, considered— the other thing that AI has, uh, normalized is the speed of migrations and the cost of— like, the speed has gone up and the cost has gone down, right? Because you can now use, you know, AI to understand the existing environment, map it, and lift and shift. We see that in Databricks as the number of, like, legacy data warehouse migrations we can do. So if your cost of migrating has gone down, your willingness to give a to try on a new vendor and pilot it has also gone up. So I think that's why I think it's hard for monopolies to remain, because many people can get traction fast. Just take a look at how many of the people you have here. Yeah, that didn't exist a year ago, and now the revenue that they have is non-trivial revenue because people are willing to give it a try and to kind of migrate off of it. I think, uh, and not all of them will make it, but there's definitely more of a willingness to try new software in this composable web, which I, which I think is interesting.

Well, I wanna, before we go into the time, 'cause it was on our list we talked before, I wanna talk about this L migration because people talk about it on X, but it's actually, you're right, living it, it's really important. I mean, for an extreme example, these guys at Lightfield here, a next generation CRM, they have Mac Minis there, and they will migrate you on site, your CRM on site. Now I don't know how many people have done it, but they will do it for you, and they will give you the Mac Mini for free.

I'm gonna guess they just won't tell their CIO that they're doing it, but otherwise, That's, that's—

Yeah, but I don't, you know, this— I, I don't think a year ago that was possible. I agree. It's, it's under understood. And okay, so it's cool that they're doing it, right? Go try it. Go, go for fun. But Databricks is operating at a slightly different scale. You're doing this, you're going into customers. Yeah. And you're offering LLM migrations that maybe a year ago were not feasible or practical. Can you just describe, because I think it, because I'm, we're finally trying to get off our 72-year-old marketing automation platform Marketo, which I don't think was even possible for us 6 months ago. We didn't have resources. Now we can do it with LMs, right? So let's just talk about that before we run out of time, because it's going to accelerate, isn't it? Oh no, it's definitely going to accel—

look, let's, let's, uh, decompose a migration into kind of its anatomy of what it looks like, right? One, you need to go analyze that person's environment. You want— you need to understand everything that it actually is doing, right? And what are all the pieces. So that's one, you have to analyze it, what's the level of complexity. Second is that you then have to mig— like convert the code that everything was going to run. And the— and converting the code, everybody's like, I don't want to lift and shift, I want to lift, modernize, and shift, right? So you have to be able to take it and map it into kind of like the modern world of what you're doing. Um, third, you now have to migrate the data, whether it's the models or pieces underneath. And fourth, you have to then go and reconcile, is the output of the new system the same as the old one, right? And that used to be very, very long processes of how do you look at it, because many of these legacy systems, the person who replaced the person who replaced the person who replaced the person who knew it just is, you know, doing their retirement party, right? So how the heck do you find— but now you can send LLMs in because the beauty of code is code is self-descriptive. It goes in, it understands what everything is doing and why, and then it can also very easily convert it. And then the key is like, how do you write harnesses to do validation and reconciliation? So we've spent a bunch of time on that, and you talk about it from BI tools, from data warehouses, from data engineering tools and all the like, and it's made it far cheaper and kind of far faster too. Because the other thing was you'd go and say, hey, here's a system you were spending a million on, I'll save you 50%, but by the way, the migration cost 5 million. Well, it just died on the vine. But if you can say, I'll save you that much and I can do the migration in a month and much faster, the willingness to to do it has just gone up. So we do a lot of those now. We invest pretty heavily on that team. I think it's truly transformed how people can kind of migrate from— so you can just—

and then we'll wrap up because we'll go over, but you can do an enterprise-grade LL migration to Databricks in 30 days or less now and commit to it depending on specifically what it is, right?

But that's a big deal. No, I think what I would say is that the time of migrations has gone down trivially, and especially like you talk about a BI tool, let's think about it like I've got a BI tool, I have Tableau as an example, and it's like I have tons of dashboards and I have data models. The fact that I can just use it, map it out, migrate it to a modern one and do it— there's always elements of what do you want to migrate, what do you want to say, what do you not want to migrate, what process— those are the harder parts. But the actual technical part of the migration has picked up, you know, significantly.

Okay, and just two last questions, uh, and then we'll break on this because I, I think this I think this is, it's so important, especially if you're trying to compete, right? How do you think about moats when, because if it's easier to move over, how do you think about moats on the other side? I mean, Data was probably thinking about moats. And then lastly too, and then we'll break, how do you think about charging? Because traditionally moving, like there's been an implementation fee and a migration fee, and sometimes I want to blow my brains out when a vendor wants to charge me a migration fee, and they want me to pay for two vendors side by side while I migrate. Yeah, a lot of topics there. But just to wrap, how do you think about moats and migration fees in this when LLMs can do 95% of it?

Moats from just a migration perspective, you mean? Or just—

Yeah, well, if it's easier to put into Databricks, is it easier to take out of Databricks?

Look, I think Databricks, we've always been of this. So let me answer it. Well, this first one on moats, but like we, we have a bunch of like really cliché sayings, you You know, many of them are— the bad ones are my fault, but one of them is like, we lock you in with value, not with proprietary APIs, right? So even today, when you move stuff to Databricks, yeah, you don't move it in some proprietary walled garden. You put it in an open data format that— so if you leave the next day and you leave and you're like, Arsalan's takes were bad, his jokes are awful, we're getting off of Databricks, there's no migration. You know, your data, you have your data, you can go move to another system. And many of our components are open source. Open source, like the open source components of Databricks have over a billion downloads a year, right? That's DNA. So the short answer is people come to us, they're like, look, they innovate fast, it's secure, it operates at scale, and all those things. I think that that's, that's always been the DNA and that's what it'll remain, right? So yes, making— as long as you have APIs, which we're a platform, you're doing— anytime you migrate something in, you can kind of migrate it out. I think that that's been the case. I think that the charging, what ends up happening now is we're previously you were worried about, it's this huge people-based cost to do something. Now you're like, if I can move something over in a much shorter period of time, and it's in a modern way, and it's less risky for you, those conversations of, well, I have to run the same system for 3 years in parallel, like some of those just naturally go away, right, as well. So I think it's an exciting time as we're seeing the pace that people are modernizing. Yeah. And look, inside of Databricks, you mentioned at the beginning, you mentioned, you know, kind of Neon and Lake-based. The fact that it used to be also hard of migration is like, okay, when I migrate, I can use a part of it on here. I have to use these other things. But now if Databricks says, well, I have ingest, I have a data warehouse, I have governance, I have an OLTP database in the apps layer, I can kind of move everything there. It simplifies the migration. That's also allowed things to pick up speed.

You know, the irony we should break— it absolutely simplifies it. It is interesting when I think, I think because we're on Replit using Neon Lake Base, right? Yeah. I think I have lock-in now, ironically. And here's why. Obviously at some level leaving would be fairly easy, right? But now that, listen, I'm product savvy, but I'm not technical anymore. Yeah. I don't know that I have the patience. There is a soft mode now. Yeah. Right. And because as you go from now, you know, as you go to allowing more less technical users to use something. Yeah. Or in Databricks, letting non-data analysts use a product. It's a, it's a soft but real moat. The context switching, the energy. When it's just two, when it's just 10 data analysts leaving, it's one thing. When it's all of 7-Eleven, yeah, I'm not sure I want to leave. Yeah. Even if it's technically possible in 3 days now or a week, there's a different moat, right? Which is this organizational understanding of this structure that I think it's incumbent on folks to build. Make it, it's the next level of making the app so great no one wants I don't want to leave, right? Yeah. Which is adding so much organizational context. I don't want to leave.

Well, I think one, you're right on that one. And I think the second thing that I would add is we talk a lot about SEO and search engine optimization. One of the big things is you go, you use Replit, like to give you an example, use Replit and you use Neon Lake Base, not because you did a bunch of research on Neon Lake Base because you're like Replit was like, this is the default answer that's kind of easy to go do. So how many things as people are kind of increasing increasingly vibe coding and doing it is just like, hey, the thing that works well underneath from a PLG, it's like whatever the AI tool that I was using recommended doing, that's what I'm going to go use. So I think that there's also a lot of locking up. Many people just like will abstract away some of those things, and as long as it's good and works well and is easy to use, they're comfortable using it. And to your point, why change? I have 50 new projects I want to tackle. Why take the time to do that unless I have a really, really compelling reason? So where you land first actually matters a bunch.

I think so. Then maybe we'll wrap. I even think that what matters even more, we land first, is what my agent wants. And my agent wants Lake Basin Neon. And what my agent wants, all things being equal, the agent gets.

There you go. All right, Arsalan, thanks.

Thanks for going over. Thanks everybody, this is great. Thanks so much. Thanks, man. Thank you. Thank you very much. Fabula.

There's something more magical happening here. It is not a pattern that I've ever seen before in the last 15 to 20 years of being in growth. This is one of the once-in-a-lifetime type of companies. We had $100 million and just 4 months later we were at $200 million. The company is moving. We're just making sure the wheels are greased. You can feel it in this office. How can we make this better? How can we deliver more to our customers? I vibe code every single week, so I'm my own customer. You have to act so much on intuition. You have to have that right vision. Vision, just knowing that the moment that you reach destination, the goalpost has changed immediately. Joining Amelia on stage, please welcome the head of growth from Lovable, Elena Verna.

All right, thank you everybody. Welcome back for from lunch, hopefully. Thanks, Alena, for being here. I'm stoked to talk to you guys today. I thought we would also do a little bit of maybe some live vibe coding like we were yesterday, just to continue the theme of the day. I've got a few points I want to hit, but the first thing I wanted to ask you is— well, actually say is thanks for being here on a big release day.

Yes, thank you for having me. I appreciate it.

For those who don't know, can you walk folks through in the room a little about what you guys released today on the AOE thing? Yes.

So, uh, I work for Lovable. Lovable is a native speech way to build any software application that you can possibly imagine. And today is our big launch day for allowing every single app to be native for SEO and AI discoverability. So for those of you who are more technical, from server-side rendering, SSR, to pre-rendering, uh, you can be discoverable. And we also integrated with SEMrush. Yeah. So you can chat and lovable about which keywords you're ranking on, where do you need to build out more content, and with a click of a button do it all. So you're not just building an app, you're actually building an app that will be discovered by customers if you so wish.

And it's a big deal. So for those who are maybe newer to Viya coding, usually, you know, you deploy an app and it's like everything's indexed all the same way because it's all Viya coded. So this is a big game changer because now you can actually get more specific with it.

Yes, you can build your sitemaps. You can fix titles on your pages. You can build more landing pages to rank on keywords, and sometimes we live in the tech bubble and we say, oh, all of the discoverability is now happening through AI. Well, yes, but still there is a world out there where people are still using search and Google, and you can't just forget that we are here a little bit more in the bubble compared to the rest of the world. So if you're building for that rest of the world, search is still very important to get right. Yeah, that's true too.

And then since you guys are on, you know, obviously you use your own tools, you Viya code a lot of stuff, like how do you feel like this release kind of impacts folks? Like, you know, maybe like we're on some of the things we still use, like we use all the Viya coding tools, but then we also still use like WordPress and Squarespace, and now I'm like, I don't know if we need these anymore.

Yeah, I mean, it depends on what is it that you're trying to accomplish. At Lovable, we're specifically focused on non-technical builder, meaning that if you want to go start your business or build internal tools for your company, you build a website, do prototypes, build a B2B app, build an e-commerce app. And if you don't have engineering background, we are the place where we make it all seamless and make it very easy. So yes, there's a lot of tools out there. This just really proves that it's a booming category and there's space for quite a lot of competition. Honestly, it benefits all of you because more options that they're out there, the harder it makes us work to be on the bleeding edge of all of the innovation and capability. So I think it's a good thing. It pushes the category to go faster.

Yeah, for sure. Yeah, I'm sure, I'm sure some of the other vibrators are going to release similar features in the next couple of days, if not weeks. And, and then something else I wanted to touch on before we get into to it too, is we were just chatting backstage, like we kind of banned this year any talks that were just gonna be like, oh, how we scaled to XYZ, and kind of background stories. But I did wanna touch on yours for a little bit because you have such a rich and interesting background at being at these big SaaS giants, right? Like Miro, you were saying, Dropbox, Amplitude, SurveyMonkey, like these are big names. And we were just talking about how you used to, managed folks there, but now you're back to being not only at Lovable, which is a live coding company, but back to being an IC. I am.

So I've been at Lovable for a year. Today is actually my 1-year anniversary. I'm very excited. I've made it. I've made it. That was my goal. Can I make it a year at a company like this? But I've been in the industry for 15, almost 20 years. So my background is in traditional tech. I'm trying to catch up and learn as we go. So actually my previous role at Dropbox, my team was a couple hundred people because growth was a really big driver for Dropbox. It was a very heavy product-led company and our team was bigger than marketing team as a whole. And now at Lovable, I started as running more of a marketing and growth. I actually fired myself out of marketing job because I knew that they needed somebody else. Else. Then I started a growth team, and then I went back into being an IC. And I'm falling back in love with work because I've always loved what I do. It's always been my hobby. But management of really large teams is more so much about coordination and cross-functional dependencies and really empowering people that you sometimes lose touch of craft. And I think in this AI era, it important to maintain and evolve and amplify your craft that I love just build— being back in the builder seat and just discovering of how many more things I can accomplish by myself as opposed to having to need a large team to make my ideas come to life. I love it and I do think that is going to be the next flex of our careers, not to go up the leadership ladders, going to those fancy VP titles and actually becoming this high-power IC that is capable to accomplish of what it used to take dozens of people to do.

Right. Now with agents, right? So for one question we've gotten a lot already in like the last day and a half is how do folks find an Elena, right? Like you're back to being an IC. I don't actually know the story of how Anton found you, but like how do folks, you know, find their own version like an Amelia or an Elena to do what you're doing, right? Become an IC, get their hands dirty, and start building with agents?

I think there's a lot of leaders out there in the companies that are deeply unhappy with the type of work that they're doing. So because how do most people get in management? You become a really good IC and your benefit of like of your appreciation for being good IC is to go into management and do something completely different versus what you're actually great at. So there's a lot of A, very terrible managers out there as a result. Honestly, I wouldn't wish management on my worst enemy unless you truly enjoyed it, because it's a hard job. It's hard. And you either have a skill set to do it or you don't. Yeah. I would say I'm like mediocre manager. I'm much better at a C craft person. Can I do it? Sure. But that's not where my superpower is. So reaching out to those people and saying, hey, would you like to would you like to exercise your craft again? You'd be surprised how many people are saying yes right now because they see it as an opportunity to fall back in love with work again. Yeah.

Yeah, I want to— we're going to switch to my laptop too a bit if we can. I wanted to show folks a few things that they could do with Lovable just to kind of ground it in reality because something you and I talked about on the prep call was how there's no feature differentiation. Differentiation anymore, right? We were, we were, we were— I like to do these calls now because it's handy, but we were talking about how, you know, any, any, almost anybody now that there is vibe coding can just catch up to everybody else in a day. So I want to show folks some of the things you could do on Lovable.

So, so yeah, first of all, it's a really funny tagline. Oh, there's no more feature differentiation, which is True in some capacity, but it's a little bit maybe overexaggerated as a statement without context. So for sure, what do I mean by that is that we in the SaaS era, because how expensive it's been to develop anything technically, have relied on feature differentiation as our moats. This is the reason that we won because we had different product offering versus the competition, because we had better technical technical teams building it. We had better product visionaries that were developing it. But now with the cost of development collapsing because anybody can go to products like Lovable and just say, hey, I wanna build this feature and AI builds it for you. You don't need years of engineering experience. You don't need, for some of this obviously more simple functionality. I'm not saying that Lovable can build everything. It will be in the future obviously. But for now it's still the commodity quality of what actually can be created by me, you, you. Yeah. Is increasing substantially. So feature differentiation is really hard to maintain as a moat. You might be ahead for a month, 2, 6 months, but then everybody will catch up because AI is writing so much of our code. In AI-native organizations, 80% plus of the code is written by AI nowadays. So imagine Imagine how many more engineers that you would need in a traditional organization to do that. So yes, you can still have feature moats. It's just gonna be very short-lived and you cannot rely on it as an ongoing predictable way to grow. The moats that are still working are still the same ones, which is hardware because it's so hard to actually create hardware. Network effects. Network effects has always been really hard to create, but once you have it, oh boy, boy, this is a gift that is gonna keep on giving. Data and security and compliance, those are still very well worth investing into because those will be more long-lasting.

100%. Yeah, you mentioned data. So one of the things I did here, just to walk you guys through what I did, if you wanna replicate it too. So yesterday we did the session where I live built an AI VP of marketing. So if you go to saasreannual.com/resources, you can grab the two things I already pre-vibed for you guys so that you have it handy. It's, it's a, it's a, it's a testament to the data. So I, I packaged up some sample data so that people could just duplicate it, but I gave it a spec. I gave it some sample historical data and I popped it into Lovable and I said, hey, here's my spec for an AI VP of marketing. You know, I— this is what I want it to achieve. And it's this whole long spec. Build your own, you know, AI VP of marketing. Start as a dashboard. Here are all the goals that it's supposed to do. Here's the the data, right? Obviously this is fake data, but when you use your own data, you'll get better results. And I think that's such a true— I'm glad you called that out because so many folks have been asking us, right? You know, we're, we're all vibing so many things so fast now. Like, what is, what is the moat? I'm like, well, we have, you know, like 12 years of SaaS data, and I don't think a lot of people have that.

Yeah, absolutely. Data is big. Brand is another one that I didn't mention. Brand is back, baby. I mean, so much of of the growth efforts now is how can you make your brand differentiate and connect with your customers? At Lovable, we're so big on building in public because of that. We want you to see who is behind the scenes. We want you to see what we're thinking. We want you to contribute to our development in real time as opposed to it being behind closed doors. And by the way, this prompt is like, it's amazing how well structured it is. So like, that's great. The way that I vibe code is a lot more chaotic. I'm like, do this, do this. No, this looks wrong. And then I just go to plan mode until I don't get what I want. So there's like very different styles to do it. Just don't think that you have to have this well-articulated prompt in order to get started. Yeah, correct.

Yeah, I, I like to, you know, I'm like, I'm, I'm definitely like you where like once I have it live, then I like tweak it too much. I'm like, do this, do that. And then I'm like, okay, but usually when I put it together, yeah, pull it, rein it in. So I like to start it with a good spec so that I'm like, okay, that way. That's the best. Which knows what its goal is and like it's not chaotic because I was chaotic. Exactly. But this is pretty fun. So if you guys haven't tried it yet, you should try to do this in your own Lovable and just prompt it, right? So it made me this. I do like that. I explained yesterday, but Tenkei, who is our AI VP of Marketing, started as a dashboard. You know, it's going to generate 4 moves. Let's see what it does here. It's thinking. But I think it's pretty cool that you can put all this in and then Okay, let's see what it said. I'm gonna see how these compare to yesterday's too. Okay, so actually very interesting, um, how different these are and actually not in the same category of the ones that I got generated yesterday from a different vibe coding platform. So this one, as our AI VP of marketing said, um, it's 4 top ideas for me today, um, to grow the go-to-market was to launch onsite registration. Okay, fine. That kind of only applies to Saastr. Initiate LinkedIn campaigns. I'm like, that's, you know, that's always a good reminder. I'm always behind on my ads. I don't know about you, but I'm constantly struggling to keep up with my ads. Sponsor revenue. Again, this is fake data, but it says send a personalized outreach to top sponsors. It can help you do that. So again, this is pretty good of like, okay, it can take a spec, it can take, you know, it was my fake data, but you guys would use some of your real data and then spit out already like 4 really good ideas that it can help, the agent can help you then deploy instantly.

And the power here is when you can actually connect to your data source. So it's live updating. So we use the word connectors in our product and you can connect it to BigQuery or Notion or Slack or wherever your data source is actually coming from, which is what makes it so much more powerful because it's live and it's living, breathing thing that can guide you through it.

Yeah. And for the folks that are like maybe more so like the founders in the room, I don't Do you know like your top use cases of what people are vibing right now on Lovable?

Yes. So we have spent quite a bit of time figuring out what is our ICP, ideal customer persona, because it's so easy to just be so horizontal and be anything to everyone, which really means nothing to everyone because you're too drawn out. So we started very explicitly with a non-technical builder. That is very focus of our persona Because we want to enable the 99% of the population that does not have engineering capabilities and knowledge to be able to build the ideas because they are experts in something. Let them materialize that into whatever is it that they need to build from software perspective. So that is like at the very tippy top of the pyramid. Then it goes into, do we have a founder persona that is non-technical founder or do we have a non- technical builder inside a company. So it's an employee persona. And the founder persona, then we're empowered for B2B companies, B2C companies, or e-commerce companies. We have a partnership with Shopify to help you create a really amazing e-commerce experiences if that's what you wish. For employee, non-technical builder, then it's more into internal tools. Hey, I have this workflow, it's so annoying, let me just go build a little software for it to solve for me, or prototyping or marketing sites or websites or satellite apps. I call it satellite apps. It's not your core product, but it's the apps that you create for campaigns or for sales outreaches. And it works quite wonderfully. So we have like this tree that we go down and then we build full-blown functionality for each one. Yeah, makes total sense.

And then for folks that are maybe starting, like, do you see them build a dashboard first? Like, What are the founders building first on Lovable?

Founders, they have an idea. They think this is the problem that I see in the market. That problem can be very local. This is happening in my daughter's soccer camp and I see that there is a problem. Can I go build a solution? For example, one of our successful founders is a full-time student and she builds an AI study buddy and she got $130,000 in AR in the first 30 days of launching it because she distributed across her school and it picked up like a wildfire. So it depends on whatever is your local knowledge is, whatever is the idea that you're frustrated with. And before you had to go and hire engineering team or find a co-founder to build it. And now you can just go and validate the idea on your own. Try to get first couple of customers of your own. I really truly believe that we're gonna be entering the era of solar entrepreneurs and running companies that are almost mid-market, potentially even enterprise-size-like, as well as the really long tail of really small, what I would almost call mom-and-pop SaaS, of something that like makes a good living for you or makes a good side gig for you. It doesn't have to be this huge business that needs to raise money, but it can be a nice source of income for you. And especially as all of our roles are are changing and organizational structures are changing, I really urge every single one of you to go and explore a side business because every single one of you knows the idea that has not been solved with software because often it was too costly to solve it with software, but now it's completely ungated for you to go and build something out on your own.

Well, not only costly, but I think so too, like when we've tried to add like feature features and functionalities or even getting this event done in the past. We sometimes just can't find the resources to do it. It's hard. Yeah.

And it's not even— it's that, absolutely. But then it's also like, how do you launch it? How do you find product market fit? Where do you get first couple of customers? Which is also something that we are heavily focused on because we really believe that building something is already becoming a commodity. Everybody and their mother are going into competition of creating and app builders, how do you create value out of what you've built, whether it's your personal value, value for your company, or value for your customers? That's where the next level of the oomph is gonna be and something that we're really heavily focused on.

Yeah, it's crazy how much, um, folks can do just by vibing, right? To your point, like non-technical folks, like I don't consider myself a technical, like I don't know how to program, I don't know how to do like traditional coding, but I, I know how to vibe code now. And I think it's one of those things too where like, I feel like sometimes folks are nervous to start it because they're like, well, I don't want the agent to like do too much. Or what if the outputs aren't very good? I'm like, it's not like my vibe coding was good 5 months ago. Like, it's very different from how I vibe code now. Like, I think the goal is just to start doing it. Exactly. It's all about just starting it.

Really, the output at the beginning doesn't matter because the one thing that everybody— there's almost like an AI confidence theater. That is happening out there because everybody seems to be having workflows and tools and like these systems and you're like, okay, I feel behind because I don't have all of these things. In reality, all of us are still learning. All of us are just starting somewhere. You might be further in one exploration versus the other, but the only thing that you really need to do is just go and learn what can they do for you.

And how does all of this impact hiring at Lovable Rights? So you guys are an AI-native company. You're a vibe coding company where you can use your own product. Like, how do you find folks that have— or that are AI affluent when it's still fairly new in the grand scheme of things?

Yeah, so there are really two types of people. There are truly AI-native people out there. Those are usually much younger. They're right out of school. They grew up with learning with AI. And then there are people like me that have been in the industry for over a decade. I'm like learning how to be AI native. We're at Lovable, really believe combining those two personas. So we push on each other because I can bring more of the patterns and industry expertise to the table, but I'm learning how to be AI native. And then there's truly AI native that people that we're hiring. Yeah. Right outta school to really go push the boundaries because they don't have the box of what's possible. To them, world is their oyster, and off they go creating and doing everything. So we try to combine different expertise. And we're really just trying to find people where this job is going to be a global maxima for their career, an opportunity to learn the most, an opportunity to give back to craft. That's why we're completely flat. We don't have titles internally for a reason. Everybody is expected to do some sort of IC work. We don't have just leaders, uh, that are just sitting on top navigating things. Everybody's just expected to build and we gonna push the boundary of what it means to be an AI native organization as we continue scaling. Because even right now we shared our figures, um, numbers in February that we've crossed over $400 million in ARR and we're just shy of 200 people still. So we are operating as a very tiny team for a reason.

And when you think about bringing people into the team, like, do you guys already know like where you do want to put those like dollars and headcounts into for folks? Or are you like, no, we're just going to kind of stay where we are for a little bit? What do you mean? Like, are you like now because you are so tight-knit, right? But you have kind of the revenue to do more. Like, are you guys putting more into marketing or are Are you putting more into product?

Engineering is our biggest priority. This is like when people say, oh, engineering is dead or design is dead. All of that I think is a lie because we had never had more need for experienced, incredible engineers that are gonna push the boundary of what's possible to create with software, to enable AI for all of you to push the boundary of what's possible at the same time. So engineering has always been our number one priority and we actually call it product engineering. Because we don't follow the ratio of 1 product manager to 7 engineers to designer to marketer. Everybody is kind of doing everything. So autonomy is very important for our team and we love to hire previous ex co-founders or founders because they're incredible at being able to do marketing and engineering and product manager and analytics all at the same time. So those blurred roles and responsibilities are very important for us. Uh, and then we do have GTM, but a lot of our go-to-market is product-led. So just building amazing product that creates stories worth telling for our customers so our customers can go tell that story to their people that they're connected with. That is our growth strategy. Yeah. But we do have incredible marketing team. We're ramping up our sales team because we see incredible demand from organizations, uh, wanting to get into VibeCon. Coding as well. But engineering is still number one priority because I say playfully that we even approaching half a billion dollars in revenue, we're still on the product market fit treadmill because category is evolving so fast, both from technological perspective, from customer perception, that we still need to keep pushing the product at the maximum velocity forward. And it's a living, breathing thing that we need to almost like recapture every single month.

Yeah, I was gonna say, can we, can you, are you able to walk folks what that looks like from like your perspective of you guys are shipping features so fast, like not every day. I was gonna say almost every day. Yeah. Multiple times a day. Every day. Okay. Multiple times a day. Like, and it's, it's, it sounds like it's due in part to how you're hiring, how you're structuring the team, keeping, you know, go-to-market maybe leaner, leaning more on engineer and more focused that are technical, technical backgrounds, co-founders, folks that maybe their roles blend a little bit, but they can help do a little bit of everything to keep the products and features deploying every day. Yeah.

So we have a very clear goal across company of where we want to end up. We want to be a lost piece of software. What that means actually changes even for us as technology continues to evolve and as we understand what AI AI is gonna be capable of in the future. But we keep a really tight 3-month deliverables for ourselves of the things, the big items that we want to accomplish. And then engineering teams and product teams, they, and even me, I ship to production myself too. I've never been able to ship to production before. It's like such, if you haven't done it, I really hope that you do. It's like the feeling of like a superpower that you all of a sudden unlocked. But everybody ships. So we have a Slack channel that is #shipped, and I love going there at the end of each day and just like going through the wall of things that have been released to production and just like, oh my gosh, it's a, it's, I've never seen this much velocity in the product. And that's also because we leave a lot of room outside of that roadmap of people just going and fixing and developing things that they're passionate about. So it's not heavily prescribed top down. It's very much much left in flex to what your idea, what idea you have of what you're capable of doing.

And do you guys have a hierarchy in like who's shipping? Because if you like, you're, you, to your point, like you're shipping features in the product, like you've got engineering shipping features in the product. Like is it just if somebody has a good idea in that Slack channel, you guys just put into production?

Yeah, we have a channel that's called #feedback, uh, that everybody is like, hey, I think this needs to be done, and everybody's is like, that sounds great. Let's go. And 24 hours later, it's out in production. So it's like it's very agile. It's like it's very quick on its feet. And we bring customer ideas. So like we constantly screenshot things that we see on socials or that come through our support queue or anybody writing to us. And if we agree, so we have like almost a principle, if you can convince one other person that it's a good idea, let's go do it. But that's because everybody holds such a high amount of agency and autonomy that even getting agreements sometimes from another person is not as easy. I have a funny example. Yeah. I was doing pricing page changes to make our enterprise plan a little bit more transparent of how we show it on the pricing page. And I vibe coded the full design and I'm like, okay, I created a PR, let's ship it. And then engineer who's a 20-year-old engineer is like, um, I would like for you to get a design sign-off before I commit it to production. And I'm like, 'Okay, let me go talk to designer.' But like everybody pushes against everybody. There is no hierarchy. There is no like, 'Oh, if it comes from above or if it comes from this person, it has to go through.' Everybody challenges each other and it's a really wonderful atmosphere to be in. Like, I love that type of mentality.

No, I'm just saying, it's just, it's wild to think about how that probably compares to like how that worked for you at previous companies, right? We're at a Dropbox, like you probably can't even get a feature if you—

Well, it would take months. And then if I had an idea, I would get very little pushback on my idea, not by design, but just because of all the hierarchies that we create in organizations and the lack of the transparency and information flow for even for anybody at the bottom of the organization to be able to argue with the leader because they are not equipped with as much information. We at Lovable push that information all the way down to everybody so everybody can make the decision.

Yeah, because typically, you know, for the founders and folks in the room who are in like leadership roles, a lot of our audience, like, it's just so crazy to think about how to like now as I see is like I've gotten a lot different in how I speak to our vendors, right? Like I ask for like API features now and I'm like, I didn't— I would have never thought about asking for like all these crazy like custom AI features and functionality. And I did it almost every week leading up to SaaS because I was like, hey, we're using it in our vibe coding or we're using it with our agent and like we're just hooked up into the API. Like, can you add this to the API? Can you add that to the API? And they're like, what are you talking about? They're like, have you read the documentation? I was like, no, I haven't read the documentation. Maybe my agent did. But like, we're literally just shipping and deploying, right? It's, it's just a dramatic shift in how like a year, maybe even two, like maybe a year ago of, you know, going, having to go to product and ask for features and having that lag versus now, you know, you guys are shipping daily and it's coming from within different people in the org and the company.

And also the things that you even like, what are you shipping? So often I could go and try to like do RFP and like buying software and I was like, oh, let me just go build it in an hour because like, let me see how far I can get there. So we constantly pressure testing our own boundaries of how how much of our Lovable we can build on Lovable ourselves, which gives us a really nice roadmap horizon of the things that we need to improve as well.

Yeah, what's the— it's a little bit on the spot, but what's the best thing you could think of that somebody internally has vibed at Lovable?

The first one that actually existed before I even joined is our internal admin tool, which I can log into any given workspace, I can grant credits alerts to people. I can create coupon codes or discounts, and we call it WOOF, W-O-O-F, but it's built on Lovable and it's just like, it's mind-boggling. So sometimes I'm like, oh, I wish that there was like a bulk export of this. And like our person from support's like, oh yeah, let me go at it. Done. 5 minutes later. And I'm like, mind blown. Like of the capabilities that you can just like add features on the spot. It's just, it's so great.

Yeah, it's crazy that it's Yeah, because you guys are building on your own builder and it's just all in real time, right? You're like, we don't— none of us need to wait on each other. We don't really have to, you know, somebody just has to plus one it and then we're off to the races.

But it's just a different way of operating inside of company. It's like it's something that you get used to because I sometimes go back to my old ways and then I'm like, no, no, no, no, no. I have to like, I have to think about this differently. So it's definitely, it's a transition even for us. For me, for sure.

Yeah, we've talked about it a lot at SaaSRev. Like, it's a— and I know a lot of folks have talked about it this week who've been on the stages of— it's a big mind shift change of— I think sometimes too, when I like observe people, when they start to vibe code or when they start to work with agents, like the way they talk to an agent is different from how I talk to an agent or how you talk to an agent. And I always like to treat ours and like how we use our like AIBP marketing or AIBP VP of customer success, is like we treated the agent itself truly as its own entity, like with its own personality. And it does. It's not something that happens overnight, right? Like when we started by putting these, some of these were dashboards and some of these were just to help us get more emails out and make sure we followed up with customers. Like it started from a place of, okay, we just got to communicate better with the folks that are in our product market and are fit and our customers and, and our, prospects, but it turned into like something much bigger now with spending a lot of time in it though too, right? It's like some of this does become like you get out what you put in, right? It's like, okay, we've, we've put a lot of time into some of our apps that we've custom build and asking for custom API calls and working with vendors that have helped us connect to our agents. And so, um, it's just a little crazy for folks to think about how much they can achieve though in real time. But sometimes Sometimes it does take a little bit more effort to get it to like where you want it or as, as more like thorough and custom. Like the more time you spend with it, the better the agents and the, and the sites can be.

Yeah, I mean, this like everybody says it, but it's all about context. Agent will just produce an average for you unless you feed it detailed enough context and instructions. So one of the biggest advices that I have for every single one of you is to start building your own context, your ideas, your call recordings, your just brainstorming, because all of us will have to build our AI doubles at some point. Yeah, just to continue scaling ourselves. But your AI double, if it's just trained on World Wide Web, it's just going to be an average intelligence of everybody. So try to start capturing that context. So when it's time for you to go into more of a gentler version of yourself, even to start performing some workflows and tasks for you. It has your thinking so it can truly replicate some of the work that you're doing as opposed to, again, just doing average. That is not going to be exciting or interesting.

Yeah, I think it's something that I've, I've definitely gotten more comfortable with too. And I'm sure you see it in your customer base of like, you know, people start with like fairly like either basic prompt or like, hey, just build me that or duplicate this. Task, or hey, take this task over for me that I don't want to do anymore. But then the more you get into vibing and like comfortable with the agent, you'll see that like the way you talk to it, you can just talk to it like it's a super— like it's the— I talk to it like it's the internet. I'm like, it's the internet plus any other context you put in there. Like if you're talking like to like an Iron Man Jarvis, if you will, like it knows a lot of stuff. Like the more context you give it and then the more you actually talk to the agent of like, hey, if these are, you know, your, your top 4 ideas for my marketing campaigns for the day, and you actually tell it and like train it like you would a human being and say, okay, maybe I really like number 1 or number 2, or maybe I'm not ready to spend so much on LinkedIn ads. Can you give me 2 more ideas? Like, this is what I think we should do. What do you think we should do? And literally treat it like a copilot or a person or like a brain. It starts to get better because then it starts to get to know you and And the magic of the agents like this and like in Lovable is it won't forget that context.

Yeah, I, my favorite prompt in Lovable is this is my idea, this is the problem, this is the solution. Now build what I say and then build two other options that are better than mine. And just then I can see like what it pulled out and then I can merge it into what I actually think is a great solution. But I often like, oh, my idea was horrible. Like, yours is so much better. But again, just, it's a thinking partner at this point. It's not something that I 100% trust because there still needs to be, what is your contribution to it? What is your context to it? But it's so great at pushing you forward and being like the biggest amplifier of what you're capable of doing.

I want to make sure we can try and hit maybe one or two audience questions before we wrap. So if you guys have a question for Elena, there's a mic. There's like 2 microphones out here in the audience. We could probably take like 1 or 2 questions before we roll right into the next session. But if you guys have a question you want to come up and ask, please do. Or if you guys are shy, then I'll keep asking her questions. Okay. I was like, you guys are all enthralled. They're like— I think they want us to invite them.

Everybody's in a post-lunch coma. I understand.

It's all good. All right.

So my last two questions. The microphones are over there if you want to ask a question. There's one there and one over there.

Oh, I see somebody coming.

Hey, am I audible? Okay, hey Elena, I just have a question because you've obviously done growth for a lot of big SaaS companies, and when you moved to Lovable, what was the biggest growth, all your learnings, shifts that you had to do? Because it's a very different sort of company, a very different avenue, right? And what is after changing that mindset, what has been maybe one most successful growth story that you've had, or growth feature that you've shipped at Lovable, which has worked really well for you?

Yeah, great question. So the question was, is what has really changed when it comes to growth at Lovable versus other companies that I've been at, let's say Dropbox, Miro, SurveyMonkey. I've advised 50+ startups as well on their growth efforts. I would say that it's twofold. Some of the things that that used to be the reasons that company won are now just table stakes for existence. For example, SEO used to be why you could win the market. SEO right now is just something everybody has to do. It is not gonna be the reason that you win. Same with SEM, so with paid marketing. It is something that you're probably gonna have to do, but it's not gonna be the reason of why you're gonna be successful out there unless you have those results that we've talked about, which is a more sustainable, predictable growth that you're gonna win. So to me, I actually joined Lovable to help codify a lot of those growth 101 things that everybody should be doing at this point because these things don't need to be reinvented, and I want every Lovable app to be equipped with those things. And then number 2, I think how important is it to be on social. I know that we've talked about it and like it's been around But having that building in social mentality and really leveraging your employees as marketing, almost powerhouse behind your brand and uniting customers behind it has been very important. And then just one other tactic that I will say, freemium has never been as important as it is right now in the age of AI. And I know it costs a lot of money and I know it's really scary to see that bill, especially of the the things that you gave away for free. However, it is the only way to get into customers' hands and to change their habits in order to do it. So we actually treat our freemium costs as marketing budget, and we're constantly looking for more ways to give away our product. We just announced a partnership with LinkedIn a couple of days ago where every LinkedIn premium member now has free access to Lovable, and those conversion rates to paid are double digit. So really get out of that mentality of I have to gate my high-cost features. You have to really ungate it in order to win in this market. Yeah.

Hi there. Absolutely love this. I think a lot of businesses right now are going through a dilemma of buy versus build, and clearly you're for the building of it. But internally at Lovable, Do you guys go through this dilemma yourself?

Are you all 100% build? We definitely go through very hard evaluation of anything that we're going to buy. But we're not fully built on Lovable internally too. So we love Linear. We use Linear for project management. It's just, it's so deep in features that we need to organize, especially across all of the individuals that are running so quickly. We love Slack. We use Slack. Level of granola for meeting recording. So we definitely pick and choose solutions too, but we've built so many of our own. So for example, we've built our own CRM internally for our sales team to use, and our sales team love it. So we're constantly pushing the boundary of it internally, but it's not a zero or one. The answer is somewhere always in the middle of the complexity of maintenance and the feature richness that needs to even function in that tool. But We try constantly to do it all ourselves. Thank you. Of course. Hi.

Hello. Hi, Elena. Loved your session. You talked about marketing structures and marketing teams in the AI era. I lead marketing for Sprinto. It's a compliance automation tool. We've been around for a long time, so we are embedded in a very typical traditional marketing structure. Your demand gen, in content, PMM. It's slightly harder, like, when you're not born in the AI era and you have that legacy. I was wondering if you had any tips in terms of thinking about team structures, marketing organization in the AI-native era. Like, it's a lot of baggage, like, because we've been around for a long time. So how do you think about that?

So I would think about it in two sides. On one side, you as a marketer are capable of doing so much more by yourself compared to before, and you should really take advantage of that. Whether it's improving your workflows and how the systems are connected so you're not working in Excel sheets and wasting your time there so you can actually do creative things, or whether it's creating launch assets yourself or apps on your own to really— you can remember lead magnets or like all of these like little tools that companies created. HubSpot has been amazing at that strategy. Now you can do it yourself. You don't need engineering talent to do it, and you should really lean into it because your marketing efforts can be 100x more important. However, marketing structures, the way that they're structured right now, are not created for the type of velocity that is introduced with the AI native development. So once your development team starts to go into AI native, then you really need to think about adjusting your marketing structure and organization because you're never gonna be able to keep up with the velocity that is gonna get introduced into the system. And then marketing goes into more of an enablement mode. How can I make every engineer into a marketer? And how do I create systems for everybody to say at least somewhat the right things while making maintaining their voice, have the tools at their disposal, and then I just pick and choose the things that I want to control and launch, and then the rest of it goes as like the ongoing noise in the system that is also its own marketing strategy on itself. But a development team really needs to lead the change in the actual marketing organization because otherwise you'll be the biggest blocker to velocity that they are capable of maintaining all of a sudden. Thank you. Yes, of course.

I think we have time for one last question.

Yeah, go ahead. Great. I'm Josh. Very nice to meet you. Big fan of The Gross Scoop. Been subscribed for a minute.

So that's a plug for everyone. If you're not subscribed, you should do that.

Curious, what's, what's one question you're never asked but you wish you were asked more frequently in your role?

What's the one question I'm never asked in my role from who? From anybody.

In general, what's one question you're just never asked but you wish you were asked more frequently?

I guess a lot of people think that especially after being in some of these logos and doing some of these things that I know what I'm doing. I'm sorry, I have no idea what I'm doing because nobody knows what they're doing, especially right now, of how much stuff we're actually figuring out as we go and we're learning. Learning, and how much of it is just wild ideas for me too. So first of all, just realize that we're all in this transformation era. I think when dot-com really— the web started coming into play, I feel like people probably felt the same way back then. But we're all figuring it out. We all don't know what's going on. And I wish more people realized that.

Well, thanks for being here, Elena. Everyone give a round of applause.

And I'll just be outside there if you have additional questions. Thank you, everybody.

Creating a culture where alignment and execution towards your mission matters more than anything else. It's how Snowflake went from zero to billions. Snowflake has always been, we are a customer-driven organization. The hard part is we need to remind ourselves to think outside in. A company like Snowflake still growing at an extreme rate. I just love the craft of marketing. I love what I do. I have to completely adapt every quarter, every year. If we're gonna take Snowflake now to $10 billion, I still need to change.

Please give a warm SaaStr welcome to the Chief Marketing Officer from Snowflake, Denise Persson. All right.

Hello, everyone. I'm back. Hello. I feel like Jimmy Kimmel here. But welcome back, Denise. I'm so excited you're back at ASTRA. I know last year we had Streethar, and that was really fun on our closing day. But I'm glad to have you back at ASTRA.

So welcome back. So excited to be here. Great to meet you again. And just love hanging out with the SaaStr community.

Yeah, I wanted to have sort of like a difference in marketing here, kind of with these sessions back to back, right? We were just talking about Viya Coding and marketing and how they deploy literally on, you know, anybody at the company can help deploy features at Lovable. And Snowflake is at such an incredible scale that I wanted to give people like a sense of what it's like to deploy agents at a company that has more like compliance and data and security, but you're still finding a way to push through and get it done. Yeah. Yeah. So one of the things, it's just funny, we were just saying backstage, like, you don't even know the number of agents that you have anymore. Can you give folks a little bit of context of what you do have going on in marketing that's agentic, just from a high level?

Yeah, I mean, again, it differs across all the different functions, obviously. And we have this vision. Of course, it's a very bold vision, and everyone might have the same vision, But it's that we want to become the most AI-assisted marketing team in B2B, essentially. So that's kind of what everyone is inspired, you know, by. And it's very much about also kind of getting everyone on board, you know, with you, right, as a team. I mean, way more people to get on board. I think a year ago there were more people maybe resisting, maybe even a bit scared, right? We saw some data around certain functions, right? A lot of people were were concerned about their jobs. I think this year is completely different. We're seeing everyone really leaning in. Everyone is kind of motivated about being part of the transformation, you know, a journey. And we just invest a lot in learning and development and a lot of peer learning, you know, across the team as well. So again, examples— I mean, I just got some recent, you know, data. Of course, one of our big spends Spent is around media, you know, buy and data display ads, you know, all of that, right? And in the past, it's been very quite fragmented. You have all these different channels with its own analytics. And what we've been able to do now, bringing all this together, it's really looking at, again, not just what happened, yeah, but really what, what, why is it happening, giving, you know, real-time recommendations accommodations, we can optimize in our spend. And that's where we've seen, yeah, now with 30%. Wow. So far, I mean, we hope to open— you know, you just— optimization, you know, happens on a daily basis. Yeah. But just looking back in the last, you know, 6 months, we've been able to lower our cost per opportunity with 30%. So that's where we're seeing real return of investments.

That is a real return. Yeah. The other thing I wanted to ask you, because it's such a contrast to from like Okay, I just showed folks like our AIVP marketing dashboard, but I know you've said like on our prep call and stuff, you don't really log into a dashboard when you wake up in the morning. No, I mean, I did. I'm like, that's so scary.

Yeah, I mean, of course I did that in the past. The thing is, when you look at the dashboard, it just creates more questions, right? You look at the dashboard, you want to understand, okay, why? Why did the pipeline go up in this territory? And most Most times, right, I would have to ping someone, have to Slack someone, hey, can you look into this? I'm just really curious to understand what happened here. Or I would call someone, or you would have meetings about the dashboards where you sit with the sales team having meetings, interpreting the dashboard. They're just talking about it. Exactly. Just talking about the time spent just discussing the data, really what's behind the data. So today, I mean, we of course use a lot of our own technology at Snowflake. But now, right, I can interrogate the data, right? I can understand again, you know, okay, why did the pipeline go up in US West, for instance? And what are the projections for the next quarter? Those things. So nobody gets any Slack messages from me anymore, right? It's quiet because I can get all the answers to my questions now that I couldn't get before. That's why I'm saying, yeah, the dashboard is dead. I don't think it's fully dead yet, but again, that ability to really understand the why and get, you know, recommendations in real time, I mean, it's—

it's because you can just talk to your data now in a native language. Yeah, yeah, talk to your data.

Yeah, I just— again, I mean, this, this time is just the best time in history to be in marketing. I mean, marketers always been about, okay, the one who's the most relevant wins, right, most relevant to the customer, and it's also about the timing, right, for that customer, and now you have all the tools to be 100% relevant at an individual level at the right time, so it's magic.

Yeah, it is pretty magical. What are some of the things on your non-dashboard that you see now that you that you maybe would have had to wait weeks for beforehand.

Yeah, I mean, even things beyond just— I mean, pipeline metrics, of course, is super important at Snowflake. Marketing owns the new business pipeline, right? So that's, that's what I would be— that's what I'm accountable for. But everything else around, you know, our business, everything going on— I mean, the market organization, we have about 700 people in marketing. We also our whole customer training, partner enablement, sales enablement team, and marketing as well. Just understanding for me in the morning, what's the health of our organization? Wow. Has someone— how many— you're asking it for internal? Yeah, exactly. Who joined Snowflake marketing in the past week? Or I mean, who has left? Why did they? Okay. Yeah. Do we have an attrition issue, you know, or not? Wow. Even things like, I mean, Traveling expenses. I can see if someone is— I hate looking at those things. That's a lot more for a CFO, but I can even see in the morning, right? I get my morning brief, my morning skill coming in, and I can see, well, these 5 people, they had booked too expensive flight tickets. It's actually quite expensive travel right now. But things like that, I get intelligence not just around kind of the business but the health of our organization. Optimization and, um, et cetera.

So I want to ask you two parts to that question because it's a little funny you mentioned expenses, but it's actually so important when you have a bigger org like Snowflake is, right? Of like, okay, being able to access, like, that's something that maybe only your finance team previously had. Yeah. But now, but now you can see almost probably almost anything across Snowflake and you can just ask your data the question of, hey, what's going on with XYZ person or new person? And it's kind of crazy to see that unlock. And then related to that, a bunch of folks have been asking our presenters this week is how good does your data have to be to get the agents deployed?

Yeah, I mean, absolutely essential, right? I mean, you know how it is. If you have bad data, you're going to make bad decisions. Yeah. If you have bad data and AI, you're going to make bad decisions faster and at scale, right? So the consequences of bad data when it comes to AI is even worse. Compounds. Exactly. I mean, again, I mean, AI runs, you know, on data. There's nothing more important to have your data estate in order. And also having that collaboration with your own data team, or your— maybe it's your, you know, IT organization, whoever runs data in the organization. But that partnership is so important because, again, in the past, right, we've been very sort of isolated, look at marketing data. The opportunity now is is to be able to combine that data with the rest of all data in your organization, all knowledge, all intelligence in your organization. That's really where the impact comes from.

And how much has now accessing this data impacted how you work with like marketing, or not marketing, with like sales and then product?

Yeah, I mean, in the past there was so much, I think, I don't know how many marketers and how many people in sales we have here, right? But think about the time you spend, right, agreeing to the data. Hey, this is what my data tells me, but mine says something else. Yes, the alignment, right? Sales would say, well, this is my dashboard and my data, this is how I interpreted this. Yeah, that importance of just one source, you know, of truth for your organization, that also saves you so much time.

100%. Yeah, I can't tell you the amount of times I'd be like, well, you know, I think this campaign, it says that is working, the sales team would be like, well, we didn't get enough, you know, close one revenue from it, or it doesn't count, or I brought in that deal. And like, the data will just tell you now, like, where the deal was sourced from, who touched it, what happened, what are the actions they took on the website. Like, that not only do the data— does the data know, but the agents know too. Like, exactly, everything ties all back together. And I mean, I haven't seen my agents duke it out yet amongst each other with the data because they, they're always falling back to what what they know to be true. Yeah, yeah.

And also, right, look at, of course, I mean, we all have some sort of funnel view, right? We have, of course, our top of the funnel engagement targets, we have conversions, we have, you know, meetings, we have opportunities. And today, also in real time, just seeing exactly, right, what, what, what is happening across the funnel and, uh, being able to address things in real time, right? Everyone in marketing have that last couple of weeks in the quarter, we're like, we don't have enough pipeline, right? It's too late. It is way too late for marketing to do anything that quarter. So that ability to really project, knowing, okay, you know, how is our Q3 pipeline gonna look like? How is Q4, right? Those are the things we want to be able to impact, not trying to fix something broken in the quarter.

Because now you can get a lot more ahead of it. Yes, you can make better decisions. And now that you have so many agents in marketing, like how, how you guys are a big org, like now how quickly do you guys move now when you see that something bad is on the horizon?

Yeah, in terms of, in terms of like if you see that, yeah, very much, very much real time. Yeah, really. I mean, everyone have their core metrics, especially when it comes to our pipeline. We're very clear targets for every single, you know, stage. Exactly. And yeah, we can see in real time, and it's also that ability to reallocate investments, right? Because in the past you only knew at the end of a campaign, right? Typically, yes. You want to be able to just adjust in real time. Wow, that's crazy. Yeah.

And then one thing I know you guys have been big on, like, like you said, like investing in brand and like data and then your team, like how much has that changed now with the amount of data that you have and with all the agents and AI that you guys are running?

Yeah, I mean, again, I mean, budgets are not growing exactly, right? I mean, it's all about— I think we are have, you know, 40, 50% growth we need to deliver on this year. But it's not like we're not getting more resources, right? You know, to do that.

You might be getting less. Yeah.

And I know exactly. Exactly. I think also, I mean, I think for next year, I don't think anyone is going to come back and say, I'm going to need more headcount for this. They're just going to look at you and like, you don't understand where we are today. Right. So I think everyone realized what's coming and that, yeah, that's what drives people to, yeah, to optimize more. Yeah. Yeah. And I know you have—

so you've got a lot of agents deployed across the marketing org. Is there one that you find is like the most valuable maybe for you? Or then maybe if you have a different answer for your team that you feel like has really identified your process at such a large scale?

Yeah, I mean for me it's more like, and I can get it today, tonight, right? Today I get, you know, the brief, you know, for, for tomorrow. I can see everything going on. I can get recommendations for how to prep for things. Just things like again, going into a customer meeting that used to take a lot of time. Time for sales organizations out there to prepare for that, right? Okay, Denise is going to meet the customer, right? It's going to take— they could spend, you know, hours on all that.

You had to talk to all the different parties, you got to, you know, listen to all the call recordings, and now you're like, well, the AI just summarized everything that happened and everything that we should be doing.

Yeah, just getting that full, full brief before those, you know, customer, you know, conversations. You're clearly more intelligent when you go into those conversations. Things like preparing all the presentations, right, for customer meeting, the time that used to take to custom that. Now you can create complete, you know, custom presentation depending exactly on that customer's, what they care about. That's, that's pretty, pretty huge time savings. It is.

Yeah. Yeah. One thing I want to get your like kind of unique perspective on is you guys have a, obviously you have a big go-to-market team at Scale. They're creating a lot of data because you are Snowflake, like how do you manage that now with all the humans entering data, but then all their also agents and their workflows, like compounding amounts of data that's just going back all into Snowflake.

No, exactly, exactly. I mean, it's a loop, right? It's a loop. Yeah, exactly. But also for us, I mean, we also have a centralized sort of AI engineering team, which is also very important to understand, okay, what, again, how, what are these agents? Doing that we are, we have developed, right? I mean, are we compliant? Is our data safe? Are these agents having access, you know, to the right, you know, right information? So there's a lot of, for us, right, there's a lot at stake if something goes wrong, if we send out wrong communication to customers or something like that. So we of course have to invest more in what we call like the control plane, right? In terms of what these agents are doing. And I think if you're a bigger company now, well, maybe you can make, of course, bigger investments. If you're a smaller company, you can clearly move faster. You can take a little more risk. Exactly, you can take more risks. And also, in a larger B2B organization, there's so many channels that you have, right? I mean, just the number of channels you have from a digital media buy standpoint to events. And in terms of events, we're just seeing the demand for events, all those in-person experiences, yeah, it's going off the roof right now. So I think 10 years ago, right, we all said, well, events are dying, and think about this marketing team spending any time on events, that, that's bad, it's all digital, right?

It's a waste of money.

Exactly. We're seeing now again that people really craving those in-person experiences, really coming out, really learning from others at the moment.

Yeah, you brought up something interesting, which is that, you know, because your data compounds so much, right? You have humans building with agents and all the data flowing back together. Like, because you guys need to be so compliant, like, is there— how do you, how do you manage that now? It's like, do you have— you still have like a human that presses go at the end of the day, or how are your agents able to go Yeah, and exactly right.

I mean, we are first— I mean, look at last year, you know, everyone did experiments. We, you know, we got some efficiencies. Today, again, most of us have been able to automate quite complex, you know, workflows. Another example for us is that most of our revenue comes from expansion within our existing accounts, and we're really then increasing consumption by activating new users. Use case. So just looking at an account and like, okay, where do we start? What use case should we promote or market or sell to which account? That workflow, which it was very time-consuming, that's a workflow we've been able to fully automate. But what we haven't done yet is of course the full activation part, right? The kind of, can again the campaign activate by itself? Certain parts within our digital media investments. Yeah, not there. So I think that's why we're seeing, we're seeing more that, you know, GTM engineer function, you know, coming in. And that's, that's kind of a function that we are expanding to really help with that component. Yeah.

Yeah. And, and you mentioned GTM engineer is something you're hiring for now. Yeah. How much has your hiring changed from like the last time Snowflake was here at Saastr?

Yeah, I know that is again, like I think the GTM engineer function has clearly is that's the one we're hiring for. Business analysts, not so much anymore. So there are certain kind of functions like, let's say, our content marketing team, they're able to do, of course, so significantly more than they did in the past. Exactly. So it won't be a function that we will sort of invest more in in terms of, you know, headcounts. Yeah. We're seeing functions like product marketing, right? We have always wanted product marketers to spend more time with customers. They haven't had time, right? The product marketers have a wealth of information. They're amazing in front of customers. They just haven't had time for that. And now they're getting that time, you know, freed up to really spend time with more real high-value customer interactions that they couldn't do before.

Which leads me to a related question that I think a lot of folks are struggling with now, which is is you've mentioned, so now product marketing has spent a lot more time with customers. How do you think about this whole debate of like FDE versus SE versus product marketer? Like how are you guys tackling that when folks are like, hey, I need a lot of help to get myself set up on Snowflake to get Agentic and be able to talk to my data?

Yeah, well for us again, like we have FDE capabilities for our customers as well. So some customers really, really want that. And of course, I think we're seeing now the world moving very much towards we're looking for specific business outcomes. And that's of course where this FD function comes in, right? They go in and like, hey, we're going to help you accomplish this specific outcome versus like, here's the technology, go and figure out yourself. We of course have a very large partner ecosystem with the GSIs and the SIs that do a lot of that for Snowflake, you know, as well. Yeah. Yeah.

And have you seen more demand from customers? Because I know you guys are obviously responsible for new business. Yes. Do you see the new business customers asking more and more for it, for something like an FTE? Definitely.

Because also when you talk to customers, right, they all— many of them now have like, they have a vision, right? They know where they need to go. And they also understand that the technology is here for it, but they don't really have the confidence yet in their teams to make it happen and at the timeline that they want to. So customers are clearly looking for more of that help to get to that gap in between their vision and execution, essentially. So customers are asking for a lot more now than they ever used to ask for. Yeah, definitely.

And how does that impact everything else in marketing for you?

Yeah, again, I think we talked about— business outcomes. I mean, we have made such a big shift this year getting the entire sales organization to really just talk about business outcomes. Yeah, and of course we're fully aligned with that. And last week we have something we call the AI Challenge on the marketing team. So every week someone creates a video about an agent or a skill that they developed. And one of our product market leaders, he's developed develop this business outcome skill. So essentially then you can put your content, let's say you've created content, right, for, it could be for a customer, it could be for an industry or a persona, you put it into the agent, the agents will basically look at, okay, how business outcomes focused is this really? What are you missing here, right? How do you need to adjust this messaging or this content? Content. So, um, just an incredible example again, right, of someone building something like that.

And since you have a bigger team, for folks that are, you know, scaling in the room or maybe thinking about, you know, future-facing of when they get to this scale, how are they going to roll out agents to like the entire marketing team? Can you share a little bit about how you've built like AI fluency without— with the entire marketing team?

Yeah, and I think that's probably been the biggest investment we made over the in a year to also really get everyone on board. There's a lot of like, you look at people who are self, wanna self-learn, right? Who wanna look for education and have that curiosity by themselves. This is not a time, right, where that traditional kind of weekly training sessions, right, is working anymore. And also when you deliver training today, right, you might have 80% sitting there saying, I already know all this, and 20%, wow, this is way too advanced for me. So that's also the power of AI now and learning is that you can learn at the, at the same pace, at the pace or the phase you're at. But, but again, this, again, overall investments in that. Again, we do have, of course, our weekly, you know, AI skills training sessions for the team. As I mentioned, we have the kind of the weekly AI challenge where someone is sharing what they built, challenging someone else to show what they built. We do AI hackathons at the function level. Let's say what the communications team needs to do versus our digital marketing team needs to do are quite different. So we have done a number of folks are really experts in helping the team leaders setting up these hackathons et cetera. We have our AI council, you know, on the team. So every, every quarter we have an AI day across the entire team that everyone around the world, you know, participates in. So, and I think that it's so important to not have a kind of like— I don't believe kind of in having the stick, right? You know, forcing people like, you have to do this. It's about really inspiring people and fuel that curiosity from within. And I think looking across the marketing team at Snowflake, right, some of the greatest innovation from an AI perspective, it happened among people that I could never have expected. People that they were the last people that I would think. If we look, let's say our leaderboard, we actually do have a leaderboard as well in terms of it is actually a sort of a usage community-based leaderboard. And, and also with that, I also tell the team, right, hey, this leaderboard's good, but it's not, it's not about how many tokens you're using, right, that matters either. But looking at our leaderboards, yeah, exactly, how are you using it? And also realizing that everyone is at a different phase of the journey, right? Yeah. And, but again, the leaders on the board are people from our brand team. I didn't expect them to be there, but the top 3, they're actually on that team, and they're just innovating a lot of things in terms of what we're doing with our website and their brand and those things.

And do you incentivize them to get on the top of the leaderboard?

I mean, I think they're excited. Yeah, we have some prizes and things like that. They just, I mean, I don't think they're like, okay, I'm gonna do this to get on the leaderboard, and I think it's I think it's super important, and again, not to have a behavior like you're just having people do busy work, right? Just use it just to show that you're using it. That's not the important part. Also, we have what we call our quarterly what matters. Essentially, it's like our OKRs on the team, and everyone across the team, they need to have a what matters for AI. And it could be that I'm gonna learn something. It doesn't matter if something basic, right? But you need to have something that brings you further, right, in terms of AI fluency. So some are way out here, some are a little bit behind, and I'm also saying it's okay, right? We just want it to kind of move. Yes. So we all have a sort of an AI skill we need to learn on our quarterly What Matters, and also something that we're going to build. And it doesn't matter if it's something small. It's more about really getting in, getting it, starting using it, and then yeah.

Yeah. How has that impacted hiring for you specifically in marketing? Because I feel like right now everyone's just kind of in like a, almost a holding pattern when it comes to marketing of like, you know, maybe we can't find— I, everyone's been saying like we just can't find enough go-to-market engineers because you post the job and people apply, but no one's been in a go-to-market engineer yet.

No, exactly. But I think some people converting into that. It could be someone former marketing operations. Yeah, we're seeing a lot of marketing ops. Yeah, I mean, think about I was in marketing when marketing ops came 20 years ago, right? Where did we find those people? Now we couldn't find— it was the same thing. These were people that saw, wow, there's a new career opportunity. I really want to lean in here. We're seeing the same thing that a lot of marketing operations, also those We said that we don't need the business analysts so much more, but the business analysts are so used to also working with data. So many of them are leaning in into this kind of GTM engineer function. I would say in the past you would look at, okay, you need to have this skill, you need to have Marketo skills, you need to have Salesforce skills. There were a lot of the skills you needed to have. Now more and more the soft skills are becoming more important. Are you— things are changing so fast, so are you someone who can really adapt with this speed? Do you have that kind of sense of fire within from a curiosity standpoint? Adaptability, you know, self-leadership, change management, collaborative skills, those are becoming even more important to get things done.

And like certifications. Yeah, to get things done now. And have you found a good way to test for that when you're hiring?

I mean, we're asking people to again solve problems, right? I mean, we're giving them more, more sort of homework. Okay, here's a problem. Here you can use, you know, Snowflake Intelligence, or you can use any of these AI tools. How do you approach this, you know, problem? And they can get, you know, a few days and can— if they don't want to participate in the exercise, they don't, they don't need to. But then they kind of sort of self-select out. Yeah. But if they're curious and want to join the team, right, they get to, to do one of these exercises and then we'll see what they, what they come back with. Yeah. How they handle this out again, their thought process around it.

And are you still, are you still doing most of that remote or now have you kind of flipped back to doing some of that in-person?

No more. I would say again, we are doing interviews, need to be very much in person. There could be parts of the process that might be virtual. Yeah, I think that's also what we learned that we have as a full mandate, right, that we cannot hire anyone before we have met this person. Because you can only really, again, judge those maybe kind of interpersonal skills, soft skills, if you're meeting someone in person. Yeah, whether it's versus sitting with someone over Zoom, right, and they're kind of, you know, using ChatGPT for every, for every answer. And how would you How would you know? Yeah, you can't do that, right, when you're in, in person. It's not about kind of, um, that you need to know it all, but it's that willingness to kind of really, really learn and want to be part of this transformation we're in right now.

And have you seen certain profiles when you're hiring now be more maybe successful than others?

Uh, again, it's not about, again, um, kind of, again, And before it was also very much like the long experience you have about something. You need to be able to kind of look at problems in a different way, right? So that's also something we're looking for. Yeah. Yeah. It's, you know, hey, you know, hiring, there's always going to be mistakes happening. You know, someone also from both sides, right? Someone maybe thought that they would be a good fit. They're not. There's always, always, that's always gonna be the issue.

Yeah, so now that, and then now that you have so many agents that between yourself and the rest of the marketing org are using on a daily basis, I just want to get your perspective on something that I feel like people talk about a lot on social media, which is like, okay, how much of marketing should still be humans and how much of it should be agentic? Like, how much, where do you draw that line of like what your team should do versus is what you let the agents or the data do for itself. Yeah.

I mean, we're seeing more and more that there's this skepticism around what's fake and what's real. Right. There's certain things which, you know, are fine with, you know, a dancing dog or something. We're fine. It's fun. It's okay that that's sort of fake. But there's certain things where we don't know, right? Is this real or not? Or not, yeah. Authenticity. Authenticity is becoming, I think, high value in that. And that's also what we can spend more time on, right? That is the human strength and value we can bring to marketing to make sure that, okay, what is the uniqueness of our brand, our company? That authenticity, right? That's where we can spend more time. Than we could before. Yep. In marketing.

Yeah. Another thing I want to ask was because like you have such a large at-scale marketing org, right? And like I know people are at like different scales when they're here at Saastr. Like, is there, are there certain pieces within marketing you feel like for driving new business they should like if they're maybe just starting out and maybe they only have one agent or two before they have, you know, multiple, you don't even know how many agents you have. Yeah. Endless, right? Endless agents, endless data. What are the ones you think they can maybe focus on that'd be the most impactful?

Yeah, I do think you need to start with the data. If you don't make that investment, it's gonna come back and bite you big time a year from now. It was the same thing, remember Salesforce hygiene, right? If you didn't get that right in the beginning when you spent time on that when you implemented Salesforce back in the day. We're at that stage now. If you don't really look at your data estate, making sure that you have all the data you will need to do what you want to do, that's, again, you're going to get very, very shallow value from AI if you don't get that right.

And how do you enforce— I mean, You work at a data company, so maybe it's just part of how you hire and your DNA, but do you ever find that you still have to enforce people across marketing and sales to actually put marketing attributed revenue or what campaigns are working? Do you feel like you still have to push a boulder uphill to get people to do that?

But that's also why that has been so much easier with Azeez, right? Things like, again, on the sales side, I mean, just updating sales forms or whatever that is, that is of course a very important system of record, that you can just do that directly from your agent. I mean, you're doing directly, I mean, it's automated, right? You just take the email conversation you just have with the customer and then you can update Salesforce automatically. So I think all that data hygiene is becoming easier than it was before.

Yeah, I think having a good foundation, obviously having a good foundation—

People skipped steps before, right? People skipped a lot of stuff. I don't wanna skip, you skipped a lot of steps and therefore you didn't have, had an incomplete picture what's going on. Yeah. But yeah, that's, we're making that easier as well. Yeah, yeah.

That's one of the supreme things I think about agents is I was saying earlier, like we, you know, when we started building out these Agentic workflows, we kind of took like a day or two to like clean all of our data because what you put in for the context is definitely affects what you get out. When it comes to your data. If it's building on a foundation that's wrong, it's just gonna duplicate that.

And I mean, it's getting so much more amplified and visible now. I think we're all getting emails or ads that are completely wrong, right? Where there's a completely different name, right? I'm getting that, you're getting the text messages, emails, and ads every day where the information is incorrect, right? You're going to be pretty kind of with that brand of that company. You're going to have a pretty, you know, first impression lasts for quite some time. It might take a while before you want to give that company a chance again. So and again, that all comes back to the data that is feeding these agents. If they don't— if that data isn't right, the agent is going to amplify that issue.

Yeah, we're going to just for folks too, if you have a question for Denise, I'm going to open up for questions in a little bit. So if you want to hit with the microphones. She'll answer some audience Q&A, but I'll ask one more question too, which is, um, is there, is there an agent now that you either you feel like is your most important or one that you're like, I, I couldn't imagine living without this one like a year ago?

Yeah, I mean, for me, we have our own agent, Snowflake, in intelligence essentially. That gives me— it's essentially like a ChatGPT interface to all our data, all our information. All our knowledge, but also external, I mean, you know, knowledge as well, that I wouldn't be able to live without today. Yeah. Yeah. Go ahead.

I know you have a question.

Can you hear me? Okay. Nope.

Try now.

Now can you hear me? Yeah.

Yep. So with a company at Snowflake scale, how do you determine the percentage of your marketing budget gets allocated to AI, then how do you manage and control that in a highly variable cost environment?

Yeah, actually for us, so the AI cost for any, again, all, all, all that is actually at an, at the company level. Yeah. At the moment it will probably change, you know, next year. But the, the, the, the, from a company perspective and from the leadership in our CEO, right? He doesn't want someone's budget to hold them back in terms of leaning in and experimenting at the moment. So essentially for us on the marketing side, essentially is unlimited access to all AI tools, essentially without getting, you know, anything allocated to marketing. Again, could change next year, but I think that was a wise decision because what's happening as well, right, that usage is going off the roof. It is, of course, you know, costing a lot of money, but, um, you cannot be like, okay, limit people, right? You know, you cannot use this anymore now, you need to slow down, right? Because then we slow down innovation. So I do think there's a, there's a quite big bucket allocated at the company level. Yeah. And I think that was a wise decision. It might, again, might change next year where we know a little bit more. Okay. What should the spend from the marketing organization be and the sales organization as well. We of course need to optimize our AI usage as well. There's clearly of course AI we're investing in or agents that are doing the same thing right across our company. So we're gonna have to kind of look at that. Yeah.

And from a full— Do you have certain agents or like tools that are company-wide that are, I mean obviously you have your own too, but like— Yeah, especially, yeah.

We have our own kind of GTM agent called Raven. That's really what our entire sales organization is using as well as the marketing organization. And that is very, again, that one is then also centrally managed. All the kind of skills within Raven is certified, because again, it comes back to, right, we don't have, we are very low risk, you know, tolerance. So our centralized data and they certify the skills within our own GTM agents, if you will. Yeah. And also making sure again that we're not duplicating, right? Because we have, we have the opportunity, everyone can essentially go out and build almost anything they want at the moment, right? So of course you're going to have duplications, efforts happening through that. So we need to kind of also rein things in. Yes, exactly. Yeah.

Yeah, you wanna go to your question? Yeah, okay.

Mic is working, cool. So I am building a GTM engineering function at a security company. It's been going pretty well, I think, but I know you have one at Snowflake. And so I was curious, first of all, how big is that team building the central services and automations and things like that? And then also, you know, because you have a lot of compliance things to worry about, obviously you can't just, give everyone access to everything. I'm a little bit struggling with that in my current situation. Like, I guess one of the approaches I was thinking about is having sort of like personal AI tools and information and stacks for each employee that they can do whatever with. And then like the central or certified ones like you were talking about. Yep. But if you could just elaborate a little more on how it's set up, how many people are kind of building those things and just a little more information.

Yeah, so we also have, again, a central AI data organization at Snowflake, and they have people on that team that are dedicated to sales and marketing. I think that might be around 20 people or so that are dedicated to sales and marketing. I mean, we have a quite large sales organization. Again, the demand on that team, right, it's they can't handle the demand that is coming in at the moment. We also now again have about 2 people on, on the marketing team that again sits a little bit closer to us today. But again, I said the demand— that you're gonna find the same thing probably, right? The demand for this person's time is gonna be very, very, very high. So someone is gonna just more again have to prioritize or Okay, what are then the agents or the different skills that we wanna prioritize, you know, essentially, right? So again, it's that, I think that's a challenge. I think everyone is gonna be faced with, they're gonna have to really prioritize hard for what that person is gonna focus on. Thank you. Yeah. Go to your question.

Hi, so my question is actually, If you had an enterprise that is big, globally speaking, how would you start your AI optimization inside the company? Because it's very challenging in big companies to have that.

So how would you structure that internally?

Yeah, again, there's nothing more important than having your data estate in order, and that's not siloed on the market organization. Right. It's so incredibly important to have, you know, you know, a chief data officer or I mean, now our chief data officer, she's also chief data and AI officer. So she is sort of running our data strategy and AI strategy for the company. And a big part is around just, you know, governance, right? Governance of our data. What— how should access control work for our data? Do we have sort of unified semantic layers of, of our data for our company? So that, that's very super important for a company, you know, at our, our scale. And I think again, right, the power of the agents we have in marketing is not just the data we have on the marketing side, it's the data knowledge that we have entire, entire, uh, have entire across, across our entire organization. So I think that function around data and AI needs needs to be a central, you know, function, if that helps. But then if you have that, right, then you can give more access to the entire marketing department to work with that data, right? So it's, yeah, yeah.

Thank you. Thank you. One thing that stood out, by the way, I love Snowflake, so thanks for being here, but the token leaderboard kind of stood out. And it might be not the most central thing to all the different things you're doing, but it's an interesting— use it as an anecdote— this tension with that and the idea of tracking outcomes. And there might be a correlation with more usage and more outcomes, but there's also the business impact of more costs. So I wonder if you've considered how you could build leaderboards more oriented towards marketing outcomes, if it's more content, more leads, rather than your current total.

Exactly. As I said, right, when I bring up the leaderboard, it's more about highlighting those individuals who kind of just, you know, leaning in, experimenting. And most of those people on the leaderboards, they are helping the rest of the team, right? So they're very much— they're becoming— that was never the intention, right? But they're probably 7, 8 people that have become true power users. Those are the ones who are now going in and helping, let's say, the communications team with their, with their AI hackathon, for instance. So they're very much like, they're not staying siloed in their function, and they're really just out helping the rest of the team. So they're true sort of power users at that standpoint. And also, I always bring that up, right, when I show those, you know, top 3 people people, right, every month. It's like, you know what, it's not about, again, the number of tokens you're using, right? It's really about the business outcomes. It doesn't matter if you only have 100 tokens, right? Again, the most important thing is that you— we're all leaning in here. We're leaning in, you know, we're learning, we're being curious, and we're always thinking about a hard problem. And it's like, is there a way for me to do this with AI in a more effective, in a way So important to not reward token usage only. Yeah. Thank you. Go ahead.

Yeah, so with the drop in website activity due to AO, GO, paid social, peer-to-peer review sites, how are you thinking about stitching together that prospector journey and then creating measurement in a way that really hasn't been done super well thus far, right? Especially like that Tofuu activity. Therefore, then for someone like you to understand like how much should we invest in Tofuu and what return are we gonna see on it, how long down the road, that sort of thing.

Yeah, I mean, we're seeing now, we're of course capturing that first engagement at different channels than before. So we were investing in other channels that we can open up optimise further. Yes, we're seeing, you know, websites engagement, you know, going down. It's more about, again, okay, how are we optimising all the other different channels that we have to kind of reach, reach our targets? And also, again, more importantly, it's of course about optimising the conversion, right? And that's actually what AI has helped us a lot with. I mean, I think actually have my head of head of growth marketing is in here, the room somewhere, but she's always looking at, okay, the leaky funnel, right? So she talks about the leaky funnel, and that's something also where AI helped us come in to really look at, kind of, okay, what is causing the leakage, right? What is the recommendation, you know, for better conversions? Yeah, so conversions is gonna probably more important than the top of the funnel, to your point.

Yeah. Lou, how do you decide who owns what?

And what I mean by that is, for example, we just gave a bunch of people a Claude license at our company.

Now there's a million agents getting built, but so much of it's cross-functional dependent as well.

And so one area I'm trying to figure out is who should own what and how do you decide?

Is it just who built the agent?

Owns it, or is it, is it a VP, or is it a C-level person? Like, how do you decide who gets to drive the direction of the growth?

Yeah, for us right now, we're basically just encouraging everyone at every level of the company to kind of, to, to build, you know, agents, to build, you know, apps, you know, on their own. It's gonna come to a point where we're gonna have to kind of rein this in a little bit, and And again, as I mentioned before, we do have also a centralized team who goes in if there is a skill that we have developed, right, that is going to be used by more people. It's going to have to be certified by them to really make sure, again, is this skill, this agent, actually delivering the correct behavior. So we have those resources or people people who go in and help us, you know, with that. But again, there's also— we, again, said we're very much encouraging a lot of experimenting, usage, building at every level of the company right now. But I think we might find ourselves like, let's say, 10, 15 years ago, right, with SaaS applications, right? It was the same, you know, when marketing started buying technology, right? And we probably had at least 100 different SaaS applications applications, and then sort of the IT department finally coming in and like, okay, we need to actually have order here. We need to understand what is being used across our company, what is really the usage of this, of this applications. I think we're gonna see probably similar thing happening in, from an agent standpoint, right? Where there's gonna have to be, you know, a control layer around this.

Yeah, it's funny. I was just gonna, I'll, I'll hit your question in one second. Just, um, one thing I was thinking, like we were talking about how like certifications are pretty much dead, like in your interviewing process, but do you, do you see yourself in like a couple months actually doing like an internal AI certification process now?

Uh, abso— we, we actually talk, we haven't, we have plans for that. You have? Yeah. Um, exactly. And, um, yeah, I think we're gonna have to definitely have more of that across the team as well. Yeah. Okay, go ahead.

Hi, my organization is currently consolidating our data and we are a customer of Snowflake. So we're incorporating our call transcripts, our CRM, we're incorporating our Power BI data all into one place. That's not the end of the game though. Like, I feel like eventually we will want to build agents that can act externally, you know, sending email to our customers or, you know, interacting with them, it'll have to pull in assets like marketing content, white papers, webinars, blogs. Do you use a content management system to house all of that? That's where my— I'm not really sure what direction to go in for our organization. Do we purchase something? We have Snowflake, we have SharePoint. Does it get housed in our CRM? I'm curious what you're doing on your side.

Yeah, we have a content management platform. Actually, I actually don't remember the name. I can We can talk after. I can follow up with you on that. We have that. And of course, we have a lot of content on Google Drive. We have different— maybe we also have different e-learning systems or things like that. So data sits on a lot of different systems. The good thing is that we can actually access that with things like Snowflake Intelligence. We can access all that information. But we're also looking at, again, OK, how do we build now our own kind of, you know, email functionality, etc. That's kind of— it comes back to kind of the activation, right? We have developed parts of the workflow, but the last part is, of course, the activation part. Email is a big part of that, and that's what we're looking at next. Yeah, go ahead.

Is this going on? There we go. Thanks. So I run a marketing organization as well. And as we get better and stronger and faster all the time, particularly at content and at launches, one of the challenges that we're running into is that we are running the risk of overwhelming sales on the enablement side. So how are you handling having your sales team feel confident and enabled at the same pace?

Yeah. Yeah, and actually just a few months ago we moved our whole sales enablement, partner enablement, customer training over to marketing. And the reason for that is also because a lot— we build a lot of the content, right? Content started to be duplicated. There were duplicate efforts in terms of content building across our company. And the key thing now with AI and content, while you build content once and you can create all these derivative assets, right, for different target segments including sales and partners. So the journey we're in right now is that still we have our— again, we have something called Snow Search. That's essentially our, you know, biweekly sales training programs that we still have that in parallel, but at the same time we're building this complete sort of self-service Luis, enablement agent for our sales organization. So they basically can get the answers and training at the time where they need it. Because again, so much time is wasted, right, by attending training sessions where it's either too advanced or it's too kind of too easy for you, right? So we're very much focused on building this individualized journeys for our sellers. Things like also now you can do, there's a platform called Jodly. I don't know if you know that one. We can do, I mean, even role-playing, right? You have an agent, yeah, but you're role-playing with— think about the time before you had to spend doing that with your manager or others, right? It was kind of pretty horrific to do that. Now you can do that, train yourself, you know, with an agent who has all the intelligence in your company, can really help you, you know, perfect perfect your pitch or perfect your presentation you have to your customer tomorrow. So I think that whole world of enablement training is gonna be up for massive disruption next. And we're just gonna be able to again meet people where they are, right? Give them the right intelligence at the phase they're in essentially. So we, yeah. We have a lot of work happening on that at the moment.

All right, we'll do yours as the last question. Yep.

Thanks, Denise, for being an inspiration to many of us in the CMO community as well. So let me lead with that. I know that Snowflake has led the discussion around one team GTM.

I'm curious to understand how you see that evolving in the AI world and what do you tell the marketers marketers.

I'm actually struggling to tell the young marketers on my team what the future of this function looks like. Yeah, I think, I mean, we, I don't think anyone can paint the clear picture of what it looks like, but we can be a part of shaping it and building it, right? That's kind of really how I try to inspire, motivating the team, right? You know, you can be a part of this transformation or you can sort of opt out out essentially. But the second part in terms of one team, you're talking about the alignment with the sales organization, right? Yeah. I don't think there's anything more important. It remains how we operate on a day-to-day basis. Again, we had a CRO that I worked with for almost 10 years. We were so aligned. We have a new CRO now. Been with the company for a long time, so I'm building the same relationship, you know, with him. And it's also important to— it's not about just about me and the CRO. It starts, of course, with us having the same exact objectives, but you also need to make sure that you have that alignment right across the entire market organization, the sales organization, right? So if I look at the sales leadership team, right, I'll make sure we have the VPs under CRO I have a counterpart right on the marketing side that is fully aligned with that individual. So it's so important to have those kind of relationships built at every kind of level, you know, of the, of the organization. Awesome.

Well, Denise, if folks have other questions for you, how can they— what's the best way to reach you?

Yeah, I mean, yeah, probably LinkedIn is probably the best way to send me a message. And again, Hey, so, so fun to hang out with all of you and the, and the SaaStr community. So thank you for having me.

It's always nice to have you. Thank you for having me back. Thank you guys.

Give a warm welcome to the co-founder and CEO of Monaco, Sam Blonde.

All right, SaaStr 2026, bigger and better than ever. Uh, thanks everybody for, uh, attending. My name's Sam Blond. I'm co-founder and CEO of Monaco. Excited to be here today. Um, I think I, uh, take maybe the world record for longest session title. So, um, by the time I finish reading it, the session will be over, but I was a VC at Founders Fund. I saw 200 AI sales startups, then I built my own. Here's what everyone gets wrong. I think a couple of things that the everyone gets wrong is alluding to. The first is the company building aspect, sort of like defining the type of company that you want to build, the market that you want to build in, those sorts of things. And then the second thing is oriented around AI and sort of like how companies are applying AI AI today within go-to-market. Um, and so I'll jump right in. Um, just quick agenda today and maybe some, uh, hopeful takeaways that folks walk away with. The first, um, lessons on company building from Founders Fund. Uh, I'll spend a couple slides talking about my learnings from meeting with a bunch of different founders and how I applied those to then building Monaco. Um, the next thing is how you should be applying AI in go-to-market. And so Monaco who does a lot of this stuff, but this certainly won't be a Monaco commercial. She's talking about like, what are the, the aspects or what are the workflows that today AI is quite good at? Um, and then actually the inverse, what is AI sort of less good at, uh, and how should we be leveraging AI today in go-to-market? Uh, lastly, um, in that sort of natural segue, try and give some tactical ideas on how to grow revenue faster at your your startup. Okay, jumping in. Um, lots of folks flowing in, so thanks for, uh, coming. You haven't missed much. Uh, we're just getting in, talking about the agenda for today, um, some lessons on company building from Founders Fund, uh, how we should be applying AI in go-to-market, and some tactical ideas on how you can grow revenue faster at your startup. Okay, uh, Again, sort of back to the session title. Um, while I was with Founders Fund, and my background is go-to-market and sales, and so I think because of that, lots of folks that were building in this space would meet with me to sort of like pitch the company as a potential, uh, target for Founders Fund to invest in. And so met with hundreds of companies in this category of go-to-market. Um, and I think that each, uh, company sort of fit into one of two different types of businesses. The first is like, this is— this should be a feature and not a company. I'll give an example of what I mean by that. And nobody should take these either personally or literally. I'm just like grabbing for straws here. This isn't a company that I'm aware of and hopefully not a company that exists. But I meet with somebody and it'd be like, we take the content from a call recording and then we update a field or a set of fields with that content in your CRM. And it's sort of like, okay, that makes sense. You're taking data in this sort of unstructured way from a call recording, and you're then updating the CRM form fields with that data so the rep doesn't have to do it. Um, and so as a like workflow, that makes a lot of sense. But as a company, shouldn't that just be a feature of either the call recording platform or the CRM platform. I don't know that this needs to be like an independent business in this concept of point solutions. I think we got like way too pointed, especially towards the end of the sort of like SaaS era and sort of beginning of AI. The second category was what I would describe in sort of a saying that we had at Founders Fund was a solution in search of a problem. Give a quick, like sort of random example here. This isn't an actual company. But something like, hey, we do video creation. And at the end of every week, we send a video to every rep on the team. And it's like a recap of their week, the things that they did well that week, the things that they did poorly that week, and maybe some like guidance on how to improve. And I'm kind of like, huh, like, okay, I guess that sounds kind of interesting. Are people really asking for this though? Like, who are you meeting with that's saying, Gosh, I'd love like videos for reps at the end of every week telling them what they did well and what they didn't do well. And it's kind of like, you know, maybe the reaction from the person that's building this company is, well, people aren't asking for it, but that's because it doesn't exist. And, you know, like if people told Henry Ford what to build, it'd be a faster horse. And it'd kind of be like, okay, I don't know that you did the like amount of customer research on this one. And so there's this like graveyard of companies that were either sort of like two-pointed as point solutions or these solutions in search of a problem. And so, um, a, a lot of this sort of discovery led me eventually, and I'll sort of like put a pin on this, do not be a solution in search of a problem. Uh, all of this discovery sort of led me to, um, starting to build Monoco. Uh, and there are a few things, and this will say, sort of dovetail into how I recommend founders broadly think about building companies. But for Monaco specifically, the founder market fit here is quite strong for me. So I'm a non-technical founder. My background is sales and go-to-market. And through that lens, I believe there's only one type of technology company that I'm qualified to be the founder of, and that is a sales or go-to-market technology company. I had between myself myself and co-founders hundreds of customer conversations just trying to understand what is like the biggest problem that you are faced with today in go-to-market tooling. And one of the things that regularly came up was this concept of there are like so many options. We're an early stage startup, we're 12 people, we have 1 or 2 sales reps. Do I go with the like HubSpot or the Adio or the Salesforce and and that only does the database side for me. Do I go with the like Apollo or the ZoomInfo to give me the actual data? And for email sequencing, I've got like Outreach and Salesloft and like I'm just in sort of like tool sort of overload. I don't know what works. I don't know how to use these tools. And, um, this is like far more complicated than it needs to be. And so that was a very recurring theme that I heard and certainly resonated with from many years of building go-to-market organizations in startups. The second thing— or excuse me, the third thing around Monoco inception. We are in the early innings now, maybe trending towards mid innings of this AI platform shift. And I think there are two applications of that for Monoco. The first is we at Monoco and the world around us, we're just able to build software much faster. And so the reason that there were these point solutions, historically speaking, is because it was hard for one company to do everything because it just took so many development resources to be able to build all of the different suites of products that would be available. And so you had this like best of breed point solution model. Today, it is very easy for us to build things that historically speaking took large teams large amounts of time. We can build things like, you know, call recorders and email sequencers and database providers and all of these things. We've done all of it inside of 18 months. The second thing that is related to this AI platform shift is we are in the sort of like early innings at the onset of what will be, what will emerge new leaders in each of the large enterprise software categories. If we think about go-to-market market technology as a category. Salesforce today is the market leader. Salesforce was, um, enabled to become that market leader because of the platform shift from on-prem technology like Siebel to cloud-native platforms like Salesforce. AI sort of rhymes here in the way that it rhymes. You have these pre-AI system architecture companies like Salesforce, HubSpot, Adeo. They were architected to be reactive to user input. What for Monaco is AI native, and what that means is we were actually architected for agents to be the first line of defense before the user has any interaction with the platform. There's already been a lot of work that has been done by technology and sort of specifically in a cliché way, agents, before the user even sees the user interface that is basically sort of a front row seat to the work that has been done on their behalf. It is very difficult, if not impossible, for a pre-AI system architecture platform to pivot to being AI native. You have to either start from scratch like we did. I, I don't know the alternative to that. Um, and so because of this like culmination of forces, Monaco, uh, was, uh, born. And then sort of applying, uh, this background on my experience at Founders Fund, the learnings in the space and then the inception of Moneco. Here are some 3— here are 3 primary lessons on company building just broadly. The first is, um, build in a space that you understand better than anyone. I, I think for me, I, I have lots of experience in go-to-market, specifically building and scaling, uh, startup sales organizations. That is exactly what Moneco does. Um, the sec— um, there's maybe like one other example that I can give here. I was at Brex for a number of years, and Pedro and Enrique, who are the founders of Brex, they had a startup in Brazil called Pagarmi, and Pagarmi was a fintech company. It was like a version of Stripe in Brazil. They came to the United States, they did YC, they started a virtual reality company called Vyond while they were in YC. No experience in the space, thought it was like kind of cool and that virtual reality was going to be the next big thing. And one thing that started happening is that— and this sort of dovetails into the second bullet point here— they were in YC and none of their batchmates could get corporate cards. And so they had this like insight that was a big pain point of a bunch of people that were surrounding them in YC, and they pivoted to this space that they understand better than anyone. This was the second time being founders in the fintech space. If we look at Brex's largest competitor, Ramp, same idea. Their founders had sold a company previously to Capital One in the fintech space. And I think there's a recurring theme here that the companies that, um, realize the largest outcomes, those founders prior to starting those companies, they have deep experience and insights into the category that they are building in. Uh, and then, um, lastly, Lastly, market timing matters. I think this is like a pretty exciting time to be building companies right now with the onset of, uh, AI and the platform shift that comes along with it. Uh, and so Monaco, so like checks these 3 boxes that I think generally apply to other folks. Okay. Um, I'm gonna start talking a little bit about, uh, the state of AI and go-to-market today. So, um, we start building Monaco, this is about, uh, 18 to 24 months ago. I'm going to sort of catch us up to lots of learnings that we have had and apply those to how, how should we be leveraging AI today. So here is a list of things that AI is just better than humans at, and we should interpret this as like, how should we be using AI in go-to-market today? The first is building and scoring your TAM. So, uh, for Monaco, we are selling to startups. Uh, historically speaking, sort of pre-AI, the way that a company would build and score their TAM is with human, uh, people. Um, and we would crawl the internet and we would run and filter lists and export the data into a database. And we would do what's called prospecting where we'd go on LinkedIn and we'd try and find companies that Did it meet a certain criteria? Um, AI is better at that than, uh, humans are. And so we shouldn't be spending our time on that. Scoring is another concept that you can just in a prompt program how you want your TAM scored. Not all companies in your TAM are created equal. We sell to startups. That's a very broad category, probably 20 or 30,000 companies meet that definition, there are companies that are better fits for us based off of firmographic information. Can be things like, where's the company headquartered? How many people does this company have? What is their business model? Are they B2B or are they B2C? Are they venture-backed or are they, uh, uh, non-venture-backed? All of those criteria lead to what should be some sort of a, a methodology in prioritizing companies. And so this is the The scoring of your TAM. Historically, this was done by RevOps people that were sort of in spreadsheets running reports, writing this algorithm. AI can do this pretty easily in like a day or two from the time that you want to apply that scoring methodology to your entire database. Enriching and overlaying signals. So we now have this TAM of companies, everyone that meets our ICP definition. And it is scored and prioritized based off of the fit for us to sell to that company. We now want to enrich these companies in our TAM broadly with lots of data. Agents can do that in a very streamlined and automated way. We then also want to pick up on signals. These are going to be things like this company just raised funding. Uh, for Monaco, that's one that would apply. Another common one is something like this company is hiring for a role that we may want to either replace that role with our technology or we may want to sell to this person as soon as they've been hired. So that's another event that could be a signal. Historically speaking, and I remember back in the days at Brex, we would have sales reps that every day they would read some like online email newsletter that would be the funding announcements of different companies every day. and then they would manually reach out to those companies. Don't have to do that anymore. AI is very good at picking up on signals that happen in real time and then applying that to workflows that we'll get to in just a moment, um, which is executing outbound. So, um, you should have sort of automated outbound that is based off of this TAM that we have built. So all of our ICP companies. It is prioritized based off of the companies that we are the best fits for. We're leveraging signals like, did this company fundraise? Is this person new at the company? Have they visited our website? All sorts of signals that we can incorporate into the messaging. We should then be multichannel. So we should have automated outbound going out at a minimum through LinkedIn and through email. Uh, and all of this should be done in a totally automated way. We should never have to think about this. We may want to monitor this and manage this and make sure it's actually happening. And when we get responses from some people, we may want to be the people that are sort of like handling those objections and scheduling the meetings rather than the AI. But everything that I've just described leading up to this point, AI is better than humans at, and we should be spending near zero time on doing these things because we're worse at it. Lastly, you schedule these meetings, you start to have a pipeline of opportunities. And AI is very good at pipeline maintenance. So what do I mean by that? We can take unstructured data from emails and phone calls and we can make sense of that and do lots of things with it. The first is we can automate all of your opportunity creation. And opportunity updates. We should no longer be like clicking plus for an opportunity, typing in the name of the company, moving the opportunity from stage 2 to stage 3. We should no longer be updating form fields with information that we have learned from the calls. We should no longer be clicking things like add a new task as a reminder to follow up with this person in 2 weeks in case I forget or in case they don't respond to my email. AI is very good at making sense of all of this structured data, creating opportunities for us, updating our pipeline, reminding us who are the opportunities that are slipping through the cracks. This should save us lots of time. And again, AI is better at this stuff. I remember as a sales rep, I would get off calls and I'd sort of be racking my brain on like, I don't know, we'll go back to the, uh, Brex days, which corporate card did they say they were users of? I can't remember, like, who they're on right now. You know who does remember is the agent that has access to the call recording. So it's not just that, like, AI can replicate the labor here. It's that AI is actually better at doing a lot of these things today than we are. All right. So what is AI less good at? And then we'll move into how should we make sense of all of this? Okay, so, um, AI is worse than humans today at meeting with customers and building relationships. We could probably imagine some, like, futuristic world where you're talking to, like, an avatar, and that person is, um, maybe selling to you rather than the, uh, in my case, founder of the company or a sales rep. For Monaco, companies like Monaco, and, um, upmarket enterprise software companies, buyers actually want to talk to a person. Uh, they don't want to talk to an avatar. Nobody is purchasing a complex enterprise software platform, um, even as Monaco, or you can go into the sort of like NetSuits of the world without having conversations about the needs of the business. How do we implement this? Those sorts of things are still, um, human-led. And, um, the, the buyers of products like this actually want to be sold to. So AI today is not a replacement for meeting with customers and building relationships. The second category that AI is worse than humans is creative and/or operationally complex campaigns. So, um, I talked about this stuff that was all online, that was effectively building our TAM, running automated outbound, those sorts of things. There are lots of campaigns that we will want to do to build our brand that are either creative or complex. I'll give an example or two now, but we'll go much deeper later on. Just at Saastr today, we have like the plane with the Monaco banner that's trailing the plane right now. AI didn't come up with that idea. AI didn't reach out out to the company that has the planes. That was all sort of like human-led and human-orchestrated. The same thing, we did a Series B announcement yesterday. We did a video that accompanied the announcement of funding that we received. AI is not going to be able to just like, you know, vibe code that video. You should be very, very thoughtful about like, what is the content and message that we want to get across? And that is worth spending a lot of time on. And humans are better at these either creative, like a video, or operationally complex, like a banner trailing a plane, campaigns that I think are absolutely critical to coincide with what I would call just table stakes, which is outbound. So we will make sense of a lot of this. Let's see. All right. So the result, and this is one of the things that I think people, some people are getting wrong today, where it's like I can just outsource my go-to-market to AI. I'm a technical or product-oriented founder. And now that there's all of this automation through Monaco-like companies, I just bring in Monaco and that allows me to then focus exclusively on writing code, building product, the things that like I'm very good at. That is, that is not the right way of thinking about AI and certainly not the right way of leveraging AI in go-to-market. So how should we be leveraging AI? And then we'll get into sort of tactically some of the things that we should do. So if you're spending time on things that AI is better than humans at, you are likely behind your competitors. And so there was this list of things that I had on a previous slide where it was specifically like, if you are going on LinkedIn and trying to find companies or buyers at companies, if you are writing messages manually to send out to people that you may want to engage with. If you are logging into some sort of like pre-AI sales platform and running lists of company types and filtering those lists and changing the filters that you have and exporting the data, um, that is, um, something that again, AI is better at today than even the best human at doing this sort of, uh, workflow. And, and that is, um, a waste of your time and there is an opportunity cost that you could be spending your time on the highest ROI activities that I'll talk about in the next two bullets. My day is full, just wall to wall of customer calls, and this is potential customer calls and existing customer calls. That is the highest ROI use of my time, and I would argue it's the highest highest ROI use of most founders' time, you know, potentially unless you're like in a PLG type, uh, uh, customer acquisition motion. Um, but I now today am able to spend so much time with customers because of the leverage that AI provides for me that I would otherwise be spending on a lot of the things that I described that AI is, um, now better at. Um, and then lastly, we spend relative to other companies, I, I suspect we spend a disproportionate amount of time and effort and focus on these like creative and operationally complex campaigns. And I'll give some examples of what those look like in, I believe, the next slide. So here's, here's like the way that we should think about building our brand and growing revenue faster. Every single month you should have a major initiative. And some of these things are organic. This is a perfect example of one of those things. So for us, Monaco is a sponsor at SaaStr. It's an event. It's perfect for us. A lot of our ICP will be here. And so this is like an organic thing that we can spend calories on figuring out how do we make the most of this event. Other examples we launched in early February, So we had a fundraise announcement that accompanied our product launch in February. Yesterday we announced our Series B funding event. These are sort of organic milestones that you should be thoughtful and deliberate about how you can make the most of them. Now, between February and today, we didn't have anything that was this sort of like organic initiative that we could orient the company around. To really stand out and drive brand awareness. So sometimes, and in fact oftentimes, you're going to have to manufacture some of this stuff. An example of something that we have done would be, we, name of the company is Monaco. I think Monaco, there's an association with like a big casino there and maybe gambling. And so we throw what we've called the Monaco Invitational. So it is a big poker tournament. We give away $100,000 to the winners. We threw like this huge party at a beautiful venue in San Francisco. It got a bunch of sort of both press and I think buzz on social media. And it added to a lot of the other things that we were doing that were organic. The fundraise announcement, the product launch, this event. And it's the type of thing where you, you, you invest in these things and they compound. So if we think about the things that Monaco has taken advantage of in a short window of time, February was our product launch and our Series A. We have billboards all over San Francisco. We have the Monaco Invitational. We're doing the second one in— second one in June. We have SaaStr. We had our Series B announcement yesterday. We're now 3 months almost to the day from the time that we came out of stealth. And we've been very deliberate about having these just continued moments of big brand awareness campaigns, and it's very intentional, and it's the type of thing where it's like, you know, it's not 1 1 1, maybe +1 if we're thinking about 3 or 4 of these things, equals 4. Each of these compounds on top of one another, and so our billboards work better because of the poker tournament. You see the billboards, you hear about the poker tournament, you see the Series B announcement. You just want to be very intentional about this steady diet of having major announcements that are going sort of viral on social and building your brand. So this probably sounds like a little bit easier said than done. So I want to talk about maybe a process recommendation on how you should apply this inside of your company so that you can start doing this. Now you want like sort of 30 days in advance for a lot of this stuff, because it takes time to plan a major poker tournament and get invites out and those sorts of things. And so each month, you should sort of like figure out what next month's big thing is. You should identify a group of people inside of the company. Maybe it's like, if it's a 5-person company, it's all 5 of you. If it's a 50-person company, it's like the 5 most creative people 5 creative people inside of that company, the CMO and maybe somebody else in marketing and maybe somebody in sales who, you know, is really creative and likes to be part of these things and the CEO. And again, I don't know the exact personas of the people in each individual company, but let's pick 5 folks and we get together and there's a prompt that's like 48 hours in advance. And that prompt is we have our Series A and product launch. Launch announcement coming up on February 11th. Everybody come up with the craziest ideas that you have that we should do to amplify this product launch and fundraise announcement. We have SaaStr. It's coming up on, gosh, what's today, May 12th or thereabouts, coming up on May 12th, and we want to go big at this event. Let's the 5 of us come up with the craziest ideas that we can have that can amplify our presence at the event. You get in a room 48 hours later, hopefully people have been ideating on the things that they think would have the biggest impact. Some of us have budget constraints, we probably wanna put those in place. Um, and then you get on the whiteboard 5 ideas and you probably pick 2 or 3 of them. And the outcome of following this process and being rigorous about it on a month-to-month basis is that you don't show up to SaaStr with a booth giving away an iPad or an iMac mini or what every other booth is doing at the event. You show up to SaaStr and you have planes flying around with banners and you have money machines and you have magicians and you have poker tables at your booth and your booth is the most crowded one at the entire event. And so follow that process, be rigorous about the sort of like, we do this on a monthly basis, basis. Of course, there are going to be capital constraints that, you know, not everybody— although it's surprisingly inexpensive to do this plane thing, uh, not everybody's going to be able to afford to do the like plane with the banner behind it. Um, but gosh, if you're doing a $100,000 sponsorship, the $15,000 plane flying around with the banner for 3 days, it's actually like pretty reasonable ROI. Um, okay, uh, I think we have a few minutes left. Um, We'll do Q&A. I don't know if we're set up for Q&A, actually. But if we're not, if anybody just wants to like stand up and I'll try and hear your question and repeat it to the audience. And we don't have to force it. It looks like no questions, so we can wrap. Thank you so much, everybody, for your time. Really appreciate it. Have an awesome rest of SASTR.