How Maven AGI is Creating Delightful CX with Agentic AI | Agentic EP 01

Maven is creating AI agents for enterprise customer experience. Part of the reason why we built Maven is that everyone really struggled because the cost of being able to serve customers is expensive. If customers aren't getting value out of your product, they churn.

Changing behavior is very, very hard. People don't change their behavior. So create a solution that injects into the existing workflows.

When we looked at the technology that was there and we said this is going to unlock complete paradigm shifts in the way that people are interacting with.

All three of you guys are perfect, well balanced.

Co founding team what is it and.

Why is it so special?

I knew I could build tech, but I can't close enterprise deals. But this guy can. This guy can build awesome product.

I think about it like a three legged stool that if any one leg wasn't here, it would collapse.

When you look at, you know, kind of historically great companies in Boston, HubSpot took them a couple years to get to a million in ARR. Klaviyo took them three years in order to get to a million in ARR. UiPath it took them 10 years to get to a million ARR. We exceeded all of those numbers and the reason why is that we've been able to be intentional about what we wanted to build, who we wanted we're targeting and how we're going to do that.

Welcome to the Agentic podcast where we dive deep with the pioneer AI founders reimagining business and investors betting on them. We bring you the war stories, insights and perspectives you won't hear anywhere else. I'm your host, George Van Darren, a general partner of Untapped Ventures, a VC fund that invests in early stage AI startups.

Let's go.

Welcome to the very first episode of the Agentic podcast and today we have three very special guests with us. They are the co founders of Maven AGI and raised close to $30 million of early stage funding in a year. The CEO was formerly the Global Vice President of Customer success and strategy at HubSpot where he led a team of 1000 driving customer retention, revenue growth and value realization for over 200,000 customers worldwide. The CTO was part of four startups with one of them exiting to Google and he holds over 55 patents and the CPO, the chief product officer is X Google Xstripe XPCG. They're a really amazing crew, so you're in for a real treat. So let's welcome Jonathan Corbin, Sammy Shalavi and Eugene Mann to the show.

So let's go ahead and just dive right in.

One quick note before we dive in. Unfortunately, the audio from my mic was corrupted. Luckily, we had a backup iPhone recording to save the day. You'll notice a difference in audio quality between me and my guests, but the good news is they're the stars of the show. And as Reid Hoffman says, if you're not embarrassed by your first product, you launched too late. So, without further ado, here is episode one.

Jonathan, maybe you could tell us about Maven AGI. What is it and why is it so special?

Maven is creating AI agents for enterprise customer experience. So part of the reason why we built Maven was because some of the experiences that we had in our prior lives. And so my experience was deep in the CX account management space, working for companies like Adobe and Marketo, Sprinklr, and most recently HubSpot. And what we found, not just at the companies I was working at, but at Google and Stripe, where Eugene and Sami were at, is that everyone really struggled because the cost of being able to serve customers, to be able to help them get value out of the product is expensive. It's 40 to $60 on average per customer interaction. And so if customers aren't getting value out of your product, they churn. And that's bad for, you know, especially for B2B SaaS companies, for consumer companies, for anyone. Right. If you're paying to acquire a customer and they leave without getting value, your product, that makes you sad. And so as we were going through and kind of looking at the challenges that we were facing in our respective organizations, we turned externally and we said, hey, you know, let's go talk to other folks and see if they're also experiencing these challenges of, like, you have increasing budgets, poor customer experience. How do we go about serving customers the way that they want to be served? How do we help them get more value out of the product? And what we found is it was a universal truth. Every company, large or small, whether they were B2B SaaS or healthcare or a marketplace, all had the same challenges. And so it's a challenge that we saw universally and said, we're going to go out there and we're going to do something about that. And so that was the reason why we started Maven. Now, the origin of Maven probably started much earlier for Samy. I know you were thinking deeply about experience during your time at mit.

Yeah, when I was a graduate student at mit, I worked on one of the first personalization engines. At the time, this is like pre Internet. So I'm dating myself and I've been doing personalization AI ever since, whether it was at my, at Lotus, the NY Next startup at Google where I was running Google News. And you just see the power of creating highly personalized experiences actually solves real problems in the world. I've been fortunate. I've done a bunch of companies, I've been involved with many large big tech companies and one of the things you learn over the years is you got to solve a real problem. And we just suddenly had this unbelievably disruptive technology that has entered into the marketplace that has commoditized reasoning at scale. And with that you're able to solve kind of this problems that have been very difficult to solve historically. But you have to start with real pain, real problems. And we've all lived this particular problem as part of our experiences. Everybody wants to create a great customer experiences, it's just hard. Nobody wants to build bad customer experience. It's just a really hard thing. So there is tremendous value to be created and just economically it's not, not a, it's not a healthy, it scales linearly with your, with your growth and it just exhibits no economies of scale.

Right. So, so I love that we, we dove right in and kind of talked about what Maven AGI is, but if we take a step back, you're starting to do that a little bit. Maybe we could talk about, you know, formally introduce yourselves, each of you and let's talk about what was your experiences and specifically the insights that led to the founding. So let's meet each of you and then talk about those insights and then ultimately dive deep into the origin story. I know a couple of you guys knew each other from the, from the Boston scene playing basketball and kind of knew of each other. And you know, I think what's really interesting as new founders are building magentic startups is like those really early days going through ideation and validation and that the product market fit journey which happened kind of quickly for you guys. But let's do it in two stages. First, let's get to meet you guys and your experience and insights. And then second part, dive deep into the origin story.

Sami Shalabi Founder CTO at Maven AGI A little bit of background about myself. Maven's my fourth company. Started my career at Lotus working on Lotus Notes for those it's not the car company, it's the software company. After seven years there I started my first company which was a mobile social network that ended up selling that to Google. Second company Was a Netflix for indie filmmakers that ended up being really big in the Middle east and is now doing originals. Involved with a third company that was helping cancer patients navigate their care. Everything from med trackers to treatment plans. The clinical trial matching grew that from zero to a quarter of the US population. And also spent 12 years at Google, started their news and magazine business. And much later, around the presidential election between Hillary and Trump, Google's relationship with the news industry went complete sideways. Filter bubbles, fake news. And we ended up trying to understand that problem and realized that people did not think of news as being the front page of the New York Times, but the things they cared about. So we embarked on building the largest real time personalization engine at Google at the time. And while I was doing that, I had infinite headcount. So not only was I hiring, I was looking for companies to buy, to grow the engineering team. And that's how I met this wonderful man, Eugene. And then through him I met Jonathan. And this was many, many years ago. Eventually convinced Eugene to join Google and run all of product personalization. And we have been doing monthly dinners for many, many years looking for excuses to work together.

Nice, nice. So before we bounce to the next person, one thing you Left out was 55 patents, is that right?

Yes.

How does one create 55 patents?

I don't know, work at a place that'll happily pay for all the cool ideas that you develop. I mean, patenting is a codific some of the new things. And I've been very fortunate working on kind of the bleeding edge of technology. And whenever you kind of create something new, there is a tremendous amount of desirability to patent the work.

Was there like some 80, 20 rule with the patents where like 20% are really valuable and monetizable or like what was your experience?

I don't think software patents are really monetizable. They're really kind of a protective vehicle to capture innovation, to kind of create a. It's a way to kind of create a moat. That's I think the first part. The second is if you're creating, especially on the startup side, it just creates an environment where you can assign value to an asset within the company. So, so there's a, there's, there's value in creating the patents. And I've been very fortunate that I've been on a lot of bleeding edge stuff.

Presumably he's also written a lot of code too.

Yeah, very exciting.

So introduce myself. I'm Eugene man, co founder and chief product officer. Maven is my third startup I try to sell the first two to Sami at GOOG and then had the wonderful opportunity to work on personalization. Google News was a personalized user experience for over a billion users around the world. So imagine 127 countries, 64 languages and when you open Google News you are not really expecting the front page of the Wall Street Journal, the New York Times or Fox News. You want to read about your own personal interests. So imagine a billion people times all of the permutations of news articles and interests. And when you open it, you want to the top three things that matter to you. Before founding Maven, I actually had the opportunity to lead applied machine learning at Stripe. And how I got there was my area of responsibility was the risk user experience. So risk at Stripe their mission is to not lose money for Stripe. Now if you are a individual or a platform that is a great customer, maybe doordash in Australia, but for some reason to the risk models looks like a credit or fraud risk. They might turn you off and you get quite upset. Actually I had a personal experience last week where amex turn off all of my credit cards because a vendor accidentally forgot a decimal point and ran a $700,000 charge instead of 7,000. All of my credit cards shut off. I call Amex like Hey guys, like this is a disaster. Like, oh, sorry. The risk model said you may be fraudulent, come back in 14 days, we might let you back on. Imagine this is your payment system at Stripe. The amount of support requests and the urgency of customers saying oh crap, something is off is pretty painful. A lot of people get upset and they email Patrick Collison saying hey, like can you please turn me back on? Patrick forwards the email to me. Eugene, what are Yahoos doing? Like, we can't be doing this to folks. And so we quickly realized that the volume and the urgency of the customers looking for support didn't match even Stripe's capability to service. That I built an early LLM powered support tool that could start working the process ended up getting access to GPT4 about six months before ChatGPT even came out. Greg Brockman was the former CTO of Stripe and gave us early access. Stripe ended up being a launch partner. And so if you all remember, the first time you used Chat GPT, at least for me, it was a holy shit moment. I was on a vacation actually for Thanksgiving in Portugal. I start using this and you can just start feeling the potential. Google of course has been doing Transformers and Large Manuf for a very long time. At News, we've had access to earlier versions of this, but this felt like a very, for the first time, something that could be practically useful. And so I reached out to Jonathan, who was the best man at my wedding. I was at his and saying, hey, Stripe is growing incredibly quickly. Its intention is to offer great customer experience, but we're clearly not there yet. HubSpot must have figured it out. Like what's the answer to do offering great support.

That's a perfect transition question.

It is a great transition question. And I started off my career writing code at a company called Sybase. And so Sammy and I like to joke we're the only people in most rooms that remember what these companies are.

Sidebase and Lotus. Yeah.

And then begin kind of the beginning of what I like to joke as being my love affair with go to market software companies. I joined a company called Unica here in Boston. They were very, very early on in terms of being of creating a, a full stack marketing software. And so they ended up getting acquired by IBM and then spent some time at wpp which is one of the largest ad conglomerates in the world, working with companies like Coca Cola on their 10 year digital strategy. And as I was working with them, I spent some time working with a variety of different vendors. One of them I worked with was called Omniture and they were a market leader in digital analytics. Really thought they had incredibly powerful technology and ended up joining them as an account manager working on their strategic accounts. Went on to lead the eastern half, the us, Latin America, Canada, started the financial service and insurance vertical and then after five years of amazing time there, I went and started a company. The company was called Veer and it was a parking app. So we understood destination tent of where people were going, what they wanted to do and so we could engage with them around that. And so we ended up exiting that and I joined Marketo, was there for about a year and then we ended up getting acquired by Best Equity Partners and then they sold that to Adobe. And I said I've been to Adobe before, I know what this is all about. I ended up joining a company called Sprinklr at about 100 million revenue. Spent a couple of years there and it was probably one of the earliest companies I'd been at besides my startup to date. And so tremendous learnings there of how do you go out, how do you build the go to market motions, how do you ensure that customers are getting value out of the product. And then after a couple of years there, I heard from this company in Boston called HubSpot and big fans of what they're doing at a number of friends who they, some of the founders had invested in. And I remember having this conversation with Brian Halligan, who's one of the founders. He was like, we're the only kind of market leader in this space you haven't worked at. You need to come join us. So I ended up joining them back in 2019 and spent four years there. And over that four year time period, we were one of the five fastest growing B2B SaaS companies ever. After a billion dollars in revenue. I mean, the amount of growth that we're going through, it's exhilarating. If you've ever been part of something that's just growing at this tremendous speeds, it's amazing. And what was amazing about it was not just like, hey, we're hiring a bunch of people, or you know, kind of all the revenues coming in. It was the impact that we're having in terms of customers. And so we went from, I've heard you share.

What was the employee and customer count when you started and by the time you left?

Yeah, it was, it was a little under 2000. It was about 7700 when I left. And my team went from 238 people when I started to about a thousand people by the time I left. You know, from a customer impact perspective, we went from one product on average per customer to three products on average per customer. And so it was tremendous to be able to see the value that we were creating for customers in terms of helping them to do a better job of interacting with prospects and ensuring that they had more insight in terms of the people that they were working with. And so for me, you know, that was one of the areas where I had very deep conversations with Kate Bucher, who's the CFO at HubSpot. And she was like, Jonathan, you need to figure out nonlinear growth. And you're like, I don't know what the heck you're talking about, Kate. Like, we have these spreadsheets and we just like, you know, for X number of customers, we add more headcount. And I think that's the way most organizations grow. And good for Kate, because she was really trying to figure out how we could create some opportunities within the business to be able to reduce the human capital that we're bringing in. And so I turned to Sami and Eugene and talk to a bunch of different companies, hey, what are you all doing differently? What are you doing differently at Stripe? What are you doing differently at Google? Sammy, how did you create personalization Such a massive scale, because at the end of the day, that's what customers want. They want to know that you understand them, that you have complete understanding of the historic interactions that you've had, the products they purchased, why they purchased it. And you want to be able to ensure that you're consistently bringing that to the people who are interacting with them. There's nothing more frustrating for a customer than going out and having a conversation with someone who didn't bother to read about, you know, what they were doing historically, why they purchased the product, the value they're getting out of it, how they're using it, who the heck they are. And so you want to be able to break those down. And when you go out and you talk to customers, the number one pet peeve that they'll talk to you about is, why do I have to reintroduce myself and what I'm doing and why I purchased a product every single time I talk to somebody? And, like, that's where the real value and information is. It's like how you're deploying it and the insights you derive from it and the actions you can take from it. And so, you know, for this particular thing, you know, it's something that I think has been fermenting in our minds for a very long time. And so when we looked at the technology that was there and we said, this is going to unlock the way that a complete paradigm shift in the way that people are interacting with technology, with large language models and transformers, right? 2017, we said, this is going to unlock a complete change in the way people use technology. You're going to go from being able to visualize things to be able to ask for outcomes. A really good example would be customers would come to HubSpot and say, Hey, I purchased your marketing products because I want to be better at marketing. I want more leads. And so you can imagine a world where you could actually, instead of saying, great, now I'm going to purchase a product and then I'm going to plug in my website and I'm going to plug in my socials and I'm going to run a whole bunch of campaigns that generate more leads where you could actually just say, give me 100 leads next month.

Yeah.

And that was the world for us that it unlocked.

And this user problem, I think, is very intuitive to all of us. We've all had frustrating experiences, maybe to pick on B2B SaaS software for a little bit. And that scenario that Jonathan just described where we all blog into a CRM, be it HubSpot, Marketo Salesforce and you're trying to create an email sequence to get more leads for next month. You know that HubSpot has the best performing subject line. It has the great emails. Why is it keeping it from me? Why is it making me learn how to do marketing, learn how to use a Salesforce or HubSpot where you can imagine that the software is actually supposed to give you outcomes. I want to grow, I want leads, I want to get organized. And so we thought about this at Stripe, where Stripe could be your cfo. It knows so much about your business, it can benchmark the others. But how do you unlock that data and guide your customer to using it? We've all traveled on pick on your favorite airline, you get rebooked, you have to call someone. Why am I talking to a human? To get this solved. And so I think we all want that concierge experience, but it was impossible three, five, ten years ago. As we're having dinners, the technology just wasn't there to enable this. And we think it's finally starting to become that where software can be outcomes based.

Okay, so. And it's clear how all three of you guys are so passionate about customer experience and, you know, it's a perfect, well balanced co founding team combo and we'll come back to that. So just to go back to the origin story in the founding to just kind of, you know, maybe you guys could tell us like when was. When did we decide to pull the trigger? And then what happened? How did we make that decision? And then how many iterations did we go through? Or did we go straight to what the product is today?

No, we kind of went straight to where the product is today. I think we probably started having conversations and it was maybe a month or two beforehand of, hey, we think this has real legs to it. And then I think we all kind of came to the same conclusion together. There's something here and we're going to go do this. And so, you know, we told our employers, hey, we're going to go do a startup. I think all of them would have liked us to stick around a little bit longer, which is amazing.

I mean, there was a funny moment. I mean, we knew there was a there there. I mean, the technology is cool, it's very capable. We knew there was a real problem and there was real money behind it and it is fundamentally disrupted. But we went through this like, I'd have dinner with Jonathan. I won't do it. If you, if you, if you're doing, not doing It. I'm not doing it. I'm like, if you're not doing it, I'm not doing it. And then I had dinner with Eugene. If you're not doing it, I'm not doing it. And then I think they had the same thing.

It's the Spider man meme.

And then we, some point, like we're sitting there, are we doing this? I'm like, okay, let's do this. And then we started to kind of think about, well, should we raise money? Should we not raise money given our network we had. I mean, we did it. I mean, we raised our seat in a week. I mean, we met 8 million in a week. Yeah, I mean, it was very. I mean, just. I mean, we're not first time founders.

So how long after the dinner when that.

We had dinner on Tuesday. We met the partners on Friday. We did the meeting on Monday. Tuesday we had a term sheet.

Really? Wow.

Actually, actually I remember a little differently because on Monday night they were like, great, where do we sign the check? And we said, oh, shit, we're. We haven't incorporated yet and we don't.

Have a bank account even in.

Yeah, yeah. Nothing. Yeah, so let's definitely drill into that. So. So you feel like the validation you had done was you were working at HubSpot. You guys were going through this pain. You guys, you were at stride. You were going through this pain. So you guys didn't go out, you didn't, you hadn't built anything, but you didn't do any. Did you do any external either validation or just like talking to friends? Just like, hey, we want to go start this company. Or it was like, we're doing it and then you talk to a local VC or something and they're like, we're just, we're just gonna take down those.

It wasn't, it wasn't a local vc. It was, you know, kind of a connection that reached out at a very fortuitous timing. And, and that we happened to have.

This was after. Right. The, you know, the, the VC party as we talked about it, when, you know, VCs were doing, you know, high on drugs and alcohol. Right. We're talking about November 2022. Right. I'm sorry, November 2023. So less than 12 months ago, or is that just.

Yeah, so we announced, I think we announced it in June of, Of this year, but we raised, we raised it. Was it probably June of last year?

Yeah.

When we, when we ended up closing it.

Yeah. Still well outside that period.

So it was. Yeah, it was just, it Was it was great timing, happened to have some connections and it was think, you know, for us. We said, we made the decision we want to move fast and we found a partner who wanted to do the.

Same thing and the partner took down the whole round.

We ended up pulling in E14, which is the MIT Media Lab arm. And Lux and E14 have been great partners for us in terms of being aggressive, moving fast and understanding our vision of what we think the world is going to look like in the future.

And so it was an $8 million seed round.

It was an $8 million seed round valuation. The valuation was 14 million post money.

So what do you think led to that? So pre product, pre revenue, pre a lot of things. You raised 8 million on a 40 million valuation. How does one do that? There's a lot of early stage AI founders listening and, and they would like that. And I know some, some of it is that your story is not typical, but what is there that that can be learned from that process?

I think when you, when you're looking at, you know, the people that you want to work with, you know, it's fortunate that we were friends and we, we kind of came from complementary backgrounds. So I think the, the team and the makeup of it really worked well. I, I know the CX space extremely well. I have experience in go to market functions. You know, if I was to build a team, I couldn't ask for better people to do it with than people who have, are able to go deep in terms of being able to not just like imagine a world that looks different in the future, but being able to actually create that world. And so I think, you know, the accomplishment, complimentary skills across the founding team is something that really contributed to that. I think the other thing is like, what's the size of the opportunity that we're going after? And so, you know, the TAM in support alone is $560 billion. And that's just on the human capital side. There's another 60 billion on the software side as well. So, you know, that's a fairly significant TAM to go after $720 billion in the support area. So I think that's, that's something that's also very attractive too. But if I was a founder and saying, hey, how do I go out there and how do I create a team that's going to be able to have an impact that people are going to take seriously? I would say how do I find complimentary skills that are going to allow me to move fast in terms of building a company?

Yeah. So the, the complementary skill set between the three founders. And it's not just that. I mean even you're a technical CEO and you have a deep domain expertise.

Sammy. Sammy's shaking his head. He's like, no, I'd never let him touch code.

He sent me his code.

You buy patents and obviously deeply technical with lots of startup experience. I'm same with even Eugene, the technical CPO with experience at different brands and with AI. So you guys have the kind of like the well known successful tech startup experience. You have small startup experience of yourself, including exits. You have the deep domain expertise and the passion with the area. It's kind of like customer experience combined with personalization coming together.

And we knew we needed each other. This is actually the other aspect of it is that Jonathan wanted to, needed us both. I knew I wanted both of them. And when, when you kind of the thing you learn over the years is like it's not, it's like the team that ultimately does that. And I knew I could build tech, but I could, I wasn't like the best salesperson. I can sell, but I can't close enterprise deals. But this guy can't. And this guy can build awesome product. And kind of the trifecta made it made us be kind of like a pretty profound group that can not only dream but also execute.

Yeah.

And the startup journey is tough. So this, I think about it like a three legged stool that if any one leg wasn't here it would collapse. And so, so many times we're texting each other. Like at least I text. I'm so glad that I'm here in the trenches with the two of you because there's so many cases where Jonathan or Sammy and one of the things founding a startup in this space is that we need to execute and grow fast. So one of the exciting parts about when we started, when that first check came, Sammy's texting and he's like, okay, how many engineers do we need? Eight, fifty? Like they're here. Same thing with Jonathan. It's like, you know, he's hiring absolutely world class people who then bring in their world class people. And so you have this chain of trust that as we're scaling maven to the team size we are today and beyond, like we're all, all thrilled to be in kind of this Viking ship together going on this great adventure. So I think maybe that's one of the reasons why we can now execute here is that the trust and the familiarity that we've had and developed each other over Time and Sammy and Jonathan with careers that they've had and to call on the people to make a lot of this happen.

I can definitely sense the chemistry with you guys. Okay, so we get the 8 million and then we're off to the races. First thing is we want to build a product, I'm guessing.

Yeah, first create a GitHub repo. So literally new repo, start clean.

I think for us, what we've tried to do is we've tried to be intentional at each stage of our journey. And so our milestones are less about funding and they're more focused on impact. And so for us, we set out and we said, great, a seed round is amazing, but what are we actually going to do with it? And I'm sure there's people come with hiring plans and all kinds of things, but for us, us, we've been very intentionally focused on customer impact. What's the impact that we want to have on customers? The outcome of hiring a great engineering team that we were talking about is that we can build a great product. What's the value of a great product? It has to solve a customer need. It has to be technology for the sake of solving and creating a solution. And so that was our focus for us when we went out.

Did you have like a clear goal of what you want us to do in order to run it to A?

And it wasn't, it wasn't around raising the A. It was what we wanted to do. The first milestone was around customer impact. We wanted to be able to prove out that we can help companies to solve their customer support inquiries using AI agents at scale. And so that was our stated goal. And you know, when we went out with our pitch deck, that was on there and I think you were asking earlier, hey, what were the, what was the idea. Ideation that we, you went through in terms of coming up with this idea and bring it to market? We didn't, like we said, this is the problem we're going after. This is something that everyone needs. It's a horizontal challenge that every single company has. We're going to go out there and we're going to solve this because we.

Were buyers of this problem. This was a problem we struggled with at our respective functions. So you see the overall value you're creating, the impact and the real pain. There is pain in this problem. I mean, sometimes we do talks, we ask the question, who has had a great customer experience? Nobody raises their hand. And if there's one person, are you really sure that was a good experience or Your bar just went super low and then they kind of nod their heads. It's just, and it's not that enterprises are looking to create like to have bad experiences. It's just the economics and the structure and the way it's set up is just, is not conducive to actually do it well. There are too many systems to learn. It's managing costs, the volumes are insane and automation has absolutely failed to solve it. And for the first time we have technology that actually has a shot at solving it.

Did you guys bifurcate, focus and spend some time on going and talking to customers while building the product or did you guys focus on let's get MVP to have something to show?

What was interesting is that, you know, kind of we said, hey, we're going to go out and we're going to like create a product that has customer impact. How do you know if it has impact? Well, customers have to tell you that, right? And so he said, great, we're going to go out there and we want to be intentional in terms of who we're working with. We want to work with people across disparate industries, multiple different verticals in order to prove our assumptions that this is a horizontal challenge. And so we worked with companies that are marketplaces, B2B SaaS, hospitality startups, publicly traded companies. Turned out they all had the same problem. And the problem that they all have is you have three different types of customers that you're working with and you're really challenged to be able to solve all of their problems because there are all different channels and different mediums that they want to interact with you around. They might do one really well. Doing all of them is really challenging. And so the three types of customers most people face is like, you have customers who want to self serve, you might call them like, you know, digital natives, right? Where it's like maybe you. Maybe. Yeah, exactly. I don't want to have to go talk to somebody, I don't want to engage with them. We've had enough bad experiences doing that that we like stay clear of it. And so you have those people, you have people who, they want access to that same capabilities. They want to be able to search for their own solution, but they want validation. Hey, before I hit this big red button, is this the right thing? Great, okay, I'll do it. And the last one I think is someone who really wants that human touch. They want to know that there's someone on the other end who's thoroughly reviewed all the information that's required in Order to be able to tell them, here's how I would handle that. They want that white glove treatment. And so you have these three different types of customers who want to be treated three different ways.

And when you say customer, you mean the end user or your customer who has a support issue right now that they're dealing with.

Yes.

And there's three Personas. Right.

That you're trying to serve. And so that's. That's challenging because you need different technology for each of them. You have different capabilities that are required in terms of the people that are interacting with them. And so you're. They're struggling. And so we said we have to be able to help people solve all of those in order to be able to introduce a solution to the challenges people are facing.

And the way we think about that from a product perspective is I think people get stuck on the features. Is really kind of embrace the fact that there is a workflow. Changing behavior is very, very hard. People don't change their behavior. So create a solution that injects into the existing workflows. And that was kind of where we started. Like would. It comes from kind of our background of building consumer grade products. Products is that you have to be very intentional. You have to re. Like behavior change is hard and you have to align with where people are and also do it in a way that makes people feel good. Which is why we ended up with a product experience that is. Starts with helping customers self serve all the way down to like how human agents solve these things behind the scenes.

So in a short amount of time, I think like what about 18 months? Less than 18 months. Months. You guys have gotten HubSpot ClickUp and few other enterprise customers, right?

Yeah.

You guys have. I think my notes are right. Over a million tickets closed saving.

Like more than that.

More. More than that.

Right now we're. We're doing millions of tickets a month now.

Millions of tickets a month now. Wow. And you know, some of the stats I saw with like 65 savings, 40% health savings things. So. And Eugene, you talked about the execution velocity. So clearly you guys have been able to execute very fast. But let's kind of talk about that. Do you feel like how did that happen? How do we build it so fast, get it working so fast technically, Sammy. And then how do we get it? You know, how do we ramp up traction and customer acquisition?

So starting with how we built it, I think one of the philosophies around building the product and the engineering team was we pulled in a pretty senior team it's not, I mean it's a team that's been there, done that, built at scale, know where the skeletons are. So we kind of walked in and with kind of a clear view from a technology perspective, what are the things we need to build in order to build an enterprise grade solution that functions at scale. So one of the first things we built, evaluation framework. If you're building AI systems, if you can't measure it, you can't make it better. One of the first things we built second and kind of progressed with that general approach is like very metrics driven in terms of how we solve problems. We work very closely with our customers to kind of evolve the user experience such that we're solving the real need and also find design and work with design partners who will actually be design partners, not look for a turnkey solution. And Eugene led the charge on trying.

To understand that you built first. Which one of the three Personas did you start with or did you build all three into the product?

So great question. I mean you should probably cover that.

Yeah, so we definitely started with the self serve user because that has instant impact. So now at TripAdvisor, for example, Maven is answering 93% of user questions. We're talking hotels and restaurant overs all over the world, asking questions in all sorts of languages. And these customers are not expert marketers. You're running a boutique hotel on the Amalfi coast, you don't have time to, you know, mess with your listings, but you really want to know that you're represented fairly on TripAdvisor and you're getting more customers. And so that was the first experience that was super helpful. The second was a co pilot for your human support agents. So you have folks today who are servicing your end customers. Some are phenomenal agents, some are early on in their agency, may not have the knowledge that they need. And so we found that Maven, at places like ClickUp is letting agents not only answer questions 30% faster so they can do more per hour, but with higher quality. And the agents themselves are happy. So at one of our fintech customers, they named a conference room after us. And at first agents were really nervous. They're like, oh, is this AI going to take our job? Is it like halfway to Skynet?

Right.

And like we're the first to go. But what they realize is actually what it's helping them do is stop doing that mundane drudgery work or that research or looking into six different systems to understand a customer profile. And they're really as individuals, as Humans being able to up level their skills and become superhuman. And so there's much more of a of job satisfaction as support agents. They're going up upside down cycle. And in terms of how we built it, I think a really important part of us being a gen native company is we didn't have to replatform. So imagine you're a legacy company who's been doing rules based workflows for a really long time and all of a sudden you're saying oh now we're, we're gen AI. We're, we're AI agents.

So one of the things that you guys mentioned in the beginning is like this idea of like hey HubSpot, you know how to get customers, I want more customers, give it to me. Are we at the point where we tune the system to go from fixing problems like problem solution troubleshooting into like strategy or like is it helping with like what's the most higher order stuff that the product is helping your customers users with today? Or is that not a space you guys have gotten yet?

That's kind of aspirationally where we want to go. And the way we think about that is that first we start with AI agents that are able to answer questions in a personalized way and take actions on behalf of users in specific functions. We started in support. In order to operate at the strategic level, you need to be able to do this cross functionally. And if you're looking at it from a customer experience, support success, marketing, sales and kind of our overall approach is start nail the functional use cases one after the other. As we start to have additional functional use cases, you can start to get into a world where you're orchestrating across functions so that you can start to solve these higher order use cases. We want to be able to get to a place where we're answering CEO level questions, not just like hey, I want to add this or that and these outcomes based agents is a world where I think like this is where we're trying to get to and we're starting to see early signal that we're able to get there. But in order to do this well, it requires we approach the word cross functionally, be able to ingest all sorts of data across all sorts of systems, start to rethink what a system of record. Because the reality is if you're trying to solve CEO level issues, the reason it's hard is because you need to think about the world of digital data across many, many silos. And bringing all those silos together enables you to start thinking At a higher order of thinking, if you think about.

The underpinnings of what's required in order to enable AI agents, whether it's in support, success, any other function across the organization, there's knowledge to be able to answer questions, there's personalization information in order to make sure that you have complete visibility to why customers have interacted with you in the past, what they purchased, what they're trying to achieve, and then the ability to be able to take actions on their behalf. And so you need to be able to pull all that information together. You need to be able to not just pull it together, but you also need to manage security for it. You need to manage access to it, you need to manage the access to who it's visible to, who you can share it to, internally, externally, all that stuff. And so there is a massive amount of effort required in order to be able to, to create a system of records capable of ingesting all that information and maintaining it.

And one thing that wasn't intuitive when we started was folks were asking us, what about the systems of record of today, the sales forces of the world? Can't that platform just slap on some LLM and you have AI agents? But it turns out the idea of the system of record is a real misnomer because there isn't one system of record. You have a CRM for your customers, you have a payment system, you have application telemetry, you have contracts. And to get anything done, like imagine you're a HubSpot user. You say, hey, I want to add five more people and send 10,000 emails next month. Well, we have to look up your application to see if you have the permission. Did you buy the contract? Are you up to date on billing? That's why there's 6 and 9 systems that support agents have to use today. So there really isn't one system of record. To be able to create true agents requires bridging the gap between all of them marrying this disparate type of information and being able to, to actually take agents. And one of the experiences I had at Stripe was when the war with Russia started, Stripe users wanted to block payments from Russia, right? But when you email support, and so how do I block payments from Russia? Go to this screen, type in.ru equals no and write like SQL or regex or something crazy like it's incredibly difficult. All users wanted was an outcome. I want to block payments from Russia, a higher level outcome. I want to be fully compliant with all regulation. And so the way that we believe that the Future of software will be will be AI agents that you can give maybe broad missions. Give me a hundred leads. Okay. I'm going to create some landing pages. I'm going to do an email campaign or help me grow this as something that is proactive. And imagine having a concierge for the software you use day to day, like an expert. If you could call Dharmesh and Brian Halligan, they would tell you what to do. But we can't all have access to Brian. If you complain really hard at stripe someone to the CEO you would get a response from really was our head of comms and he would look like a support agent. But all of his emails are phenomenal. And I always wonder like, why can't all support agents answer like this person? Well, because he was the head of comms, that was what he did. But you can imagine a mental model where you can turn that into an AI agent. That's what you train on. Empower that agent to have access to take action. And now we all have to access have a much better experience.

Yeah. And like one of the other things that I'm. That's on the horizon that I'm excited about is like moving from it being initiated by the human to right. The agent, the AI agent initiating. Right. So if it's like going back to the HubSpot, you know, if there's a way we do the SEO on the pages, the agent notices, a couple pages got posted with that proper SEO and then it either notifies the customer or ideally you just go like applies and says hey, this is what I did. You may want to take a look at it. So I think this proactive agents that understand what the company's objectives are and then can start to take actions autonomously and or with the human in the loop is part of what I'm excited about.

And the reality around that is that if the technology existed today, if the technology existed today, it still would have difficulty adopting because we're in kind of this business transformation people are still trying to get like, is this Skynet? I just saw ChatGPT and it feels like it's alive, but it's not really alive. But people don't know that. It's just a token prediction engine. And so you have to kind of embrace the fact that as you're building these AI agents and deploying them, we are dealing with kind of business transformation at its core and many things on how people think about these capabilities so that you can get to these really interesting use cases. We can do these things today. But are people ready for it? I think it's. It's shifting and it's changing, but I don't think they were ready for it last year when we started, but now it's evolving.

I think what's interesting is I remember going out and having conversations with people that, you know, we wanted to get feedback from as design partners. And, you know, we would do a little demo for them, and some of them were personalized based on information that we pulled, you know, off of the web. And people would panic, oh, my God, where's this coming from? Did you steal our information? Are you training a large language on? And you're like, no, no, no, calm down, it's okay.

It's just we didn't hack into your system.

You contrast that to today where people are feeling much more comfortable with, like, yeah, I'm okay with putting an AI agent in front of my customers. And I think, like, the trust and safety aspect of it are what have to be fulfilled in order for people to be able to be comfortable with that. And we've done a really good job of making sure that that's embedded in our product. And so we're able to go out there, share information about some of the folks that are using it and getting great results, but at the same time ensuring that they feel secure, their data is secure. We're not training a model on their information. I think that's really important, especially from an enterprise use case.

So let's continue on that. Grab Jonathan. So Sami and team built a product, right? And we talked about, like, we started with the automated or the self serve first. And from a customer acquisition standpoint, how did that go? You know, walk us through from customer one at a high level to today where we're at. And has that evolved over time? Has it gotten easier or harder?

You know, for us, in the early days, what we wanted was feedback. That was the most valuable thing that we could have. We weren't looking for customers customers. We wanted for people who were also thinking about generative AI and the impact it could have on their customer experience. Just so happened that everyone we talked to said, what are you guys doing? We talked to them, here's what we think the future is going to look like. And they said, we want it, we want it. We're signing up for that. And so we were very fortunate in terms of the customer demand for the products that we're building and how we're thinking about it. And to Sammy's point earlier, we were consumers of the product that we were building. We live the challenges that these customers that we were talking to had firsthand. And so for our ability to be able to understand the landscape, the challenge they were having, the types of customers they were trying to interact with, and to build a product that solved that problem reflected very well in terms of market demand for it.

And one of the advantages that I think Maven has is that folks like Jonathan and Leila and AJ and our team were leaders in the support success, post sales space. And so we can, can really empathize with both their desires for a future and their concerns. I think Sammy says a lot. It's not a technology problem, it's a people problem. Because this really, this new technology requires organizational change and a rethinking of what humans do. Like, okay, fast forward a year. All these AI agents exist, can do all these things. What does it mean for my org? What does this mean for my leadership position in Org? I'm a VP of support, but how am I a VP of a bunch of AI agents now? And I think what we believe is that it's going to be a transformation that humans are going to do. It's not that the humans are going to go away, it's that the humans are going to do a different job. It's humans are going to be empowered to no longer do this like input, output, you know, mundane, painful support job, but something much more level up, leveled and helpful. And we think that the role of support and success in organization is going to evolve from today it's somewhat of a cost center. People want to minimize the cost of support, but in the future it's very much a foundation for your organization, what's possible because you think about if you're, you can do support well, you could also do customer success well, that's just outbound support. You can do sales because you now understand the product deeply. And so for us, it allows us to create these functional A agents and then effectively orchestrate and merge them together. So from a customer perspective, it's just one AA agent you're talking about to Stacy. And if you need something, Stacy helps and she's incredibly thoughtful, knows you personally, knows the product deeply. You no longer have to deal with marketing and then onboarding and then support all these different functions with an organization who seemingly don't talk to each other. So we think that the future world for us all as end customers will be very customer centric from the point of view of support. And these, these functional leaders of today, they're going to be doing a different, different job. But A very interesting and important job in the organization.

Somewhere along the way between when you started and got the seat and now there was a series A. So I get the sense that it's going to be a lot like the seed story, but maybe you could tell us about that, how that came together and what we can learn from that.

You know, I think we were very fortunate in that regards, where we had an opportunity to be introduced to some great investors who were interested in what we were doing. Their view of the future matched ours. And so they came to us incredibly well informed about the space. They'd done a ton of research. They understood what the challenges were the customers were facing, how people were looking at what they were people looking for in a solution, and then what our existing customers, who they were and how they were using it. And so I was very impressed by them. And I think we all were like, they're. They're going deep in the space, understanding why Maven had an opportunity to win in this space. And so they came to us and said, we want to be part of this journey with you. And so.

And how far along were you guys at that point in terms of number of customers, like, number of tickets, like, roughly, like when they contacted you? And when that started, I think we.

Probably had five or six different customers. And there were all, like, pretty large. They were using our products, they were getting value out of it. And to that point, we were still in stealth mode. We. We weren't talking publicly about what we were doing. I don't think there is very.

We haven't launched the company at all. We didn't really have a website.

Yeah, so they. They were very good at finding us and then engaging with us.

Yeah, we. We made it on a market map and that. They created some. Yeah, we made it on the Sequoia market map. That's how we got on the race radar. There was a. I mean, there was outreach, and we were very impressed by their depth.

And I think, you know, it was really interesting is they. They actually came here and we got to spend a lot of time with them and said, we think there's potential for them to help us accelerate our journey. And so that's what led to the a. So maybe a bit of an abnormal story in terms of kind of going through two rounds of funding in less than a year, but for us, it was all about choosing the right partners to help us accelerate our journey. Journey.

Yeah. One of the things that's interesting, right. And kind of goes back to that last point, which was. And we Talked about it earlier which is this the horizontal versus vertical kind of debate. It's probably been around before SaaS in different, you know, business categories. Certainly with SaaS we saw that with different horizontal plays like Salesforce and HubSpot and then the vertical players coming in and taking market share from some of those companies. Companies as a, you know, the best CRM for financial advisors, the best CRM for insurance agents. Right. But yet the, the verticals have still grown. Right. And now we see, we saw it with AI application layer horizontal versus vertical plays and now we're seeing it with the Gentic. So yeah, I'm curious to get like what are your thoughts on that? Obviously you guys started horizontal by default but I like what you said that you intentionally went the validation they cross customers in different verticals. So yeah, what are your thoughts generally? And then specifically for you guys, do you think that you'll be able the customer support. We're an early stage AI agentic, AI VC and I would say the customer support use case is the most pitch decks that we get. Right. Everybody's going after E commerce customer Support to the AI agent B2B SaaS if we understand your retention insurance success function and Right. These are entirely different businesses and so you know, they're tallying why their vertical solutions are better than the horizontals. Say what are you guys thoughts on that?

I'd be curious if we all are aligned on this but I'm happy to, happy to go first for us, we're students of history. We look at the great software companies that have been built. Why were they successful? They solved the problem that everyone had. You look at Salesforce and HubSpot and Adobe to name a few that you know, we're all familiar, familiar with. And I said we're solving a problem that everyone has. Now there's certain instances where, you know, maybe if you're a financial advisor, you want certain bells and whistles that not everyone else does. But I think that's where building a great partner ecosystem comes into play where you can customize it based on what people are looking for. And so our job as a horizontal player is to be able to create 80% of the features and functionality and the things that are required in order for us to be able to fulfill the requirements requirements that people are looking for. And I think that's where, you know, building out a partner ecosystem, having great partners you're working with comes into play of being able to take it the rest of the way. That's, that's how I view the World.

From the product perspective, customers come to us and say, I'm special. I'm sure you can do L1 tickets, but mine are highly technical. There's no way Maven can solve it. Or they'll say, well we're in fintech, we demand the highest level of security. Security. There's no way that you're secure enough to be able to see my bank balances take actions on my millions of users bank accounts or my customers are unusual. They speak in the vernacular, they have their own terminology. There's no way AI could automate this. But we've tried, you know, vertical after vertical, segment after segment. And we found actually the more complicated and thorny the issue, the better maven and probably LLMs are at solving these problems, let alone humans. Because the thing that we're wanting to improve on is human performance. And so imagine large organizations with thousands of support agents. You get a curve, a distribution curve of performance where great. The tenured ones, ones that have been around for 12 months are fantastic. On month 13 they quit because it's basically a sweatshop. It's a really tough job. So you get these new agents that have varied performance where with something like Maven, you train it once to be able to serve all functions. And so we've done done a series of verticals from marketplaces to SaaS to FinTech performs great. We've done L1 support tickets, effectively the first stage in a support journey. We've done effectively professional services like the most advanced form of support, technical, non technical, Maven's writing code. So in our experience it's a horizontal problem and the solution works for everyone that we've tried so far.

So a little more color on the solution. One of the things when you kind of think about product and engineering is you take complex ideas and you deconstruct them into building blocks and building blocks that you can then use a data driven model to be able to evaluate and whether it's like working or not. And as we were thinking about this as a one of the things when we were talking about it, it was actually important for us to get diversity of use cases so that you have a clearer understanding of what are the building blocks necessary to solve diversity of use cases. And this is actually where some people go like pick one use case and go super deep. You actually end up always building better software if you have a little bit of diversity. You build it once, you build it twice. By the third time you know what you're doing and how to generalize it. And as A result, it's actually had profound implications on how we've engineered the product. What are the constructs that we expose from a user experience? What are the requirements from a platform capabilities? And it kind of gave us a deeper understanding of what we need to build, how we needed to expose it and really kind of identified what are the appropriate building blocks to create a tremendous level of customizability while being understandable, while solving the most diverse set of use cases.

And I think the first question is vertical versus horizontal. The second is product or platform. What we've discovered that it's not just a single use case product for support, but as you start doing support first with inside the organization, you get pull. Folks are asking well okay, you did support but like could you help automate this success function? It's exactly like support, but just go outbound or sales or another. And so what we realized, realized what was needed was not just this, this one product, but a platform to build AI agents. Because the support agent and the sales agent both ingest salesforce data, both have a personality, both talk to a chatbot. And as a customer, like let's say a TripAdvisor, you're not going to put five chatbots on your site. You need one chatbot that understands when a pre sale customer shows up, then when the same customers now post sale asks a question, you need to serve that as well. Well and so we've built this platform that lets you and you, any non technical person without a, you know, MIT PhD to be able to create and test and validate agents. And so the platform capabilities, things like evaluation where you can know that it's working well or connectors with 50 plus integrations are ones that all companies are going to need. So maybe the answer isn't even just horizontal vertical, but should it be a product or a platform platform?

I, I like that you added that dimensionality to it. The product versus platform. Yeah, you know it's interesting at SF Tech Week, you know we, I led a couple panels, a VC panel and a founder panel on agentic and vertical AI and we talked about this horizontal versus vertical. And then the next week at LA Tech Week we did the same. And you know, I think what I've learned, my takeaway after diving deep and getting a lot of perspective is both conversations win. Right? Both have their playbooks. It's a matter of executing the playbook properly. Some deep domain experts start off for like an AI agents for insurance, but now it's like okay, well what's similar, let's expand from There tell a VC like me the story of ultimately how does this scale? And then on the other side is kind of what Jonathan said, right? Like with solopong, everyone has let's build the Y bomb, do it better than everyone else, let's be there before everyone else, import the bloody red ocean and try to become the next Salesforce or HubSpot. So. So on that note, so we're in the middle of a big product launch as we speak. So Eugene, you want to share a little bit about what are you guys working on? What's, what's new and exciting?

Yeah. Sammy, do you want to tell us about the platform?

Yes. So we, we are. I mean early on it was very clear that in order to get to this world where we're solving outcomes based agents, we had to build a series of things. First, we needed to build core agent technology that can solve answer questions, personalize the responses and execute actions. But as we were developing that it became very clear and we've kind of lived through it is that to build world class agents that solve this horizontal problem, we need to think about integrations as being a first class citizen when building agentic experiences, especially in the enterprise. So what is something we've been working on since the beginning is how do you. We take this core agents and turn it into a very easy to use environment where we can integrate with anything out there such that agents are able to access all the data necessary to be able to answer in all its mess and all its glory and all its diversity to be able to tap into our customers customers user data such that these experiences experiences can become highly personalized and unique. So everybody gets a personal agent that understands them. And then the third part is to be able to take where agents can actually start to execute actions on behalf of users. And we kind of evolved our product architecture such that we have a platform where you can build these integrations in a very fast and easy to use way. And we actually trained Maven on our API so you can actually create these integrations using agents, which is kind of cool and really kind of dug into the security and compliance requirements necessary for these integrations to operate in highly confidential and highly secure environments in the enterprise. So very excited this platform's coming out now. We already have 50 integrations and if I were to kind of simplify, like the, the visual I want to leave everyone with is if you remember the movie the Matrix, Neo wakes up and he says I know Kung Fu. Our platform will have the capabilities. As you install a Maven app, I know Snowflake, I know HubSpot, I know Zendesk, I know Salesforce by simply just installing the app and it just becomes a new capability that's out of the door. And it's also designed such that these things can be done and built in a highly personalized way.

I think the other thing that, you know, we've all realized is that as much as, you know, tech is a way for us to interact and engage, it's the world is not just text based. You need to be multimodal and that's the future of customer experience. You need to be able to have voice, you need to be able to use pictures and video, video, not just texts. And they all need to be kind of wrapped up together. You need to be able to engage with someone on a cell phone, be able to hand off to your laptop and be able to carry on the conversation through email. That's what's required in order to maintain continuity. And so it's exciting for us as we're launching this platform to be able to share more and be able to unlock it's exciting for us as we're launching this platform to be able to share more and be able to unlock additional surfaces for folks to be able.

To engage and imagine you are a functional leader inside an organization. You've had tremendous success with maven, you're just sharing it with all your others saying like, hey, this thing is awesome. Next quarter, every functional leader is going to want their own AI solution. So now you're the CFO or the cio. Do you share your data and integrations with five or ten partners and you know, go through that same security review? Or is it better to, to work with one organization who has a platform that can serve these multiple use cases? See, at the end of the day, there's just one set of enterprise data and one set of actions that needs to take and not so many surfaces that you can have five different tools. So we believe this platform needs to exist. One, to support the integrations, two, to be multimodal and three, to serve the variety of use cases that you're going to want to have in the enterprise to have AI agents do work across function.

So as we're nearing the end here, you know, I'm confident I'm sitting in the offices of, you know, going to become one of the most successful agentic AI company so far. You know, there's elements of it that, you know, I think you said fortunate a number of times, right? You're fortunate. You got inbound, contacted by The CPC and the Series A. And fortunate to have some great design partners was everything fortunate. Has there been anything that has been a little unfortunate? Like have you guys just been like this totally exception startup journey with like not a single thing has gone wrong.

The way I think about it is like imagine we're trying to take a rocket ship to Mars. You have to wait for the orbits to align. So we've been having dinners every month for 10 years talking about the problem, you know, was as soon as the technology came out, Sammy I were talking to each other about LLMs. And so firstly we all discovered these are need. So that's one planet aligning.

The technology there is for founders to be able to find people that they like and trust and then start building those relationships and having those, those dinners and hangouts and fears way in advance of when the time comes and exploring.

The problem, talking to customers so that when all the pieces unlock, maybe it's kind of like a key going into a lock that all the pieces start unlocking, locking. We're fortunate because at one point the key worked and everything unlocked. But Maybe it was 10 years of us bullshitting about this problem over whiskeys until finally everything came together.

I think the other thing is like look like nothing's perfect, right? And so we've said we're fortunate because we know how hard this journey is. Like we've done this before and it's not all kind of ponies and rainbows, right? There's some hard problems we've had to work through. And it means like we're working long nights. We're like the technology, right? You know, we first started working on this problem. Nothing was off the shelf. Like Sammy Gene had to go out there with the engineering team.

I mean it's like blocking and tackling and getting capacity. So we're talking about the volumes of like enterprise scale volumes. We're doing Sammy's texting with the VP of like OpenAI VP of Azure saying like we need capacity right now. And they're like, oh great. Well so some of it, some is available in Canada three months from now.

And on the sales side, right, like we're trying to sell technology that didn't even exist a year ago. And so you go in and you're like, hey guys, we're going to sell you a magic thing that does whatever you want it to do. And they're like, hold on a second here. We've heard about the hallucinations, we've heard about, you know, the Google telling people to go eat rocks and like free airline tickets for people who are, you know, made up.

Bereavement policy.

Exactly, exactly. And so, you know, like, you go out there and you have really difficult conversations with security and legal and compliance people. You know, for a lot of the customers that we, that we signed up in our early days and even today we're going out there and we're having conversations with C level executives or meeting with the executive team, we're going out and having workshops with them. Let us help you to understand better how generative AI can make an impact in your business.

This is probably, I think the hardest part was realizing that there is a tremendous amount of education to deal with the fear we are doing. We're not. It's not just, hey, here's some maven and try to use maven. It's like, here's AI101, here's large language model 101. Here's how to potentially think about business transformation. This isn't just about workforce force reduction. It is a new function. So, so there is an element of human education to kind of make your buyers more educated about how to even buy the stuff. The questions we were getting at the beginning, I mean the stuff was just like, did not, like, did not match.

Yeah. And some of the people were maybe true believers in the technology, others were skeptics. And so we would share with people, oh, and you know, some of the folks who work with are getting these results and we'd have VP support. Locust guys say, you two are effing liars. And I've never had like been accused of lying because the, the results seem so fantastical that it's like.

And I mean like one, one thing that has come up is that as disruptive as this technology is, it has also reduced the barrier for any developer to have access to world class reasoning engines. So OpenAI's APIs is like available. Anybody can call it. So then the next logical thing that emerges is that Company X decides, oh, it's just an A. Oh, we can just call, we can just do this ourselves through ChatGPT. And like, here's the API. And what ends up happening is they go down this build thinking that it's easy because it's just an API way they most almost actually always three months later they come back like, holy moly, this is hard.

Right? And then I wonder if you guys are seeing some of this. When I was moderating the, the SF Silicon Valley DC panel, they started sharing a little bit about the other end of what they're seeing. So like we talked about, there's fear of skepticism, there's some early adoption and some overexcitement when you first started, right? And now what like Khosla and others were sharing is AI burnout. And there's too many options. There's too many options in support or, or, you know, it just. There's been some failed AI projects and enterprises are starting to take a little bit of a pause and also just getting indecision over too many options there. That was the first night air to that. But Kokla and some of the other VCs are seeing that in their portfolio, which I thought was interesting.

I think what's interesting is, you know, a year ago if you went out and you talked to a VC and you said, I'm doing a large language model model powered X, they're like, oh, you're just a wrapper on top of ChatGPT. And so you've seen a very similar thing cascade down to buyers where they're like, hey, I was online and I was asking chat CBT questions and by the way, there's a line of people out the door who are saying the exact same thing. They all have four wheels and they look like cars. What's the difference? And I think you're having the kind of the exact same thing. And I think, you know, of course VCs are ahead of the curve in terms of being able to educate themselves and understand like, what's the difference between each of them? And so, you know, buyers are still getting to the point where due to the proliferation that you're seeing happening across the vendor landscape, they're still sorting that out. And to Sammy's point earlier, the education is first, great, what the heck are large language models and what do you do with them? Secondly, like, how do I go about purchasing a product that is gen AI native? And so I think that second aspect of that is still people catching up with that.

I mean, in some cases, Jonathan's written a recommended RFP process to help people figure out how to even buy this stuff. And I have to say, like, some of the customers we've worked with that have actually went and did a build and then came back to buy ended up being much better buyers because they just got smart about the stuff and they understood what worked. And so this is, I think the evolution is as more people build and fail, you will end up with a better buying ecosystem because people get it.

I think the other thing that you're seeing is that you have a bunch of gen companies that are going out there and they really build a product. They're going out there and saying, hey, you know, whatever you want, we'll build it and we're going to have an engineer sitting on site and they're like fingers on the keyboard coding whatever you want. You're like, come on guys, this is actually doing bad things for the industry and for other companies that are actually building products in this space.

Yeah.

And support automation isn't new. A lot of the folks that we're talking now have been burned by last generation machine learning learning solutions that are not LLM native. And so maybe rules based systems that sound great at first but you know, a couple years later and 10,000 rules in this crazy chart, they become brittle and they stop working. And so this area is clearly big and clearly something that's very important for all of us as consumers. So there. But there is a little bit of these like waves in. Yeah.

So what would be each of your advice to early stage founders that are building a jpeg? Whether it's, you know, AI agents that support humans or like autonomous agents that can fund processes. But in, primarily in that space, the Jentic space, maybe we could broaden that a little bit. Just have vertical and horizontal AI, although most of them are essentially agents. But yeah, what's the. Maybe we can kind of go to each of you, starting with you Jonathan, like what would be your advice?

My number one thing would be solve a big problem. Like I see a ton of people going out there and they're solving very niche problems for a very small segment of the market. I'm sure there's an opportunity there to build a lifestyle business, but probably not one that you'd want to fund as a vc. And so I think for all of us we said, hey, if we're going to leave the amazing companies that we are working at and we're going to go out there and try to do something, we want to be impactful. We want to be able to empower customers to be happier. Not just for five companies, but for an entire cadre of companies. So that's my advice and I think it's interesting is like, you know, for a lot of companies you see TAM expansion over time. You know, if we were to look at what we thought the TAM was for Gen AI last year versus the this year, it's changed significantly.

I'll probably build on that in a slightly different way. One of the things I felt I learned at Google is that you can actually dream really, really big. This is probably one of the most disruptive technologies that I think we have seen. I would argue it's probably bigger than the Internet, even though many aspects of it feel like the disruptions that emerge with the Internet is that yes, you might be solving a problem that's right in front of you, but there's a real opportunity here to change how everything operates. And I don't be shy to have a vision that is world changing. Even though you might start with a very simple, very targeted use case, that's a real problem. But the vision really needs be to, to be world changing for it to really have the material impact or be even interesting at the venture level.

That's great. And then would another piece of advice be. I think from what I gathered, correct me if I'm wrong. Sounded like you didn't do the typical co founder and CTO thing where as soon as you guys got the money, you start building. Even if without the money you start hacking things out. It sounds like you went right to recruiting like a world class team. Do you think that assuming the startups raised like a little three seater seed round, like do you think that. I guess what's your thoughts on that? Is that, was that strategic and is that what you would recommend that will build V1, that the CEO will go sell to the first 5% customers?

In my particular case, I had a very strong network and I, I had, it was the right, I mean so in this particular piece of technology it has unnaturally gotten a large swath of engineers to be very excited to be builders again. And it was just very opportunistic. While also like there was just like a lot of great people I have worked with who are kind of like frustrated where they were. It was just perfect timing to like, let's bring the band back together and let's, let's, let's do something crazy. We raised a large enough round that we could afford to do that. I think this, that particular question is really a function of can you afford it or not.

I think your ability to recruit is also a byproduct of the size of the opportunity you're going after and how interesting the problem is.

Yeah.

So if you're focusing on something that's very niche, it's a small call them like probably not something people are going to get excited about. And so if you want to go out there and recruit a world class team, you have to be solving a world class problem.

Yeah. And then so that, so, so it was kind of an opportunistic time. The other thing is that I would argue if you're trying to build a world changing company. Building the recruiting muscle early is a critical checkbox. If you're an uncomfortable doing it, you better to do it quickly and kind of get the reps in because these, this particular industry is moving faster than anything I've ever seen before. So the ability to recruit quickly is kind of critical because you need to be able to scale fairly quickly. I mean we're. What is it? How old are we now? 13, 14 months old and we're almost 40 people. And if we were doing that like with three of us and I'm coding behind the scenes, we wouldn't be where we are today.

And it goes back to like, we founded the company with intentionality. We wanted to go big and we wanted to go fast. In order for us to be able to do that, there are certain things that we're going to have to like, move faster than you typically would. When you look at, you know, kind of historically great companies. In Boston, HubSpot took them a couple of years to get to a million in ARR. Klaviyo took them three years in order to get to a million. AN ARROW ARR Uipath, not Boston based, but like it took them 10 years to get to a million ARR. We've exceeded all of those numbers. And the reason why is that we've been able to be intentional about what we wanted to build, who we were targeting and how we're going to do that. And so, you know, for us, if Sami was coding this thing for, you know, the next three years, like we would be restricted by our ability to be able to move fast. We can't build 50 integrations. One person coding, you know, for, you know, Sam mean, he wrote a whole bunch of the early code. But in order for us to continue, he needed to be able to bring in people to help him with that. And the same with like founder led sales. Of course, you know, I'm out there on the front lines. We're having conversations with Sami and Eugene, talking to the people on the technical side and the business buyers. We were the first people in the door and our early customers were all sold by us. But once you get signal from the market, hey, we have product market fit and we're ready to, to go. Great. We have to move faster here.

That's great. I love that. Thanks Sean. Eugene, your tips for the founders out there.

So my advice for the founders out there as technologists, it's really easy to get enamored by this magical technology. Like maybe this LLM thing is really alien Technology from the future, like it very well could be and it's I think magical. But we need to develop empathy for real users problems and not just our customers, but their customers and how they're thinking about to make sure you deeply understand your customer, their workflows and how this new thing fits in and not just the end state of where you're trying to reach, but what are the steps to get there. We didn't start with selling AI agents that are going to take over and like have your support team go away. We started with a chat experience. That chat existed as a surface and we started with Copilot for agents as the entry point product into support, support as the entry flow into the enterprise because we understood how the buyer would think about this and had true empathy for their problems. And so for the founders that are touched or kind of inspired by this technology, often also get inspired by the user problems that you're trying to solve with this technology. Don't just show up with like, oh, this is a cute toy. This is a cool tech. I don't know what, what would you want to do with it, but bring the, the full solution.

That makes a lot of sense. So last question here is what haven't I asked that I should have asked you guys? What else is on your mind that you feel like you want to get out? That would be interesting for the viewers to figure.

One of the questions I was asked the other day and I was in a customer recall and they said, oh, your name is maybe AGI. Like AGI isn't that presumptive? Hell yeah, it's presumptive. Like what is the thing that we're aspiring to build? What is the thing that we want to create? We want to be able to build that outcomes based agent that's capable of doing incredibly complex work at scale. So I think that's the only thing that I would add to it of like business AGI. Business AGI, right. Like Eugene was talking about it earlier of like imagine a world where you have an AI agent that's a CFO for you that can give you the reports that are required in order to run your business. And I think the same thing is.

Going to be true Artificial business.

I like it. This is an innovative show right here. I love it.

Okay, cool. Yeah, I was wondering about that too. But that makes sense. Owning the business, that's an even bigger cam than, you know, what you're looking at with the customer first experience, one that we talked about. Jonathan, Sami, Eugene, thank you so much for sharing your journey and insights. I know for all of our listeners that are AI founders that are building in the AI space, aspiring or early stage, there was a lot here and this is a great example of a great example, right? And I think there's a lot to be learned, a lot to try to emulate here to make the story, the journey, journey easier. Going after a huge TAM with some amazing team members, well balanced co founders. If you enjoyed this episode, don't forget to subscribe. If you're an agentic founder, reach out to us. We're aggressively investing in the space.