The State of AI + Software: Where It’s Going - Fast

So think about that for a minute. I mean, in the old days at, at, at $50 million, you'd probably have at least 100 sales reps in the old days because at $50 million you want to go to $100 million. So that's net $50 million of bookings, $500K net per rep with scaling and turnover and churn. That's pretty good. So you would need 100 sales reps at $50 million. He's gonna have 5 and a team of AIs. It's not that he doesn't need sales reps. It's not even that he's not selling. He's actually a classic SaaS guy turned AI guy. He's just doing it not only with less humans, but for purpose of this, he's doing it so much more efficiently and he has 10,000+ inbound leads a month. Hey everybody, it's Saster. Connect data, automate busywork, and empower teams like nobody's business with the one platform that grows with you every step of the way. Learn how Salesforce works for startups at salesforce.com/smb. That's salesforce.com/smb. Hey everybody, it's Saster. Fin is the number one AI agent for resolving complex queries like refunds, transaction disputes, and technical troubleshooting, all with speed and reliability. See how Fin can deliver the highest resolution rates and highest quality customer experience at fin.ai/saster. That's fin.ai/saster. Hey everybody, get excited. SaaStr AI London is this December, December 1st and 2nd, and we're on track to completely sell out with the playbooks for AI and B2B. Join 2,000+ BI and AI leaders for 2 days, 2 days of practical advice on scaling in the new year. We'll have speakers flying in from OpenAI, Wiz, Clay, Intercom, Fin, and all your favorite B2B companies, including yours truly with Harry Stebbings and more doing our live 20VC podcast. It'll be fun, all in the heart of London. Don't miss out. Get get your tickets and you can still go, go to podcast.sasterlondon.com. That's podcast.sasterlondon.com for a special discount just for you. Welcome everybody. Thanks for coming. We have a full day lined up. I think this is the second AI Day we've done this year. Amelia, does that sound right? Correct. Yep. Yeah, we will try to, well, I think we will do Essentially, we'll try to do this every quarter one way or another. In May, it'll be live at SaaStrangle. So we probably won't do a digital one, although that will all be streamed with 10,000 folks. But there is just so much going on in AI and go-to-market in particular that that's what we're going to focus on at SaaStrangle. I'm going to talk about, do a deep dive on everything that I'm seeing in AI and software and go-to-market. We're going to do an even better version of it live in London on December 1st and 2nd. If you want to come and learn how to do AI and GTM together, London, it's cheap, buy your ticket now. But this is a theme we're going to do for the next 12 months. But the space is changing so much. I mean, just on our little team, you know, at the end of Q1, we had no AI agents in production. We had nothing. Then we added a general agent for support. And now we've got almost 20 AI agents. We've got 4 different AI SDRs running. We've just rolled out Salesforce AgentForce. We'll talk about that. It just started yesterday. So we need a little bit of time. We'll share all the data. We've got an AI BDR from Qualify. The team from Qualify is going to hear how that really works later today. We've got a slew of AI agents and it is so much different even on our little team than 90 days ago. And it's going to keep changing. And I think we, even little team SaaS was probably a little behind the curve at the start of the year. Now we're kind of at the bleeding edge. And so we want to drag everybody along with us because everything's changing. The models are changing. The tools are changing. Things that didn't work last year can work really well now. I mean, everyone complained about how crummy AI SDRs were last year and there's There still are a lot of issues, but now we know how to train them. Now we know how to iterate with them. Now we know how to make it work. There's so much more coming. And marketing in some ways is even further behind sales, but it won't be. And this whole space is going to radically change in the next 12 months. And to be honest, I think to succeed for many folks in GTM in general, you're going to have to become an expert. It doesn't mean you have to be a vibe coder like I've become. It doesn't mean you even need to play with things. What it really means, if you really want to, if you feel behind in AI and go-to-market, if you feel behind in anything, the real answer is be a part of a deployment. Go buy any tool, go buy any tool that, that comes and talks today. Go buy any leading tool in the space. It's some, at some level it doesn't matter, but don't just buy it. Be part of the deployment, train it yourself, be part of the onboarding, be part of the errors and the issues and the daily iterations that you have to do to get it going. Don't just, don't just set and forget and buy. You will learn nothing. You will learn absolutely nothing. Be part of a deployment to see how it really works. Otherwise you'll never really learn. So my meta advice, and we'll dig into that all day today and then live in London in December. But what I wanted to do today was Iconiq Growth, which is one of the top, the leading late-stage B2B growth funds forever from the early days when it managed Mark Zuckerberg's family money. To being in so many leaders from Canva and Anthropic and on down. They put out this great report. I wrote it up on SaaStr and I, and I linked it to the right. It's almost 100 pages and there's a lot of detail here and stuff a lot of you don't really care about, about public companies and nuances. But I picked out, I thought, the 10 most relevant things to SaaStr folks for B2B that show just how much the world is changing. I think a lot of folks are behind here. A lot of folks, and it's okay. It's not necessarily fatal. We're not all being killed by ChatGPT and Anthropic, but the world is changing. I thought this was a great set of data. This is all data to show you just how it's changing and how we all have to adapt. And this first one, it kind of explains a lot of things that are just really confusing on social media. We're like, oh, these, these AI-native companies, they're burning all these tokens. Their gross margins are low. Their revenue's not sustainable, it's not real ARR, and all that stuff's true. And we, we'll have a little bit of time for Q&A at the end. We can talk about that. Here's the non-obvious thing. Yes, a lot of AI companies, but not all, a lot of them have huge token costs. Certainly like Replit that I use a lot and lovable. Are at the extreme end, they use a lot of AI. There's other folks that you would think use a lot of AI and don't. There's folks like Higgsfield and Opus Pro and NVIDIA that are pretty cash efficient. It's all over the place. But no matter what, whether the COGS are high or not, the thing is these AI companies grow so fast that their burn multiple is much better. It is much better. And, or at least not worse. So if you look here that you show that what's happening is yes, they have higher costs in many cases, but they're growing so quickly they're growing so quickly that that is outpacing the costs of AI. So it doesn't necessarily mean that they are less efficient. And this is one of the many reasons VC money is all in AI. Like there's just no interest in traditional slower growing software because yes, the costs are higher, but the cost for that incremental dollar of ARR is often lower, right? And you can see at $100 million ARR, The burn multiple falls to 0.4x for AI-native companies versus 1.6 for classic SaaS companies. They're 4 times more efficient, 4 times more efficient in adding ARR. Will all that ARR stick? You know, I think if you, if we look back at our, over the last 2 years, a lot of the pre-ChatGPT stuff didn't stick, right? A lot of the early apps that were, that let you, but before, before, before ChatGPT was really strong, there were sales apps that just used the open AI API and could kind of help SDRs write emails. A lot of that stuff died, but I'm not— I'm seeing NRR is not what it used to be. It has a looser definition, but most of the high-growing AI companies I'm involved with have triple-digit NRR. They have triple-digit NRR. And even when they don't, I think churning through some of your low-value customers, but keeping your high-value customers will mean a lot of this revenue is pretty enduring. I'll give you a personal example. If you follow me, you know, we've launched 8 applications on Replit, even though our first failed somewhat spectacularly on social media. We've launched 8, and there's a lot of, there's a lot of posts recently showing how a lot of the vibe coding apps are seeing some plateau in growth, not revenue growth, but in searches like, and maybe, but maybe that's good because folks like us, we will never churn from Replit. I have 8 apps in production. I'm not taking them down. Our AI valuation calculator has been used almost half a million times. Okay. We're not, we're not going to take that down. Now, other folks, Amelia tried to do something different on Lovable in the early days, a landing page. It didn't work, right? We ultimately moved it to Replit. Both are great. We moved it. That account probably churned, right? So there is a certain try-churn level, but once you get through those triers, the core audience is often triple-digit net revenue retention. So the thing is, yes, token costs are an issue. Yes, AI isn't cheap. Yes, in some ways, the only person making real money on AI is NVIDIA. But they grow just so fast relative to the amount of capital that this is where, this is why all the money's going there. Okay. And related to this, this is related to the prior point. This is super important. Magic number. When, when do AI companies get profitable on sales and marketing expense? And this is one of the reasons they're more efficient than you think. And you probably intuitively know this, but this helps in the numbers. The reality is the best AI native B2B companies just have insane demand. Okay. They have insane demand and it's because they do something that's not like a little bit better, like a little bit better way to look at your leads and contacts and opportunities, or a slightly better way to do analytics. The reason they have insane demand is they can do things you could never do before. Like look at Sora that just came out. It's a number one iOS app. You can now do stuff on video you literally couldn't do a year ago. You can't do it to air. Let's do a more practical B2B example. Like we talk a lot about Gamma. We use Gamma to create all our sales collateral. A lot of the use of Gamma is B2C, but Gamma makes slides with AI and you can just tell it what slides you want and it can pull data from all different data sources. And we use it so that every SaaS sponsor gets a custom deck made in minutes. It's super cool. But Gamma, but we found Gamma's ROI is so high to us. We found it and we paid for it. They didn't have, they don't need a legion of of endless marketing spend and endless sales spend. We found them and the ROI is so high that they don't have to spend as much. So if you look at this chart, what's happening is remember, you know, a magic number better than 1 means you're making your money back on sales and marketing in less than a year. And look at the ones that are rocketing at $100 million in ARR, or rather above 1, they're at 1.6, they're going profitable in 6 months. Versus look at what happens traditionally at scale in SaaS, 0.5x. That means traditionally at scale in SaaS, and this is true across almost all public SaaS and B2B companies, takes you 2 years to go profitable on a customer. 2 years. Now, to some extent, when you're big, when you're public and you're doing $500 million, it's okay because your base doesn't cost that much. So yes, your customer acquisition costs actually in traditional SaaS go up as you scale. But it's tolerable because you have higher NRR and so much of your existing base stays with you. But the reason everyone is so— it's not just that these AI companies are growing so quickly, it's that they don't have to spend much in sales and marketing to get there. And I'll give you like another, another personal example. I was talking with one of my old team who's now head of sales at an AI B2B company that just crossed $50 million. Amelia talked to him too. I think he has 2 people on the sales team, Amelia, at $50 million. 2. 2. And what did he say he was going to hire? 3 or 4 this year? 5? Maybe have a 5-person sales team. Yeah, but 1 to manage the AI. So 1 to manage all the AIs and like 3 human salespeople, right? Yep. So think about that for a minute. I mean, in the old days at, at $50 million, you'd probably have at least 100 sales reps in the old days because at $50 million you want to go to $100 million. So that's net $50 million of bookings, $500K net per rep. With scaling and turnover and churn. That's pretty good. So you would need 100 sales reps at $50 million. He's gonna have 5 and a team of AIs. It's not that he doesn't need sales reps. It's not even that he's not selling. He's actually a classic SaaS guy turned AI guy. He's just doing it. Not only is he doing it with less humans, but for purpose of this, he's doing it so much more efficiently and he has 10,000+ inbound leads a month. 10,000+ inbound leads a month. Now listen, they have COGS and they have token costs and you can knock them and they have competition. But that massive amount of inbound and the ability to close with so few headcount just makes these companies, at least from a sales and marketing perspective, and this is not intuitive, often radically more efficient than the traditional B2B motion where honestly, in some ways you're trying to jam a slightly differentiated product down people's throats, a slightly better marketing automation tool, slightly better drip marketing, slightly better cadence tool, the best of the best. And this is what I would challenge everybody to do. You're like, Oh, I launched a copilot. Oh, I launched, I launched an AI chat. My revenue hasn't grown. Have you done something that is so disruptive with AI it couldn't be done before? That's where the massive pull is. That's where the massive pull is. And yes, have I had a few bumps with Replit? Yes. But have I been able without an engineer to put 8 apps into production that have been used half a million times? Like, that is so disruptive. Or the fact that we can now, instead of having to send the same boring deck to every single sponsor at SaaStr, everyone gets a real-time deck that's customized for them with almost no work. This is just so disruptive that that's what you need to find in AI. And if you find that today, and eventually I guess this will all get mature, if you can do something with AI that is utterly disruptive with instantly perceived ROI that has never been done before, the demand is insane. It, it is just in insane. It is off the charts. And even a lot of the folks that are gonna speak today, a lot of, a lot of vendors and others that will, that will talk during the day, actually many of them have more demand than they can service, more demand, and they're turning away customers that aren't the ideal ICP or that they think will turn over, or in many cases, they just don't have enough forward deployed engineers to roll everybody out. They're literally turning away customers, which is in traditional SaaS, right or wrong, you take every deal that comes in the door. So radically more efficient for now, not, not obvious. Okay. This one's really interesting from Iconiq. Yeah, it, look, it ties to, to this, to the strong market pull, but another reason you can do more with less in go-to-market is more deals are closing, more deals are closing. So if you just, there's a lot going on in this chart, but they're basically showing you classic funnels. And if you look here, it's especially prominent at scale north of $100 million, AI-native B2B companies. Close turned 56% of their free trials to paid versus 32% of non-AI. Now look, not all free trials are the same. Some AI companies make you pay almost immediately. I tried to use an AI geo, like not SEO company, but, but for a geo for to see how, your AI SEO works. I couldn't even get any results without putting in my credit card. I bounced. Others might pay, but other AI apps are incredibly free for a long time. Like you can do so much on Opus Clip for free, make free clips for of your B2B content for social media. You can do more than you think if you're careful on Replicator Lovable for free. You can't do good. You can't do that much, but you could actually do quite a little bit. There are many apps out there that are surprisingly free in AI. So I think on the blend, they're probably comparable, but think 56% of your free trials closing versus 32%. I like that. That's radically different. That's radically different. So this is another reason they're more efficient. There's just so much more demand and people will convert. Okay. This chart, again, this is a theme and then we'll move on to another theme. But, um, this one is really interesting what you're seeing. It ties a little bit. It's the other side of the story we just told of someone on my old team that's at $50 million with 4 folks in sales. Going to add a few more in AIs. It's not that we have no humans in overall, the overall sort of sales and post-sales. It's just we're putting much more energy into forward deployed engineers, which we'll talk about next, and into post-sales. In other words, we're putting a lot more effort overall in AI B2B to getting folks trained and onboarded. And it's a massive change. So if you look at this in, in this chart, post-sales is 31% of AI native companies versus as low as 22% in traditional SaaS, traditional SaaS. And it's probably even more than that because if you think about what's traditional SaaS, well, I'll sell you this product. It'll take you like 3 months to, to pilot it, the rest of the year to roll it out to the rest of your team, and then 2 years to get ROI. And maybe even I'll have an agency at HubSpot or Shopify or Salesforce that will help you deploy. That's just not flying here. What customers are expecting, or they will churn with AI B2B companies and is required to train them, is that when we go, we go, it works. It works. So Amelia's talked about this. We've done a lot on our SaaS stuff with our 3 AI SDRs and AI— 5 AI SDRs and BDRs in total. They all took about 3 to 4 weeks to train and then iteration daily after that. And that training, that onboarding, we did it all, even us, and we're pretty AI savvy. Amelia is about the best that there is. We still did it together with the vendor. For weeks. And so you need— if you— if so many of the horror stories you hear about an AI SDR failing, I ask the folks, how much time did you spend training it? How much information did you— how much iterations did you— how much? None. Well, of course it doesn't work if you don't train an agent with a lot of data, especially in GTM. It just doesn't work. And so we all figured this out. And so as we'll see on the next slide, the biggest hiring in B2B is in forward deployed engineers. And if you, if you haven't been through it on the other side, forward deployed engineers, the term started at Palantir where they would do these, these 8-figure, 9-figure contracts and everything was semi-custom. And so they would have forward deployed engineers instead of having a business person that barely knew the product, they would put reasonably technical people sitting in their customer's offices, getting Palantir to work before they went live. And in some cases before they charged them, this has sort of become productized, not for these huge deals, but for $50K deals, it's hard to do it for for a $5K deal, but a $58K deal, the vendor will help take ownership of the fact that all of your data is in the AI and tuned and trained and working. If you don't do this, you're, you're, you're going to fail in these projects. It's not going to work. They do not work out of the box. This is one of the biggest lies in a lot of AI B2B applications that they magically work out of the box without training. They don't. So people, the smart vendors that are succeeding are putting less efforts into sales resources that frequently don't know the product whatsoever. And they're putting those efforts into folks that can make sure the customers they do have, because there's higher demand, are trained and onboarded extremely well in a way we've never done in SaaS. We've never done in SaaS. We've almost, frankly, most of SaaS, we've forced the customer to do it all themselves. It just fails in AI. So it doesn't work. So if you look at this next slide, it's just interesting. The forward deployed engineer is by far the strongest hiring trend in the last 12 months. It's off the charts. Everyone has figured this out. Um, let's, if you have a very simple low-end AI product, you don't need any forward deployed engineers, but for probably most people on this or watching now or later, you have a workflow product. If you have a product with workflow, a classic SaaS product, and you want it to work in AI, it is going to require training that agent. We're not yet at the point in B2B where AIs can train AIs or where so-called evaluations can do it all for you. No. Even on our own AI agents at Saastr, we have 10 years of data on 50,000 people that have attended our event and 500,000 people in our database. All that data has to be ingested. It has to be QA'd. It has to be tested. There, there are still hallucinations or hallucination-related issues. That all has to be iterated. Also, that when an email goes out, it doesn't say crazy things. If all you do is buy an AI SDR tool and don't train it, you will just have Saleslofter outreach from 7 years ago. That's all you're going to get. It's going to be no, it's going to be marginally better. And so you will think it doesn't work. You have to train these tools. And the answer is this huge trend of forward deployed engineers. And you know, it was funny, we did a 20VC with Harry, me, and Rory, maybe a month ago with Mark Benioff. It's a good one if you want to watch it. And he said, this is the number one thing he was jealous of Palantir was one of it was how, how, how, how well they charge their customers, but he was most jealous of what they've done in forward deployed engineers. He said, what I would love at Salesforce is that everybody's Salesforce AI works before they even pay us. It works before they even say go. Mark said this, that's what he wants. And it is so radically different from the way we've bought B2B software. When I was back, when I was a VP at Adobe, it took us almost 5 years to go live on Salesforce. 5 years, a lot of business process change, a lot of this, a lot of that. Business process change is still a big deal in the enterprise, but people with AI, the expectation bars have gone way up. They want something that is a step function. And so you've got to help them train these AIs so that when they go live, it's magical. Mark Benioff wants it, you should, you should want it too. Okay. A couple more points. This one's just Captain Obvious, so I won't spend a lot of time on this. It's still early in the AI journey and to some extent I would not be surprised if when a lot of the leaders go public, they start to have roughly similar headcount to what we're seeing in classic B2B companies because there's a little bit of convergence. B2B companies got the fattest and least efficient ever going into 2021. Like public SaaS companies often had less than $200,000 in revenue per employee because the markets didn't care. And if you, if you're one of the folks that still want to go back in time to 2021, it's never going to happen. It's never going to happen. Not only has AI changed the world, but we're never going to live in a world where $200,000 per employee is tolerable when you're public. It's now $400,000 and up, $400,000 to $500,000, but The pre-AI companies have figured that out. They've gotten lean one way or the other, but the AI companies are often hyper lean. I mean, hyper lean, not all of them glean, glean 1,000 employees at $100 million. Maybe that's a more, more classic look, but Lovable getting to $100 million with 45 employees. Cursor, I don't know if they really had 20, but certainly less than 100 to 150 to $100 million. ElevenLabs, the same. We're getting there with radically fewer employees. In the beginning. And there's often really just, it seems crazy and it is crazy, but there's really just two root causes, two reasons if you look at it. One is, um, they're often essentially single products. So that certainly helps. It, it pre, if, if pre-AI, the demand was lower. So you often had to have 2 or 3 products to get to $100 million in revenue. So you need 2 to 3 teams and 2 to 3 different sales teams and often 2 to 3 different engineering teams. That alone takes up a lot of people. These are often, Now that they don't have different features, but these are still mostly single product companies. And also a lot of them have very, very lean sales to marketing teams. So traditionally half your headcount or more is in sales and marketing. Traditionally 30 to 40% of your company is just salespeople. You're just not seeing that. So being single product and having almost having a much smaller sales and marketing team enables these companies to be smaller. Okay. Next point that I thought this was really interesting from the Iconiq data. And it's a reminder that AI washing don't work. Putting AI in your website don't make, don't, don't make you a rocket ship. 94% of public B2B companies now mention AI and say they have AI agents. Adobe on its last earnings call, and Adobe is I mean, it has some AI tools in, in creative, but it's, it's not ahead of a, of a, of, of, of a lot of the competition. I think it said it had $5 billion of AI-influenced revenue. AI-influenced revenue. What malarkey. But my point is everyone is either is an AI company or they're sort of an AI company. Everyone has a copilot. A lot of copilots are terrible. A lot of them don't work, but everyone's got a little AI, a little Clippy 3.0 waving at you to help you. Everyone's got some AI. It's not enough. It's not, it's not, was this, uh, could you AI wash late a year, 14 months ago? Maybe, maybe you get people's attention today. Everyone's an AI company. So what really matters is, again, have you built something that is so disruptive with AI that it will generate massive market pull because it wasn't done before? Because if it just saying that you have a, everyone's got an AI agent, it's just not interesting. 94% of public companies are a claim. They're AI companies now. Okay, maybe just, uh, uh, 4 more points and then we'll open it up to questions. Um, I think we mostly hit this one on headcount, but this sort of summarizes it. Startups are going, as they scale up, have, have error per FTE has gone from 182 to 237. Again, once you're public, it's more like 400 to 500, but this is, this is materially more significant. While operating expenses has remained flat, even with inflation. So we're spending less overall per employee and they're 20, at least we're at least there, we're at least getting 20 to 30% more ARR per employee than before. So, and we're not paying them more. Net-net doesn't mean that there's not salary inflation, but net-net we're not paying them more. So if you're not working 20, at least 20 to 30% harder than you were 25 months ago, you're behind the curve. And if you feel like you're working 20 to 30% harder than 24 months ago, good, because that, that is the minimum required to be successful in today's world. Okay. Just, um, two more, then I'll share where I think it's going. And this one is minor, I think, compared to the others, but I still thought it was interesting. I'm not sure I like this term offshored. I think we, we all have distributed teams now and most of us have globally distributed teams. So. Iconiq used this term, not me, but it is interesting to see that going into 2026, um, where we have far, far more employees that are what they call offshored, right? And, uh, it's mostly within engineering, has it been, but this trend is increasing. This trend is increasing from 24% of headcount, so-called offshored outside of your primary location and country, international headcount up to 30%. So. You know, just sort of interesting in that it's almost two countervailing trends. The whole world in AI is coming to SF. I put together, I should have put it to this slide, SF, almost all the hiring and net hiring in tech is in SF. It's twice New York. And basically after SF and New York, there is no net hiring. It's net negative in Austin. It's net negative in Miami. All of it because everyone there is flocking, but at the same time, We're using distributed and international folks of our team even more than ever before. So both, both are happening. We're getting more San Francisco-y and we're getting more global at the same time, which is, which is somewhat interesting. So think local, go global, think local, go global. Okay. Final one is Captain Obvious. I've written, we've written on this on SaaS multiple times. It's been on multiple Workshop Wednesdays, but I felt I had to pull this slide one more time so that you So that especially founders and executives know this in B2B. I know it's Captain Obvious, but all the venture money is into high-growth AI companies. High-growth AI companies. Last year was $363 billion, which was a massive jump from 2023, as you can see. But already this year, it's exceeded all of last year just in the first 6 months alone. $377 billion in the first 6 months. So the amount of capital into AI will be be off the charts this year. You can see it on Twitter. You can smell it. You can feel it walking in the streets of San Francisco. But here are the raw numbers. As crazy as last year was, off the charts, it's already more than doubled. And it is 70, 80% of all venture capital. It's not going into— just adding a copilot is not enough. 97% of public companies have an AI copilot or an AI agent. It's not enough. VCs are looking for what we've seen in this slide. They're looking for faster growth with better sales and marketing efficiency and huge inbound demand, huge demand from the markets. And this is one that I, again, I think a lot of folks are struggling with because many folks in B2B have had to generate a lot of that demand and that still works. That's still important. The leaders in AI B2B are doing events. They're— ElevenLabs is doing a huge event. I think in a couple of weeks, OpenAI has done one just the other day. Everyone's doing events, everyone's doing webinars, everyone's at the best events like SaaStr Annual and AI Summit and SaaStr London. They're all doing this stuff. But on top of it, they have massive inbound demand. And it's just what VCs want or what is on this. Fair or not, if you just look at some of these charts, ask yourself, if you, if you had a lot of money to invest, where would you invest? You would invest here too, even if there's risk. Even if there's risk that this revenue isn't that sticky. Okay. So just to kind of summarize before we take questions, what are we seeing at SaaStr? We're up to, again, adding agent force this week, depending on how you count, is either our 12th or 21st AI agent at SaaStr. And it's our 6th sort of AI SDR/BDR that we've added. It's just starting, guys. This is just, I think we're at the bleeding edge. We have other webinars and presentation with all of our data, all the vendors we use, everything. But it's just getting going. It's gonna be so much better in 6 and 12 months. And we're so early that, um, if you were a skeptic 6 months ago, please don't be a skeptic today. Please embrace the future. It's okay if you're in a conservative industry. It's okay if you hit your number without any help. I was just this week with a VP of, uh, support, uh, at a very, very fast growing B2B company. And I asked him what AI agent he used for support. He's like, no, we don't need anybody. Honestly, our product, I, we don't, we, we, we do a million interactions a day. We only get about 200 tickets. We have a team of 15 global and they can pretty much cover that. I'm like, maybe you're the one B2B startup I know that doesn't need an agent. He's like, can you answer every single customer query in real time? Well, actually, you're right. We can't. I mean, then at least have an agent do that. We're just getting going. We're just getting going. It's so early. And the second point, and this is one Mark Benioff made when we did this session with them. And he, you know, Salesforce does have a lot of challenges today. It's a $42 billion business and, um, there's a lot to do, but his point is so many of their customers have just started with AI. They're, they're so early compared to what a lot of us are doing. What it means is the good times are, are, are still to come. I mean, the penetrate, like from the first two points, the penetration rate is so low that just as we saw in the prior Cypher VC, 2025 is the first half already eclipsed 2024 for agents. And B2B AI, like 2026 is going to be an order of ma— maybe it might be an order of magnitude bigger than this year. It should be an order of magnitude. We've just gotten going. And so the third point, I've, I've tried to kick everybody's arse. You know, if you came to SaaS Annual, an AI summit this year, I told everyone you got to get working much harder. I've told people if you didn't get your AI, AI, AI product out by June 30th, you were too late. You were behind. There's no excuse. I believe in all that stuff. Okay. There's no, like, this AI stuff ain't new. Okay. If you're not in market today, it's pretty embarrassing. Your team isn't good enough. They're too slow. Okay. Having said all of that, and I stand behind all of that, when I look at where we're just getting going on agents and enterprises, it's not too late for anybody. We are just, no matter how it feels with Lovable and Vercel and Clay and all of this stuff, it's great that these folks have exploded and shown us the way. For most of you, your customers are still early. Most, most restaurants and beauty salons and regulated industries and big enterprises, they've barely started. So if you're behind, it's, it's not as bad as I thought, but it's time to catch up because 2026 is gonna be 10x larger for AI B2B than it is this year. At least, at least 10x larger. And the last point on this for the skeptics. For the skeptics, if you still don't think AI works, if you don't think it works for support or sales or marketing or anything, try a well-trained one. Go, if nothing else, go to sasser.com, click on the bottom right where we have our general AI agent from Delphi and talk to it. Share your B2B issues. Talk about a candidate you want to hire. Talk, ask, share a sales script, do whatever and interact with it. It might be great. It might be decent. I don't think you're gonna think it's terrible. And in fact, if you have a sales marketing go-to-market customer success conversation with Saastr's AI, which is trained on 20 million words of content, it's trained on every tweet I've ever written. It will be trained tomorrow on this video automatically. Everything I say will be ingested into that AI, AI tomorrow. That is gonna be better than probably 95% of the customer support you reach out to and see today. But it can show you that you can do the same. You can do this. Stop saying that this crap doesn't work and go try one that does work and start with us. It's free. Go to saster.com or saster.ai or go to Saster AI. It's really the same thing. And, and, and, and click on AI Mentor at the top. It'll bring it up. Ask any of your questions. Pretty sure you're gonna think it's— you may think it's— we have folks that are on this all day long. It's a little, it's a lot. Okay. Asking questions all day long about their teams. We have folks that come and go. 95% of folks say it's pretty good. So try one that works before being such a, such a Debbie Downer and a skeptic. And then a couple last points here, and I think the second one is the most important. And a lot of, a lot of sales folks are either struggling with this in one way or the other, or adapting, or just trying to adapt to it. But the reality is almost every VP of sales and CRO I work with wants a leaner team now. Now last year you could see this with CEOs. Uh, it started, it, It started pre-AI. People thought Elon Musk was crazy when he bought Twitter and laid off two-thirds of the company. He probably had no choice because it stretched him financially at the time, and Twitter was pretty unlean, just like everybody was. Everyone was unlean in 2021. Everyone had twice as many employees per dollar of revenue than they have today. People are twice as efficient as they were in 2021, and people thought Elon was crazy. Maybe Elon is crazy. I'm not, I'm not here to talk about politics or other things. The guy is pretty successful, but he might be crazy. Maybe we all are, maybe everyone in tech that's successful is crazy, but it presaged what was happening. And last year, every, you know, then the public markets put pressure on folks. So folks had to get leaner, right? And at first, CEOs in some cases were reluctant to do layoffs that they didn't want to do or to shrink teams because they wanted the kumbaya days of 2021 to last forever. But going into 2024, every CEO I know wanted a leaner team. No one wanted B folks anymore. No one wanted complainers or folks that took 3 weeks to get something done. Just CEOs were done, but I didn't, but I found sales leaders were the most reluctant to have smaller teams and marketing leaders. A lot of CMOs still needed 10 folks to run a campaign and they needed all of their agencies. And a lot of CROs were still, were stressed that without enough capacity, they could not hit the number. In other words, to go from $50 to $100 million, I need at least 100 sales reps. That's the math. 500K in capacity for reps and you better have 100 reps or I'm not taking this job. Right. You better go raise some more money. Even now, I think everyone has gotten religion here that they want leaner teams. They are tired of folks that don't crush it in sales and sales jobs are harder. So just be aware if you, if you, if you're still looking back on the good old days of these massive teams and not really knowing the product working, that no one wants this anymore. Not even VP Sales Heroes. Okay, this next point is probably the, the most important one I want to say to GTM leaders, and then we'll, then we'll open up to questions. It's calmed down a little bit, but if, but certainly the first half of this year and late last year, the vibe on LinkedIn was it's not working. Outbound doesn't work anymore. Everything's harder. SEO doesn't work. Nothing, nothing's working, guys. What do you guys know that's working? Because nothing works for us. Our terrible text messages to folks aren't responded to anymore. Our generic emails with 11 fonts and 4 colors, um, with, with that, that talk about customers that aren't even in your industry don't work anymore. It does, it doesn't work. Um, hiring the 20-year-old, uh, SDR with zero training, sending 1,000 emails doesn't work. Woe is me. Nothing works. It all works. Again, look at the leaders in AI and AI B2B. They are doing webinars. They have sales teams. They have, they may have 1 SDR human and 9 SDR AIs, but they're doing, they're doing outbound. Okay. They're doing outbound. They're doing events. They're doing field marketing. They're doing demand gen. They're doing multi-touch. They are doing S-content. They're doing content marketing. They're doing video. They're doing all of it. Because all the old plays work, just the playbooks don't. The playbooks don't. And so especially as CEOs, you gotta be just more— it's always been the, the number one risk hiring a sales or marketing or customer success or any GTM leader of hiring someone that just wants to bring the playbook from their last company. It's almost never worked, especially if they worked somewhere much bigger, but it works even less good today. Even less good today. I know less well, less good today because those playbooks are just stale. The plays work, but they have to be run differently. They have to be run more intelligent. They have to be run with AI. They have to be run at scale. Everything has to be trained. And so don't stop running the plays, folks. Don't stop showing up. At the end of the day, you have to, one way or another, you have to build awareness for your app. And even if your app has massive viral pull, Or word of mouth pull like a lot of apps. You need multiple touches. You need to remind things. Why is Sam Altman everywhere? That dude is everywhere. Okay. And OpenAI is the most successful startup of all times. It, there's at many levels, it's because multi-touch works. You've got to, it really helps to show up. And a lot of the, the most successfuls, Anthropic, OpenAI, these folks, uh, look, they're there one way or another, or they're all over the place. They are present because it works and they are doing events and they are doing webinars and they are doing marketing and go Go to the hottest AI companies and you're going to look and just calmly look at their marketing site. You're going to see all the old plays just often done really well for the AI age. So, um, make— don't use the old playbook, but don't not use the plays. They all work. And finally, we hit this one, but I, I, I know I've said this too much, but I, it's just not enough folks listen. There is no interest in classic SaaS companies from VCs. Growing at pretty good rates. If you're growing 80% at $20 million or 70% at $50 million, it's, it's not cool, dude, but no one's going to fund you. Nobody. And, and actually it's worse than that. And maybe I'll end on this. It's worse than that. And this one took me a little while to figure out. We, we talked about it a lot on 20VC this week, but, um, not only are VCs not interested in a company at $20 million growing 80%, there's something much worse. And this is just, I wish it wasn't the case because I don't have any, uh, silver lining here. Private equity firms aren't, aren't interested either. And there used to be from 2012, 2013 until 2023. So there was a decade where if your growth was decent, but not great in SaaS, but your burn rate was low and your NRR was high, maybe VCs wouldn't touch you, but a private equity firm would come in and buy you for 6 to sometimes 10 times revenue. You'd get to $20 million revenue growing, say 50%, 60% and cash flow neutral. Look, no VC was going to fund you there, but a private equity firm might buy you for $150 or even $200 million sometimes. I don't see any of those deals. They've disappeared because private equity isn't immune to the fact that public SaaS companies have seen growth decay, that NRR isn't sticky, that AI, new AI vendors are putting the old guys at risk. So PE has disappeared and VCs are flocked to hypergrowth AI companies. And this may or may not change, but at least don't live in a dream world where the next round, you think the next round's gonna come because growth is pretty good. It won't. If you have any doubt, we got the numbers. Go to Saastr.ai and click AI VC at the top. And we, and literally we added a new benchmarking. It's our third tool there. Upload your latest investor update or board deck or VC pitch. We will tell you the exact odds you get funded. The exact odds. And if you don't like it, don't shoot the messenger. It's based on all the recent data, 5,000 rounds and more. And I talk to way too many founders that think they're going to get funded in B2B and they ain't. So just at least find out. We built the tool, it's benchmarking, it's cool. And thanks, everybody. And, uh, Amelia, do we have any questions we want to get to? I know you got to run in a minute. Yeah, I got to run in a minute to go talk to Snowflake, but what's your opinion on how do you start to thoughtfully introduce AI SDRs to the sales team? Without scaring off the existing SDRs/sales team. Yeah. And literally, I was funny. I was at— it's funny, I was dealing with this yesterday. So I was at the company meeting for a company I invested in called Mango Mint that is SaaS for salons and spas and doctors' offices. Great company, great, great CEO, great culture. People were just so excited at their all hands and this CEO and I are really close. He said, and I had to do a presentation. It wasn't like this, but it was kind of where agents going. And he asked me, just do me one favor. Don't scare the SDRs. He said, don't scare them with all this stuff. I'm like, okay, I won't. But I actually, I'd already had lunch. I sat down at tables with all the SDRs. So I'd already actually had the conversation with SDRs. They understood that the best, this is the best have as more important a role as, as the rest and the others don't. And the VP of sales sat down with me and she gave the same message. She's like, we're, we're aggressively in on all these tools and we are growing. We're growing triple digits at, at, at 8 figures in revenue. And this is the future. We have to embrace it. So look, if you're, if your SDRs are going to quit because of AI, they're going to quit anyway. And the tenure of an SDR is very short. Just be clear, we need everybody that can perform. And it was funny, we were sitting there at lunch. And with, with the SDRs, and one of them was, was one of the best ones on, on, on Michelle's team. I'm sitting next to her. She's already got her laptop up. She's, she's scanning a salon that she's doing outbound to during lunch. Okay. She's literally doing pro— true outbound research during lunch. She's got a job forever. The SDRs that just want to run random emails and never learn the product don't. So I think this is not— we're past the point where you got to worry about scaring the team. Be kind. Make sure that everyone that crushes it— be clear to the folks that everyone that crushes it has a role. This is the thing. You need everyone that is A-tier on your team. That hasn't changed. And it's even harder to find the A-tier folks. But if folks don't want to adapt, they won't have a future. And I think you're better off being straightforward with them. So just do it. Bite it off. And here's the other thing. So many folks have learned it. We've learned it. Here's the other thing, reason it doesn't matter. As soon as you actually roll out the tool, someone's going to quit. This is what every company I've invested in has had the same story. Even at Saastr on our little team, we rolled out a tool called Momentum for AI or from Attention. They're great too. Even our little team, the day we rolled it out, someone on our sales team quit. The day we rolled it out. Why? Because now all his actions, all his data were going to show up in real time in a report and in Salesforce. You couldn't hide. You couldn't pretend you did 5 calls this day when you didn't. You couldn't pretend you did. I, many, many stories when you roll out this tool, somebody quits. It's all for the best. It's the truth is it's all for the best. We got to get out of kumbaya land because this is a world. This is a new high growth world. You got to be a part of it. Yeah. Related question. How much should we be estimating for cost to invest per year in AI to get started? It's a good question. Yeah. Okay. What's the cost? Listen, it's a great question. And, um, I think it's even better than it sounds. I've written this up. We— SaaStr is tiny, but here's, here's kind of a— I'm going to write a longer article on this next week. We spend maybe $10,000 a year on Salesforce because we're a tiny team. Maybe, maybe we spend a little more and I've been a Salesforce customer. Good God, for 20 years, guys, since I was 16, I've been a Salesforce customer for 20 years. First, first piece of SaaS app I ever purchased for real. I'd never paid for anything before Salesforce. And, um, but we spend maybe $10 grand. Notionally, notionally, we spend $500,000 on our AI agents across these, these 21 agents. $500,000. Think about that ratio for a minute. There's so many learnings in that ratio. There's so many learnings in that ratio. It's why Salesforce has to own Agentforce. They have to win. It's why HubSpot has to win here. Okay. But we are spending far more on agents than we We are on CRM. So that's, that's a meta thing to think about. The second thing to think about is these apps that need to be trained with forward deployed engineers and work. I don't think very many of them are less than $30,000 or $50,000 a year. And a lot of them actually try to kind of have a price point that's approaching $100K, like $60K a year base, base, and then $20K or $30K they want you to pay for the forward deploy engineer and the training. Sometimes they include it. Sometimes they charge for it. It's whether they do or don't, it ain't free to do it, right? For me, I would certainly subsidize it if I were a vendor, but it costs that amount. Like having a really great technical resource help you train your AI for a couple weeks, that ain't free. It's pretty, pretty damn expensive. So, and this relates to the Meta issue. You've got to invest the time and there's a reason these apps cost $50,000 to $100,000. They are a lot of money. Be wary of super cheap apps. Be wary of super cheap apps. It's not that we're not getting there. It's not that everything isn't going to get better, but it's— you can't cut the corner on training. So if instead of $50,000 a year, you're buying a solution that's, that's $500 a month or $50 a month, I'm not saying it's not going to work, but you're going to have to do even more training. You're like, you better take on a massive amount of burden to train this app because you're not getting the benefit of the forward deployed engineers and all the training, the tuning, the warming, and everything the vendor does. There's no shortcut here. There is no shortcut here. So it is, it is an existential issue that this will change over time. But right now the agents just can't train the agents themselves the way an FTE does. So it's, it's tough to get away for less than $30,000, $40,000, $50,000 for any of these tools really when a month of training is required or at least several weeks. All right. So thanks everybody. And we'll keep this conversation going and let's go learn from Snowflake. Thanks, guys.