He spent $0 on marketing for 2 years—then raised $400M at a $3.25B valuation. | AJ Loiacono, Co-Founder & CEO of Judi Health

Pablo Srugo (00:00:00) : How fast did it grow? Andrew Antos (00:00:01) : We did like $350k or so, maybe even a little bit more of ARR in the first. Not even six weeks, like five weeks and then we backed it up. The next quarter we did like over a million, you know, then we did some experimentation and then it went through there. You truly find the core of the product market fit once you sell a hundred customers. It doesn't mean everybody's super happy and everybody's at a huge deal size but around a hundred. You actually have a hundred goes at it and then the motion kind of reveals itself. My mindset has completely shifted. I think it's nine out of ten things don't work and they might look like they're working incrementally but that really means that it's not working. But the one out of ten things works so well that it carries everything else. Previous Guests (00:00:49) : That's product market fit. Product market fit. Product market fit. I call that the product market fit question. Product market fit. Product market fit. Product market fit. Product market fit. I mean, the name of the show is product market fit. Pablo Srugo (00:01:01) : Do you think the Product Market Fit Show, has product market fit? Because if you do, then there's something you just have to do. You have to take out your phone. You have to leave the show five stars. It lets us reach more founders and it lets us get better guests, thank you. Andrew, welcome to the show, man. Andrew Antos (00:01:18) : Hey, Pablo, good to see you. Thanks so much for having me. Pablo Srugo (00:01:21) : No problem, I'm excited for this. You just went through a big rebrand, so it's an exciting time. Andrew Antos (00:01:27) : The timing is impeccable. We just did a big rebrand yesterday to actually reflect the product market fit that we found in the last eighteen or twenty months and so I think this is a perfect timing. Pablo Srugo (00:01:39) : So, I mean, you'd mentioned to me that you actually found Product Market Fit three separate times through your journey with this company. We're going to talk about all of them individually, but let's talk about the last time that you found Product Market Fit. Because that's the beginning of the latest part of the journey and leads to the rebrand, and all those things. What happened? Take me to that day, tell me what happened and why it was like Product Market Fit to you. Andrew Antos (00:02:04) : Yeah, so we used to run this business that was basically like a mini Palantir for complex financial workflows in enterprise and so we were running the business. And one of the core ways how we would grow the business is we would do two big conferences a year. Where we would invite a lot of our customers to speak on stage and a lot of existing, and potential customers. And so we have this event, it's almost five hundred people. And we launch all this cool new stuff for that core business. And as a part of it, we also launched a thing that was an internal tool before. It was an internal tool that basically solved a problem around implementing AI for complex processes in enterprise. Where we would map out really easily, you know, work and just observing the work or uploading videos and things like that. And it was just this little tiny thing because our hypothesis was, hey, we can accelerate our implementation. It can be something where customers might even find new processes for us to automate for them and so we launched all this super complicated, cool stuff that we've been working on for a year. And everybody's like, OK, this is cool. But everybody wants to talk about that one little thing that we launched on the side. Every single person, every single discussion was like, oh, yeah, it's cool, all of these complicated things, it's really cool but that little thing that you're calling the architect is really, really, really what I want. How can I get my hands on that? And so it was this really important qualitative signal super early on. Because people were just interested in talking to you about it. Pablo Srugo (00:03:38) : And did you launch it as a paid product? Andrew Antos (00:03:40) : Not really. We didn't launch it as a product, we launched it as a widget kind of on the side. But people were proactively asking us how much does it cost? How can I buy it? How do you license and so on. And so, we were coming up with pricing and structure on the spot. The first twelve or fourteen deals that we signed in the first six weeks after we launched that thing. Every single one had a completely different pricing model because we were iterating through it, through those conversations. It took a while to clean up. Pablo Srugo (00:04:12) : A lot of times, you know, you launch a very sexy new product and you get a lot of hype, and attention but it doesn't turn into revenue. How did that translate into actual people saying, you know, I actually want to buy this, I'm buying this, swipe the credit card sort of thing? Andrew Antos (00:04:26) : So what we've learned from the previous times is people oftentimes get excited but sometimes it's too complicated to buy, right? Early, and so, what we've done is we basically did these early specials and we said, hey, if you really want to move forward. And you can move forward by the end of October, we're going to give you really unique pricing. You have lots of volume basically for a very small price, to get a signal, right? Like, do people truly mean it and the signal is overwhelming, right? So we took all the complexity of negotiating price and trying to sign big deals out of it. And we just said, this is the deal if you want to sign by the end of October. And people did, and so I think that gave us a lot of signal. Pablo Srugo (00:05:09) : How fast did it grow? Andrew Antos (00:05:10) : We did like $350k or so, maybe even a little bit more of ARR in the first. Not even six weeks, like five weeks on this thing and then we backed it up. The next quarter we did like over a million, you know, then we did some experimentation, and then it went through there. Pablo Srugo (00:05:27) : This was probably what? Two years ago, so 2024, but you've been at this since like mid 2017. Give me a little bit of background on just what you were doing before you started this business and then just the origin story on kind of the V1 idea. Andrew Antos (00:05:42) : It was interesting. So I'm originally from Europe. The only real job that I had before starting the company was I was a corporate attorney for a couple of years and I came to the US to do grad school. And I cross registered for a startup course at MIT as part of it. And that's how I met my co founder Nischal and we met in this entrepreneurship class at MIT. We immediately hit it off. We walked around Cambridge for three hours after that and we both had this feeling that AI is happening, and it's happening without us. Pablo Srugo (00:06:13) : Even back then, in 2017? Andrew Antos (00:06:14) : Yeah, because there was a lot of this behavior of the super early people in AI. So professors would have sign-ups for machine learning courses at 07:00am or 07:30am in the morning and there would be students in sleeping bags sleeping overnight. In front of the rooms to sign up for those machine learning courses, to get in and so we were like, if there's this much interest and this much demand in AI. Then it must mean that it's happening everywhere, right? And so we had this feeling like, oh, it's happening right now, and so we felt like we need to start something. Pablo Srugo (00:06:49) : It's interesting, by the way. That that would have been how you noticed AI. I mean, AI has been talked about for a long time. We invested in AI companies back in mid 2010s, but the deployments were still not that interesting, that compelling for the most part. But in universities, it would have been an interesting place to see kind of the future just by the nature of how much interest there was in the potential of it, at least. Andrew Antos (00:07:15) : Yeah, and I think that's actually. I mean, this is a Product Market Fit Show, right? So I actually think that's a very important concept for finding product market fit, is look at what people on the fringes are doing. Because oftentimes what starts as a fringe activity or as a fringe interest then becomes mainstream, right? We've seen there was a couple of companies that we worked with in the very early days in the 2018, 2019, time frame and they had AI CoE. Centers of excellence, already, right? These were big companies. Were they doing a lot? No, it was really hard to do it but they were already thinking about it as well and so you can see these fringe ideas. And oftentimes these fringe things over time become the mainstream, and you want to be as early as possible in the fringe idea. Pablo Srugo (00:08:04) : So taking back, I took you off the timeline. Andrew Antos (00:08:07) : No, so we're there, right? We're like, we need to start something and so the only thing that I knew was the legal world, right? Because I was an M&A attorney and so the first iteration of the company was, OK, what if we can automate contract review for law firms and in house legal teams? And so, we launched that and we built kind of a product. It was like LSTMs and CRFs. It was very early AI days. It didn't generalize very well, but we were able to sign some really big customers, right? We had Salesforce was an early customer on that, a couple of other really big companies and it reached some level of success. But not a huge level of success because the market just, people were not very interested and the product was really janky, right? Because AI just wasn't generalizing very well. So it would be a complicated implementation for a fairly small problem, right? Obviously, ten years later, now it's working. Pablo Srugo (00:09:02) : Timing is everything, crazy. Andrew Antos (00:09:04) : Timing is everything and so we went through that period. That was the first couple of years of the company. We're just iterating, also learning a lot about how do you find product market fit and also developing our skill sets. Pablo Srugo (00:09:14) : I'm curious, because obviously all these phases are, they're not super relevant today but they're instructive. What was V1 of that product? What were you able to do? I'll give you an example. We actually were investors in a company called Blue Jay, and they do like tax research. Answers for tax accountants, for example, and they've been around since like the mid 2010s. For a long time it was a struggle because they were able to answer a small subset of questions and so it never became this product that the customers could rely on. When GPT came out and they kind of went on it. It worked for everything and they got that growth that they always expected to have. And now it's been, because it just works flawlessly across the board. I'm curious kind of what your experience was back then. Andrew Antos (00:09:53) : That's exactly the same concept. So the first thing that we launched was an NDA review tool. So the problem that a lot of big companies have is you're doing business, your customer sends you an NDA to even have conversations and you have to have an expensive lawyer review that. And so, what we would do is, let's say that you're using the product. We didn't even have an application that you would log into, it was just an email. Dedicated email ID that we would give you, so it would be like nda.productmarketfitshow.com, for example. And so, every single time a customer would send you an NDA, you would just forward it to that email address. It would hit our service, it would get reviewed, it would get marked up, and you would get an email back three minutes to fifteen minutes with a reviewed and marked up NDA, right? But to your point, it didn't generalize well. Doing it with something that actually matters like an MSA or DPA or something like that was extremely, extremely complicated. It was rules on top of rules because it didn't generalize well, the accuracy went down. Pablo Srugo (00:10:55) : And you were selling to companies or law firms? Andrew Antos (00:10:57) : We started with law firms and made zero progress. And so we were like, OK, let's take the same product and start selling it to in house legal departments. Because they're not billing by the hour. Pablo Srugo (00:11:07) : Well, this is another thing that changed dramatically is law firms were like laggards. Because of the billing issue. Andrew Antos (00:11:12) : Yes. Pablo Srugo (00:11:13) : And that conflict. And the latest wave, they've just seemed to say, it's kind of you have to do it. Because everybody's doing it and that's unlocked a lot of spend. Andrew Antos (00:11:21) : That's exactly right, right? And I mean, they're also in the direct path of AI, right? But anyway, so that was the first iteration. So then we went mostly to in house legal departments and again, it wasn't OK, but it was basically a lifestyle business, right? It just didn't grow fast enough because the problem is not big enough for a sufficient number of people and so that naturally caps your growth, right? And that's how we really understood the dynamics of a market, right? You need to be number one problem for lots of very senior people. If you're selling into enterprise, so you can grow fast enough, so you have high enough deal size and you have lots of deal velocity. Pablo Srugo (00:12:01) : How long did this last, this first iteration? Andrew Antos (00:12:03) : Two and a half years. Pablo Srugo (00:12:04) : So this is where I'd really like to go deep on the story. Because it's incredible the number of founders that, this is like one of the most common problems. I mean, you put something out and it doesn't work at all. It's very clear, and at some point you're gonna give it up or you're gonna try something else. If it just takes off, amazing, good for you, right? But being in this situation where you've got something that kind of works, just enough to keep you going but really not enough to make it worth your while is such a common pitfall. What do you do, like two and a half years in. When do you start to realize that it's an issue and what sort of actions do you start taking to change, you know, pretty dramatically, I assume? Andrew Antos (00:12:44) : Yeah, so we basically just realized that this is a problem and we always had this view of, we just need to run a bunch of experiments, right? To basically get feedback from the market, right? There's no other way to do this. You cannot hypothesize your way out of this problem. So we would always come up with new ideas. I think what we realized after two years, it's not just a product problem, right? Because we were doing everything that we could with the technology at that point. But it was also a market problem. People just didn't want this at a high enough rate, right? It wasn't that thing, and most of our discussions with lawyers were like, hey, we're doing AI for law. And they're like, why would I do that? Why should I care? That's not a thing and so, we realized we need to get out of that buyer, right? That was kind of a critical thing. Pablo Srugo (00:13:33) : And you're where, by the way? ARR wise? You're doing half a million, a million ARR or less? Andrew Antos (00:13:36) : Yeah, close to a million. Like, maybe $800k. Pablo Srugo (00:13:39) : OK, so you've got something that's worth defending sort of thing but not enough. Andrew Antos (00:13:43) : Yeah, and I think it's also the size of our ambition, right? I think for a lot of people, they would just call it like, oh, this is a good lifestyle business. You know, next year, I'm going to be at a million and the year after that, I'm going to be at $1.3 and the year after that, I'm going to be at $1.55. Pablo Srugo (00:13:57) : Correct. Andrew Antos (00:13:58) : But I think both my co founder and I, we're extremely ambitious and so we were just like, this is clearly not what we want to do. And so they were like, we need to change who we're selling into, right? And so we looked around and we're like, OK, we have some tech that works well in terms of understanding documents. Which was a scarce resource, looking at 2019, 2020. So we have that, we have expertise in NLP, which also was a scarce resource and so we were like, who else has problems that are related to documents? And we sliced it by industries, and then we sliced it by functions. So we developed these hypotheses, we put all these sticky notes up and basically what came out of it is like we want to be function specific rather than industry specific. Because we don't have any unfair advantage in a specific industry. It's not like one of us has been in banking for twenty years and have a great network and we can just easily sell to banks. And so we said like, OK, we don't have that kind of an unfair advantage nor we're in a place that has that. So we looked at, OK, what are the functions? And so within functions, we basically realized, OK, the people that have the biggest problems with documents other than lawyers are finance teams, right? Because they're looking at the same documents, contracts, orders, POs, invoices, expense receipts. Everything is just some kind of documentation of a transaction that happened and all these finance teams needed to account for those transactions properly. And so we were like, OK, let's focus on finance teams. Let's do a research, and initial research was like, yes, you have teams that are reviewing contracts for revenue treatment. You have teams that are doing order management to book and build deals. You have teams that are doing the expensing, the expense receipts. You have AP teams that are doing invoicing. So there's a lot and those teams are pretty big. And so we were like, OK, there's a logic here. And so what we've done, we were like, OK, let's build a product. Let's spend three months on this. Let's build something on top of the technology that we already have, launch it and then we said in six months, we need to be $250k ARR. So this is 2020, so very different numbers than what you see in the market today. Pablo Srugo (00:16:11) : Thats right. Andrew Antos (00:16:12) : We need to be a $250k ARR from minimum eight customers, right? Then we put that guardrail as in we're finding product market fit now. The fact that we can deliver services. So we set that guardrail and then we just managed to that, right? So we went out, we tried to sell it to everybody who would be willing to speak to us and we ended up hitting eight customers. But like $350, almost. Like $340k ARR in six months and so that told us that, hey, where people are willing to pay for this. It's a bigger problem. So they're willing to pay more, even though they were super early and it also kind of scales pretty easily, right? If you can go from zero to eight enterprise customers in six months, that's a good start. So that's kind of like the first guardrail, and then the business continued to grow. Pablo Srugo (00:16:58) : And I'm curious, before we move forward. The analysis of all the different industries and the functions, and then deciding that you're better off going after a function, picking finance. Because they have potential problems with documents. That all makes sense. That's the theory side. Then you go out and you said, you started talking to these customers. Do you have a sense? I'm curious what that was like, how deep, how many customers, what kind of conversations, what kind of things did you ask? Coming up with a story of something that sounds like it's going to work is so easy and it's so easy to delude yourself, right? And so I'm curious what you did ahead of it to increase the odds of the story you told yourself actually work and ultimately, you know, got you to the $350k or whatever. Andrew Antos (00:17:40) : I think it was just crisp articulation. So there's, I think, maybe two components to this. So once we wrote up a document, right? Like the actual pitch. What would this actually be? If you put it in writing, and I think putting things in writing really matters. Because once you see it, you start iterating through that. But we basically wrote it down and we said, hey, your problem statement is you have teams of people who are reading documents, again and again. If we can cut that down by whatever it was, seventy percent. Then you get this much time back and what is the value, right? You don't have to hire as many people as you're scaling. You have better employee experience because people don't enjoy doing this thing anyway. You have better compliance because you have fewer errors. So we wrote all of that down and then I think I fundamentally believe there's two really magical numbers that you really need to care about and that's, thirty and one hundred. Because what we found is, again and again, you want to talk to roughly thirty people to validate an idea or invalidate an idea. Ten is not enough because you can always find Ten friendlies or ten unfriendlies. Fifty or a hundred is to difficult, too many, right? But you can easily have thirty conversations in a couple of weeks and thirty seems to be a really good number to get that initial validation. So we talked to about thirty people to revalidate it and then I think one thing that we found multiple times now is, you truly find the core of the product market fit once you sell a hundred customers. It doesn't mean that it's like everybody's super happy and everybody's at a huge deal size. But again, like around a hundred, you actually have a hundred goes at it and then the motion kind of reveals itself. Pablo Srugo (00:19:23) : Is this a hundred sales that you made or a hundred attempts? Andrew Antos (00:19:66) : A hundred sales. Pablo Srugo (00:19:27) : Gotcha. Andrew Antos (00:19:28) : And so, that's kind of like what we saw. So to pursue this business, we get it to about a hundred customers. Pablo Srugo (00:19:33) : And what exactly did it do for these finance teams and these documents? Andrew Antos (00:19:36) : It would read all these documents automatically so people don't have to, and it would kind of fill in a checklist. Because all of these finance teams have some kind of a view. Like, you're looking at a revenue contract to understand if you need to account for this contract in a different way. Or the management teams are literally looking at order forms and it's like, does the data that we have in Salesforce or ERP agree to what we have in paper, right? So we don't have some kind of a misstatement. So it's all of that stuff and so we would be able to usually punch out sixty percent on the low end and ninety to ninety three percent on the high end of all these documents that didn't have to be read by people that would drive efficiency and compliance. Yeah, and so we pursued that business. It was doing decently well. It was like 2-xing, 2 and a hal-xing year over year, depending on the year. But what we found out is kind of, I think, two things. So the first one was this moment where we launched this other thing that just took off immediately, right? And if you're looking at it, the numbers are actually very similar. We ended up doing like twelve or fourteen customers in the first five and a half weeks compared to doing eight customers and getting $350k in six months. Pablo Srugo (00:20:51) : And before that was probably $350k in a year or whatever it was, like for product one. Andrew Antos (00:20:55) : Yeah, six months versus six weeks, which is a huge delta and then I think the other thing that we also found. Which people talk about less, is product market founder fit and what we realized is the business that we've had before. We loved the product, we loved the business, but there was a really big component of professional services and delivery, and implementation. We were trying to crack that problem every single day for a couple of years, and we were just really struggling with it. We were not good at it and then you sometimes saw people who are really great at professional services and they just blaze through it and they do it with elegance, and they do it with this lightness. And we were just struggling with it so much. And so the combination of, oh, there's this thing that is just growing so, so, so quickly. And the combination of we might not be the right people for this particular type of business. Basically made it very easy for us and said, OK, we need to focus the business on this new product. Pablo Srugo (00:21:58) : And where was the existing business, in '24? How big had it gotten at that point? Like $3, $4 million? Andrew Antos (00:22:03) : Oh no, it was like $10. Pablo Srugo (00:22:05) : OK, so it's sizable. Andrew Antos (00:22:07) : Yeah. Pablo Srugo (00:22:07) : What do you do when you see this new product take off? As much as it's maybe more interesting and more aligned with you, do you just pause the other one? Do you do both at the same time? How do you think about that? Andrew Antos (00:22:19) : So it's yes and yes. So, initially we ran both at the same time and even though the people were buying the new product, right? Which is a transformation tool that allows you to understand how people work, turn it into a company brain, then allocate the work between people and agents, right? So ultimately that you can, it's incredible. Our customers are building these dream teams of people and agents with it now. But we had these two things next to each other, and all of this was not obvious yet. We just saw people are buying it and people are using it a lot. But like I said, you need a hundred customers to actually really understand the product market fit, the market speaks to and reveals itself. And you start seeing these patterns of value propositions and why people are buying, and how they're deploying. What value they're getting and how they're expanding. How they're renewing. You start seeing all of that and early on, we didn't see that. And so we always believed to just run four to five experiments every single quarter. Especially early on, to really figure out what's the shape of the product market fit. So we launched it, a bunch of people bought it. Then a bunch more bought it the next quarter and then we're like, OK, let's build a hypothesis, right? We had one hypothesis was this could be a compliance tool for internal audit teams and SOX compliance teams. One hypothesis was this could be a risk management tool for banks and especially places that have a lot of risk. We said this could be purely a post M&A integration tool, right? Where it's like, oh, I acquire a company, I need to merge the operations of two businesses. This could be an ERP transformation thing. So, for implementing ERPs or making changes to ERPs, or it could be an AI transformation thing. This helps you build agents, right? They're like, find where you should build agents, quantify the value, and then actually build them. And so, we built this hypothesis and we went to the market, and then basically saw what's happening, right? We got a bunch of these, a bunch of interest and so on. And then you start seeing how much are people using it. Are they loving the tool or liking the tool or not liking the tool? What's the average deal size in any of these hypotheses? And you start really seeing, you get some amount of data, and then it's easier to say, OK, we're going to go in this direction. For us, it was the AI transformation direction, right? We found actually AI transformation and ERP transformation actually pretty close to each other in a lot of ways. But that was the thing that was working better than anything else. And so then you iterate. So then you say, OK, we're not going to focus on these, we're going to focus on AI transformation, ERP transformation. OK, and then you iterate again, and then you build more hypotheses. Pablo Srugo (00:24:59) : When do you decide, like have you completely abandoned the finance product? And if so, when do you decide to do that? Andrew Antos (00:25:05) : So what we ended up doing actually is we always want to optimize for customers and the team, right? You always need to optimize for the customer. Because the customer keeps your business alive, and then you optimize for the team. Because the team supports the customer, right? Everything that we've always built, our philosophy is to be a customer first organization and so we saw the other business just skyrocketing, and it overtook it in about a year, right? Pablo Srugo (00:25:31) : Crazy, how fast did that grow? From zero to, was it zero to twenty in a year and a half or something like that? Andrew Antos (00:25:37) : Yeah, close to it. Pablo Srugo (00:25:38) : Crazy, insane. Andrew Antos (00:25:40) : And so, we saw that. And so, we were like, well, how do we do right by our customers, right? Because these are people who believe in you. They give you money to solve their problems. You always need to optimize for the customer first. We were weighing these options, right? Should we be supporting this internally? Should we be doing something else? What we ended up finding is we have this incredible partner that was supporting some of our implementations already. We worked with them for a couple of years. It's a professional services firm and they were like, can we take over this business and manage it for the benefit of your customers? That was a win win because they were able to support our customers really well, they were able to understand the product, they were able to continue the operations without any interruption, as well as they were able to take care of the team, right? So they literally took every single person who was supporting the product and brought them over. And so we optimized for the customer, we optimized for the team, and so we effectively sold the business to the partner, and focused the entire company on this new incredible business. Pablo Srugo (00:26:47) : You love this show. You don't want to miss the next episode. Why would you? So hit that follow button. Trust me, it's in your own best interest. So what I want to go deep on now is back to that product fit moment. You spoke a little bit here and there about what that product was, but I'm curious on the origin of that product. You mentioned you built it for yourself. I'm curious, what it was, how you used it and then. How you thought about and the way that you demoed it, and launched it. Because obviously that provoked a lot of interest. Just love to learn more about that. Andrew Antos (00:27:15) : So we were implementing AI in practice, right? The way I describe the business is like a mini Palantir for complex financial processes. And so, we would have the same kind of motion where we would have FDs or consultants coming in and they would spend a lot of time trying to understand the process. How do you actually do work, right? Pablo Srugo (00:27:34) : This is for the finance teams that you were serving. Andrew Antos (00:27:37) : Yeah, yeah and it's one interview after another, right? Imagine having ten Zoom calls a day or eight Zoom calls a day. Each an hour long, and just asking people, OK, can you show me how you do work? I was like, oh, you click on this in Salesforce and what does it do next? It's like, oh, and then you do this spreadsheet and then you do this pivot table, and then you send this Slack message, and then you do that. And these complicated business processes in enterprise run on exceptions. You have one happy path and that applies to twenty percent of things and then eighty percent is an exception of some kind. It's like, oh, if you see this customer, you need to do this or if you see this partner, this button doesn't work. You need to put it in a spreadsheet and do something else. That's how the reality is and so, it was always really painful and we would just be writing that down, documenting it and so on. And then we got some AI note takers, it was a little bit better. But what we've built was this little tool. Where you could take a Zoom recording with a customer that's showing you the actual thing and turn it into a business requirements document. Basically write up, this is how the process works, with screenshots, exactly what's happening, all these exceptions. Which would be a few hours of time of an FDE and we would just do that in two, three minutes. And it felt like magic to a lot of people. So that's how it started. So we solved the problem of how do you go from, I want to automate something in enterprise, find what I want to automate, how do I want to do it, and actually build the agent or set of agents to solve it. Pablo Srugo (00:29:10) : Is it similar to like a Scribe AI? We had her on the show. Andrew Antos (00:29:14) : Not really, I think about that more like a standard operating procedure tool. Pablo Srugo (00:29:18) : Right, yes. Andrew Antos (00:29:19) : This is how people do, I want to produce a standard operating procedure. The way we think about it is more like we're a context layer or a company brain. It's a new data plane, how work gets done, how people make decisions, how they get approvals, how long things take, and do that at scale through observation. Rather than building SOPs to help people onboard. Pablo Srugo (00:29:43) : And so that is what you demoed at that event? Andrew Antos (00:29:46) : That is what we demoed. It was a super light thing. It could do one small process at a time, right? Didn't have all this sophistication. We have customers that have tens of thousands of processes in the system now. Didn't have any of the sophistication and clearly it hit a nerve. That's what people actually want. That's what they need. Pablo Srugo (00:30:03) : Given that you've gone through this three times and obviously had different levels of inflection with each product launch. One of the things I've noticed speaking with a lot of founders that found Product Market Fit is, Product Market Fit has more often than not little to do with go to market. VCs especially, and I think a lot of early stage founders, first time founders, obsess over go to market. How are we going to grow? Should we optimize this? Optimize that? And a lot of the people who are in the position that you were with. Especially your first product, where it's growing but just not growing that fast. Their first idea is like, it must be a go to market problem. It just must be like, I'm just not that good at sales. I need a new salesperson. It must be that I just don't know how to get demos. Oh, my demo close rate is just not high enough. It's got to be something in the demo and my view has become, yes, you can optimize from forty percent to fifty percent through better go to market. Let's say on a demo to close, you can do that. You're not going to get from ten to fifty. That's not going to happen. I'm curious what your reaction to that is. Andrew Antos (00:30:58) : I totally agree. I mean, that was our first reaction too was like, OK, let's figure this out. Ultimately the team sometimes makes fun of me. Because I wake up early every day and oftentimes think about the market, and the team makes fun of me that I say, well, the market speaks to you, right? Or things like the team is saying, Andrew is just listening to the market in the morning. But I completely agree with you, because as long as you're showing it to people, as long as there's a way for them to discover. There's some basic level of discoverability and you're doing something that people really want. Then it takes off, right? And then I think if you're good at sales or good at marketing, and good at building product. That's like a difference between good and great. Pablo Srugo (00:31:45) : Yes, it's an accelerant. Andrew Antos (00:31:47) : Exactly, but it's not a difference between good and bad. Pablo Srugo (00:31:50) : So let's go into two tactical areas that we're going to go deep on. The first one actually gets a segue because you're talking about listening to the market. Everybody talks about listen to your customers, listen to what your customers say, voice of customer, all these sort of things. You've kind of built that into almost a science. I mean, you told me about a whole process that you have of making sure that you're constantly getting what your customers are saying and how your customers are talking about your product. Tell me kind of what you've set up there and then maybe we can go in more detail. Andrew Antos (00:32:17) : Yeah, basically, so what we do is we have every single conversation with customers recorded, right? Obviously you always have consent, but everybody's being recorded now, right? So whether it's a Zoom meeting or Google Meet meeting or whatever that is. There's some kind of recording, typically Gong or something else and then it all goes into Salesforce. And then from Salesforce, what we then do is we take all of the data. We enrich it with usage data, we enrich it with how many tickets and what type of tickets through Linear people are submitting. How many times they're reaching out to support over email or over phone. We basically take all of the communications with customers and we put it into what we call C360, Customer 360 database. So it's a super rich database of basically all interactions with the customer and how much they're using the product, and what ways they're using the product. So we have that data plane and then we basically built a bunch of different agents on top of it that are constantly looking and giving us these, weekly reports and insights. Like what do people hear? So do they have a positive sentiment or a negative sentiment? You can then slice the data. Is financial services trending positive or trending negative? Is software trending positive or trending negative? Are you hearing specific phrases, right? One of the things that you can do now is, give me basically an analysis of similarity of how the customer is talking to us, versus is it highly variable, right? And industries where the customer is saying the same words effectively again and again are usually industries where you can grow faster. Segments where you can grow faster than when you see a lot of entropy. They're describing the problem differently and so on. So we have this kind of monitoring layer that gives us these continuous insights, or continuously tweaking the messaging, the decks, the way we speak, the way we describe the problem, the way we ask questions and so on. And all of it is just about, OK, how do you go from that good to great? How can you increase conversion rates? How can you support your customers better? How can you drive more value? What are the ways that you drive value to the customer? And so, that's kind of the heart of the company now. Pablo Srugo (00:34:33) : So let's kind of break that down. I mean, the first one is you got to get this data. Obviously everybody's recording everything these days, so that's a no-brainer. But it's more about when are you reaching out to customers? Why are we reaching out to customers? And when you do that, what sorts of things are you purposefully asking? Because at the end of the day, we're just going to record a conversation. If the conversation is of little value, then you're going to get a transcript that's of little value. So you've got to elicit that. So I'm curious what you do on that side to capture stuff that matters. Andrew Antos (00:35:00) : You need to have that enrichment layer. I think people make the mistake that they're like, oh, I just record everything and then I'm going to reason on top of everything. Because also a lot of people don't want to reject people, right? And so you might have a very positive sales discussion, sales first meeting, but it doesn't translate into the next step, right? And so, you actually need to build the data layer with all these other types of data. So you can see, OK, well, whatever an existing customer is telling you has the highest priority, the highest weighting, right? Whatever a customer that's quickly progressing through the pipeline or progressing at the rate that you want them. We also treat that at a very high rate. If you have somebody who just gives you a first meeting and then it goes away. Then we're discounting that opinion. We're still digging into it, right? Because it could have been something, could have been our targeting. It could have been the way we pitched somebody from that specific industry and so on. So we still look at that, but in aggregate that's a little bit less valuable than what the people who are already working with you are telling you. Pablo Srugo (00:36:00) : That is interesting. Obviously, stitching it together, I could see the value in that. Because then you can slice it by what you just said but also, hey, of the customers that are sending us the most customer support tickets. Who are they and also what are they saying, and so on and so forth. But let me ask it like this, what sort of things have you gotten from there in terms of end outcomes, that because you're doing this, you've seen this or you've heard that. I'm curious to get that kind of tangible piece of it. Andrew Antos (00:36:27) : I mean, so for example, one of the core things is if you look at our website The entire story is really three steps. We understood how customers are talking about the problem that we're solving for them, right? That's the most important thing for every company, to actually truly understand the nature of the problem and you can start seeing these patterns of people are like, yeah, I was running this transformation project and it took $5 million and it took you eighteen months. And it took us six months just to interview a hundred fifty people in my organization to try to understand what they're doing, and it was still interviews. It was not the actual thing and then, we build something and it was the wrong requirements. So the most important part of this is actually understanding the customer pain, right? People are really good about talking about things that they know, right? And it's their problems. And so actually deeply understanding the problem across hundreds or thousands of customer conversations and seeing, are there any changes to the trend? Is the pain evolving in any way? I think that's critical and so our ability to articulate the pain to the customer and be like, is this what you're experiencing? Is this your problem? If yes, we can help you. If not, we don't want to waste their time. So that's number one. Number two is the way people talk really matters in terms of how do you articulate the problem in their own words then. I think that's really important as well, right? We had all these ideas initially. We were like, oh, should we be in this business of adaptive operations or an intelligence layer and so on. And then the customers were just describing it as, it's a transformation problem. And we would describe technically what it is. We built a context graph of how work gets done and they're like, oh, so it's like a company brain. And we're like, yeah, it's like a company brain. And they're like, why don't you just call it company brain? Now I can explain it to other people at my company. So you get these tidbits, because oftentimes when you understand the problem or your own technology in a lot of detail. You want to describe it precisely but what the customer cares about is, OK, what's the representation that I can then take and go, and talk to other people about? Pablo Srugo (00:38:40) : And how do you use this regularly in the business? Is it every Monday morning, everybody sits there and looks. You know what I mean? How do you build this into the process? Andrew Antos (00:38:49) : So each team, right? So marketing team, sales team, value delivery, customer success, product, they have their own agents. Where they're running the types of insights and the type of data that they want from this, right? So for example, one of the things that we really care about right now is driving the usage across our customers of our MCP connector. Because people who are using MCP are shipping more agents than people who are not, right? So we really care about driving the MCP usage across our customer base and so the product team that's working on MCP gets a report every day in the morning. And says, this is how much usage there was. This is the sentiment. This is the feedback that the customers are giving. These are the types of queries, all of these things and so, every single day they can just prioritize and iterate. And make the product better for the customer in the right direction. Pablo Srugo (00:39:38) : Now let's jump to the second tackler you wanted to dive into. You mentioned earlier, that the way that you perfected, like you did that launch and then the way to really perfect it, and hone in on what you now do. Which is transformation, was through this idea of four to five experiments a quarter. Four to five hypotheses that you're testing out, and you have leading indicators that you look for. Andrew Antos (00:39:58) : Yes. Pablo Srugo (00:39:59) : Tell me how all of that is set up. Andrew Antos (00:40:01) : So we do quarterly planning sessions. We don't believe in longer than ninety day cycles at the company. So we do these quarterly review sessions. It's basically with an extended executive team. We actually use this database thoroughly and basically see, OK, what are the core patterns? What are the core problems? Or what are the things that are working really well? And how can we double down on them? Make a shortlist, right? And then circulate it to everybody that we talk about it, and then we typically pick a couple of bets. And some of these bets have more of a go to market or customer success shape. Some of them have more product shape and we say, OK, this quarter we're gonna validate these. Usually it's four to five hypotheses. We assign it, we say, OK, what would success look like. I think that's critical. I think the qualitative stage of product market fit happens only very, very early, right? You launch it in an event and everybody wants to talk about it. It gives you conviction to take the next steps. But once you become a little bit bigger, once you launch more product and so on. Then it becomes tricky and you need to be very quantitative about it. And so we typically just spend a lot of time on, what's the startup metric that would give us confidence to basically resource it on an ongoing basis? So one of the things, for example, when we launched our MCP. We're like, we need one hundred users in nninety days. Otherwise, we're not going to consider this to be a huge success. Maybe we just are not that useful, right? And so then we focused, we had a clear person owning it, and we're like, OK, your job is to figure out how to get one hundred users on this. And we define the users like somebody who uses it this much, this frequently, and so on. And so setting that initial metric directionally. So then you can clearly say it's working or it's not working. So you can either kill the bet or you can resource it more heavily, is critical. Pablo Srugo (00:41:53) : What are some other examples of experiments that you've run, especially things to improve? I imagine once you're over $10 million and you're kind of scaling. You're taking every single part of your funnel and you're trying to tweak things. You're like, how can we get a little more leads? How can we improve conversion from MQL to SQL? How can we improve lead, whatever it is, right? I'm curious on maybe some of those where you're like, oh, this part of the funnel is not great. Let's try two or three experiments and then kind of how you thought about that. Andrew Antos (00:42:18) : I can actually give you an example. I'm sitting in one of our experiments right now, so. Pablo Srugo (00:42:22) : There you go. Andrew Antos (00:42:23) : So we did this rebrand, right? To really show, A, we're not finance only, but we're supporting HR and sales teams, and operations teams, marketing teams. Kind of across the entire enterprise operation and then the second thing is, I think it's reflecting our ambition, right? We have a $100 billion ambition that, you know, we think we can really build this company brain for every single Fortune 500, for every single Global 2000 company and so we were like, OK, what are some of the unique ways we can communicate these two ideas? And so one of the ideas that we had was, we did this report where we surveyed a few hundred enterprise executives and we built this entire report. It's called the State of Agentic Work and so we built this report. But so many companies do reports like this, and then they end up like a brochure that you can pick up somewhere, or they end up as a PDF that you can download behind a gate. And so what we've done is we took the twelve most interesting data points or learnings. We commissioned artists to create a visual representation of each of these twelve metrics. We've rented a space at the University Avenue in Palo Alto for an entire month of August. It was like a pop up store. It's a pop up data art gallery. Everybody should come, by the way. It's open every day, twelve to six, for the month of August and just experience the problem statement that we're solving, and the solution in kind of a completely different way, right? And we're tracking very closely now how many people come in. How many are our ICP versus how many people are just visitors and ICP we're tracking, both for hiring and for enterprise. And so you continuously need to do kind of interesting, unusual things and nine out of ten fail. But the ones that work, they work exponentially better. So that was one of the big bets for our Q3. Pablo Srugo (00:44:17) : Gotcha, I guess the other question I was going to ask is, I assume every single experiment has at least one metric that's assigned to it. That's specific, that kind of has a clear failure/success bar? Andrew Antos (00:44:27) : Yeah, correct and you can see the ramp usually very early, right? I think doing this for a while, you know, usually you don't actually need to see all ninety days for something to play out. You see the ramp, right? So you know, OK, my expectation is this, this would be success and I think my mental model has changed a lot. I thought when we started ten years ago, my mental model was most things will work incrementally, right? And some things will not work but most things will work incrementally and my mindset has completely shifted. I think it's like nine out of ten things don't work and they might look like they're working incrementally but that's really. It means that it's not working, but the one out of ten things works so well that it carries everything else. Pablo Srugo (00:45:17) : It is very, I mean, most people will talk about it like. There is no silver bullet, you just have to do everything. But it's almost like you do have to do everything in order to find the silver bullet. That really does kind of move the needle. Andrew Antos (00:45:29) : Yeah, exactly and it's a silver bullet for part of the business, right? And so you do one, so you need to continue, you need to build a machine that finds one or a few silver bullets every quarter. Pablo Srugo (00:45:41) : But it's, for example, your problem is specifically, I need to get more demos booked. Once I get a demo, it's great. Let's say, OK, you're going to try ten things but it's not like every single thing is going to get you. Most things are going to barely move the needle, and therefore, in your opinion. They're basically useless and then you're going to find one or two things. Those are going to move the needle, and then you just go all in on those until they stop working. Andrew Antos (00:46:03) : Exactly, and I think it's because people are applying how big companies are working. So if you have lots of scale, if you have one hundred thousand employees, you're a really big company. Then doing these incremental things really matters and you need to cover all your bases. You need to be omni channel, and you need to have ten different routes to market. And you need to have fifty different support models, and you need to do all of that. And that's true for a big company. But when you're a smaller company, and you want to grow fast. And you want to become that big company that can do everything. You need to figure out these things that work exponentially better. Because, I mean, to your earlier point, you cannot increment your way out of slow growth, right? It's usually something to do with the market structurally, like you have some kind of a problem. So even if you increase your conversion rate by five percent and you do all of these things. It's not going to magically 10x your business and so you need to continuously find these new things. Completely novel ways to address the market, to get that outsized, you know, distribution and value benefit. Pablo Srugo (00:47:07) : I see this especially, I would say, on marketing. I even experienced it firsthand on the Product Market Fit Show in the sense that I could do thirty things to try and grow it. Let's cut it into shorts. Let's put it on YouTube and LinkedIn, and Twitter, and this, and that. I see it in startups all the time, where it's like you hire a VP of Marketing and they're like, OK, let's do all the obvious stuff, right? Nothing wrong with trying it, but the idea has to be in order to cut everything and then do one or two things well. Because doing twenty things that are all incremental that also everybody else is doing. It's not, it's not going to get you where you want to go. Andrew Antos (00:47:39) : Exactly, right. For example, we don't run any Google ads. We have zero ads, right? Yeah, and exactly like a traditional VP marketing comes in. It's like, oh, you're not running ads. That's part of the playbook. You need to be running ads, right? You know, one of the things that we're finding is like, well, you know, we'd rather allocate capital to do events, right? Whether our own events or third party events. We spend enormous amount of our marketing budget on events. Because that's the number one way to do it and so, and sometimes like interview these people, right? I'm telling them about the business. Like do you worry about the fact that everything is one channel? I was like, no, because the channel is working super, super, super well. So I don't worry about that. I would worry if it was not working. Pablo Srugo (00:48:23) : Perfect. Well, listen, let's stop it there. I'll ask the question that we tend to end on, which is what would be your top piece of advice for an early stage founder that's still looking for product market fit? Andrew Antos (00:48:33) : Listen to the market. I think people, and it's kind of an abstract idea but I feel like if you have enough repetition. Like, when I was talking about the thirty and a hundred, really focus on these numbers. Always when you want to validate an idea, talk to thirty people, and then you can only start feeling reasonably secure about the nature of your product market fit once you have a hundred customers. It doesn't matter if it's a hundred customers who are paying you $10 a month or a hundred enterprise customers but unless you get to that scale. It's very difficult to feel good about your product market fit. Pablo Srugo (00:49:05) : Perfect, well, Andrew. Thanks so much for spending the time and it's been great. Andrew Antos (00:49:05) : Thanks so much, Pablo. This was fantastic, thanks for having me. Pablo Srugo (00:49:05) : Wow, what an episode. You're probably in awe. You're in absolute shock. You're like, that helped me so much. So guess what? Now it's your turn to help someone else. Share the episode in the WhatsApp group you have with founders. Share it on that Slack channel. Send it to your founder friends and help them out. Trust me, they will love you for it.