Live from SaaStr AI London GTM Stage

co-founder at Filevine.

How are you? I'm going to set down my water here. It's good to be in London. Thanks so much for being here. My name is Ryan Anderson. I am the co-founder and CEO of a company called Filevine. We are an AI legal company. And I'm gonna talk to you about how we became AI native. First, just a little bit about Filevine for those who are not familiar. About, oh, I don't know, uh, 3 years ago, pre-GenAI, pre-ChatGPT, we proudly considered ourselves a pure play legal tech SaaS company. And all of a sudden that moniker feels very old and very dated. But as a legal tech AI SaaS company, we were quite successful. We have 6,000 customers, 700 employees growing extremely fast. Uh, we will be, uh, kind of well above $200 million in ARR this year, growing above 50%, close to 60%. But we have successfully now made the transition to an AI-native company. And how can I tell you that? Well, it's in the numbers. It is very plain to see that the numbers back up that we are now doing far more revenue on a new quarter-by-quarter basis in AI products than in our SaaS product. Now, that's not to say that the SaaS product is in any way less successful. And in fact, it's still growing. The SaaS product itself is growing at 35-40% year over year. We are just growing so much faster faster on the AI side of the house. So the question for all of us in this room that run SaaS companies, that are leaders at SaaS companies, is sort of how did we do this and how should we all continue to go AI native? It is a very tough transition. And I'll tell you the trick number one, and the hardest part is that nothing is sacred. Let me tell you what I mean. We're going to talk about a very old movie. Now, I don't know if any of you have any idea what this bridge is, but there is a movie made— it was made about 30, 40 years ago called Bridge Over River Kwai. It is an awesome film. If you are a leader of a tech company and you are making this transition, go watch the movie. It's 2.5 hours long. It's very long, but it is well worth your time. The movie is about a general and his soldiers who have been imprisoned on a— as soldiers of— as prisoners of war in a tropical island in World War II. And they are assigned as prisoners of war to build a bridge. Now, obviously, these are British soldiers asked to build a bridge for the Japanese. They have no interest in building a bridge for the enemy. They're not too excited to be prisoners of war either. But their general says, hey, we're going to take pride in the work we do. We're going to build the best bridge we can possibly build and be extremely happy in the work that we do and proud to have done it for the people of this island. And at first, you're sort of very moved by this general who is so excited to build this bridge and do it the right way, even though the circumstances are bad. But slowly, you start to notice that this general is a little too prideful about this bridge he's building. And in fact, they build the bridge and the soldiers become very happy. They're extremely proud of the bridge that they have built. It works. Morale is high for all of these imprisoned soldiers, even though they are building a bridge for the enemy. But at a key moment, you notice that the British spies are about to blow the bridge up. and when the general realizes that the bridge he spent so much time on is about to get blown up, he commits treason, and he actually goes against the British Army and tries to prevent the bridge from getting blown up, from getting torn down, and many, many soldiers wind up being killed because he changes his mind and actually switches sides. Why? Because he is so prideful about what he has built. And doesn't realize that the larger mission has to be sacrificed. That is going on right now at your companies. You all have teams that have built incredible things. They are justifiably so proud of what they have built. How could they not be? You have products that are working today that bring in a lot of money, and you are going to have to have conversations with your leaders about tearing down what they have built. Literally large components of your codebase, not all of it, of course, but meaningful components of your codebase that work today, that make you money, will have to be torn down. Now, the trick, of course, is that not everything needs to be torn down, only the things that make sense. So deciding what should be torn down and what should not be torn down is a big deal. It requires a good amount of judgment. And at Filevine, we have come up with what we think is a pretty simple 4x4 matrix to decide what should and shouldn't be torn down. The number one thing you have, the number one advantage you have is that you have all this data and a working system of record if you are a SaaS company. So we say the things on the Y-axis that are critical to your competitive advantage and things that don't slow you down, that keep you fast and moving fast in the development of AI. Those things, those are easy calls. Upper right quadrant, don't tear that down. If anything, preserve it, make it better, fortify it. This is an area where you can succeed. This will be the cornerstone of your AI-native movement. The other easy call, the bottom left. Now I say easy call because logically it's easy. Emotionally, it will be just as hard. But the bottom left are things that don't help you, that are not important for your competitive moat, and slow you down. Why in the world would you keep those things? But I promise you, promise you, when you talk to your technical leaders who are building these products, some of them will tell you, we absolutely have to keep them, but they're slowing you down and you shouldn't. And these are emotional things. Of course it's emotional. Think about somebody who's worked 5 years on building a microservices architecture that does not serve your AI needs. And yet, You will have to take it away. It will have to go. So you're gonna have to tear those down. The upper left, bottom right, those are more challenging calls. Those are more nuanced. You're gonna have to decide, hey, does this thing that doesn't slow me down, but isn't really helpful for my competitive moat, do I really wanna keep that or tear it down or not? Those are gonna be hard calls to make, but they're much more nuanced and based in kind of judgment that's situational. That's fine. What's important though is that you look at those logically, and coldly and not based on somebody else's feelings. I often say that one, one of the challenges in working in any company, but especially a tech company, is our disposition to be agreeable. Unfortunately, in this instance, agreeableness is not going to help you. You're gonna have to be disagreeable as a CEO or as a technical leader making these calls to go AI native. All right, but your SaaS application does have a ton of value. We want you to move from content to context. Right now, your SaaS app has a ton of content. You have built an application that is very good at bringing in lots of data from lots of sources, cataloging it, storing it in the right places, making it easy to pull data out and be viewable by your customers. That's awesome. But in this world, the value shifts. We no longer care so much about the content itself. Self, we care much more about the context. And what do I mean? Well, again, we have tuned all of our systems for kind of ingestion of data via a keyboard. And as I say those words, everyone in this audience is like, yeah, that's pretty old school. I mean, geez, ingestion of data via keyboard. We all recognize that that sounds old, and yet we are all going to do that hundreds of times times today. So there is a big shift at play. We all know this is coming, that the world that we see today is not the world we're going to see in 5 years. Most data will come into applications not via a keyboard over the next 5 years. So we're gonna have to take systems that are tuned to get data in one way and get data out another way to be totally different. Now in a context system, and you have an advantage in building a context system, your application should ingest data agentially and take action with context. One of the best SaaS leaders, I think, in the world, he had a bad weekend. I don't know if you, if you heard Shopify went down, but Toby Luca is still an incredible leader. Uh, and he says that his number one job is the art of providing all the context for the task to be plausibly solvable by an LLM. So he gets it. He says, hey, Shopify's got all this data, we want it to use LLMs. I've gotta give it the right context. Walden Yam, the co-founder of Cognition Labs, the makers of Devin, an incredible AI tool, says context engineering, context engineering is effectively the number one job for engineers building AI agents. So you have all this content, it really can act as context for an AI agent. You are so, well positioned for the future. And yet I hear it's very kind of in vogue to say, you know, SaaS is dead. We don't need the database anymore. We're not gonna need all this stuff. We just need AI agents doing everything. But anybody who actually thinks about it for a little bit knows that that's not quite right. And I'm going to use a very simple analogy. Like most things in life, it comes back to the 1990s hit film Clueless. Uh, so we'll talk about this for a second. Imagine for a moment I came to you and I said, hey, good news, we have an AI agent that can pick out your outfits in the morning. If you're like me, I don't like picking out my outfits, it takes too much time. I would be like, awesome, that's awesome, done. I'll sign up for the AI agent that picks out my outfits. And in fact, Claire actually had, or Cher, excuse me, already had an AI agent that picked out her outfits. Um, but if you then said, oh, by the way, now that you have this AI agent, You don't get to have your closet anymore. We're not gonna show you your closet. You can't see it. In fact, it's just a bunch of unstructured data and clothes in a mess, and you never get to have it anymore. You'd be like, hold on, wait, wait. I would actually like to have my closet and the AI agent. Can I have both? Of course. Of course. That is the better world to live in. And that is the world where we as SaaS leaders actually have a significant advantage. Your SaaS application is like like Cher's closet. The agent helps take action based on the content inside the closet. It's a simple analogy, but that is actually the world that we're headed to, and it's actually the world where we are far more advantaged than our AI-only competitors. But to do this, our architecture will have to change fundamentally. This is not a small change. This is a big deal. I had an engineer say to me just a few months ago with a ton of with pride, mind you. We have built an incredible SaaS application that makes tons of money, grows fast, customers never leave it. Five9 has almost 96% gross revenue retention, 124% net revenue retention. He has every reason to be prideful of the application he's built. And he said, and the great news is now we get to sprinkle AI on top. That is fundamentally incorrect. We all know in this room that if you are going to build an AI application today, you can't simply say, hey, we've got all these APIs, OpenAI has APIs, done. We'll connect them. We've got an, an AI application. That is not going to cut it in 2025. It certainly won't cut it in 2026. To be AI native, you actually have to change the architecture of your system. It has to flip. The old way, the way that my first kind of one of my top engineers said is, okay, you have an AI layer that sits on top on top of, you know, your, your basic core services, your AWS, your data, your code base, the other kind of traditional services, maybe, you know, a calendar application, all, all the typical things that a CRUD app will have. You have to move from that to an architecture where the AI data layer sits right next to the AI application layer. And why is that? Because the ML engineers you hire are going to demand and they're going to require that they can tune and change the way data goes into those AI applications all the time on nearly a daily basis. And they can't be going to your old traditional tech team saying, hey, can you please change how the API provides me this data? That's not going to work. You are going to lose, you are going to go too slow. So we introduce an AI data layer and this layer owns how information is prepared for AI, how you ingest, process documents, emails, messages, and events. For example, we sell legal software. A legal case is just a graph of people, events, claims, outcomes. It's the exact same kind of material that you would use to build an AI data layer, but the ML team has to own that data layer. And when they do, I can tell you from personal experience, the AI applications you can build When you format it like this, dramatically better, dramatically more reliable, higher context, more complete, more accurate. Your customers will be much happier. You will actually be able to compete with the AI native applications. So together this makes up the core AI applications layer for things like Copilot, search, summarization, recommendations, and reporting. Think how important this is for a moment. Reporting, search in a world where you have AI, you want reporting and search to be AI native. You don't wanna even think about whether or not a user should semantically search or should use traditional search. Both I think are pretty darn important. You're still gonna want traditional search. There are reasons to do traditional search, but you don't want your end user to have to think about that choice. You wanna design with AI in mind from the very beginning. All right, to build all this, you're gonna need to hire AI natives. The problem is they don't wanna work for you. They wanna work for AI companies. So why in the world would AI natives want to work for your old SaaS company? I've got one piece of very good news. Besides wanting to work for cool AI companies, the good ones actually wanna work for companies where they have more access to data. Let me tell you a little bit, a little story. Uh, in our world, there are a ton, legal tech is such a hot area for AI today. There are so many competitors that have said, hey, give us your documents, we will run AI on top of those documents and we will sell you AI outputs. There are hundreds of these companies. The problem is that they think that that provides an answer to the customer. But we know at Filevine that to actually get an answer to a legal question, you need way more than the documents. Let me give you a simple suggestion or a simple explanation. Let's say that you wanted to ask our system, what's going on on this case and what should I do next? Now, your application might, might be, what should I do? You know, tell me about this opportunity for this customer and what should we do next? Is a very common question that you would use a chat application, a Copilot to ask and get an answer. But to get a good answer to that, uh, question requires way more information. It requires you to know who's done what. Those are audit trails. It requires you to know who is who, who is doing the things in your application. It requires deadlines. It requires a calendar. For us, it requires conflict checks and contact information. You have to know so much to give a complete answer. And in the world of legal tech, and I believe almost for all of you in this room, an incomplete answer is actually worse. It's worse than an inaccurate answer. Because the customer doesn't know what they didn't see, and that is deadly. But the good news is, because you have such rich data, you can sell this to the AI natives you're recruiting, and then you can sell them on your distribution. And here's what happened. This is distribution from a real product that we launched with, uh, just recently. It went from, what, 5 or 10 people using it a day to hundreds of people using it every single day in just a few months. This can happen to all of you, and your AI team will love using products that absolutely rip because you have distribution and data to make these products completely different than your AI-only competitors. The problem is you are going to have to acquire talent fast. I have a confession to make. We had an ML team. It was fledgling. Now we simply bought a company. You might have to go buy a company, and I realize that may not be an option for all of you, but we bought Parrot. Uh, it's an AI-native company, and now we've merged the two teams, our, our traditional data science team that we had before with, uh, the Parrot team that is AI-native today. AI-natives wanna work next to each other, sometimes physically, but they wanna work with other AI-native talent and build something amazing together. We now have that opportunity. We now have a critical mass of AI talent at Filevine that can move very fast. Now, you have a branding problem. You have a branding problem in the employee talent markets, and you have a branding problem because you have an, a SaaS application today. At least we did. Maybe none of you do, but we certainly did. Um, a lot of our customers would say, we love you for what you do, but you don't really say AI, uh, sell AI, right? We had to like convince them over and over again, we actually do have great AI products, and that took a rebranding. So the perception of who you are needs to change. And look, it can sound so simple to say, okay, we just updated the colors in the logo, but we did more than that. We planted a freaking flag. We are different now than we used to be. The old logo very much called out like a military legal vibe. We wanted to say, no, no, we are moving fast. We are up and to the right. We are bolder than we used to be. Let me tell you, this change, it meant a lot to our customers and to the people looking at us, but it meant even more internally. This change in our mark has told the people who work at Filevine every single day, the old mark is from a traditional SaaS era and the new one is from the AI era, and it actually matters. It is highly symbolic and you should have no problem telling your team, We are moving, and you need to give them a symbolic thing to look at for that change. But what gives? Why is any of this important? None of it actually matters, folks, if you don't build great AI products. Now, we have this sport called baseball in America, and, uh, I don't know if this works, uh, in cricket. I don't know the exact rules, but in baseball, the tie goes to the runner. So the person trying to get the runner runner out throws a ball, and if you, if the runner gets to the bag first or at the same time as the ball, the runner wins. Here is the good news for all of you. The system of record, the SaaS app, is the runner. In other words, we win in a tie, which is awesome. So if you are up against a competitor who's an AI-only product and you have the system of record, I promise you, your customers do not wanna leave. They are far more interested in staying right where they are and having their data and their documents and all the workflows that they've built on your SaaS application do all the work they were doing before. They want to stay with you, but your product can't be 90% as good. It can't be 95% as good. It has to be at least as good. Obviously we want to build it better, but it has to be at least as good. The good news is though, we do win in a tie. Now you have to obsess over usage if you— to know if your product is any good, are customers using it? At Filevine, we are religious about this. We do not let our teams roll out applications beyond beta without audit trail logging to know exactly who's doing what with these applications. These are real numbers. From our AI Fields product. It's absolutely massive. I think we've had 150 million actions taken on our AI Fields product in just a few months. Um, this is our DocReview product, also growing extremely quickly. We always are thinking in the language of daily active users, weekly active users, monthly active users. We think about this all the time. And our blockbuster AI product, this is by far our best one, is an incredible product, Chat with Your Case. It's a simple copilot. Pilot, but it does so much and is relied on by our customers for everything. It is fun to see them, you know, you'll see a customer use it like 5 times, then the next day they'll use it 8 times, then the next day they use it 20 times. And we now have a customer using it, I think, like 2,000 times a day. This product, just this one product, is growing in usage 10% week over week. You can build these kinds of products, but you're going to have to have a new team and a new culture. To build them. All right, assuming you've done all that, you have all of this data, and I'm now getting into the most controversial part of what I want to talk about today. Uh, I like to talk about controversial things. I think when you talk about something controversial, you know, you're actually saying something interesting. So leverage your data. You all have a bunch of data sitting in these SaaS applications. Again, we got really good at ingesting data. Investing data as SaaS leaders. Your data is your company advantage, but you are going to have a bunch of righteously indignant AI-only companies come to you and say, "I can't believe you have all this data and you won't give it to me for free whenever I want it. It's your customer's data. How could you possibly be acting this way?" Well, I've got a couple points that I'd like to bring out. Number one, first of all, It is entirely fair that if we spent years building a product and ingesting data and cataloging it and categorizing it in a way that can be easily ingestible by AI applications, that we want a relationship and we want to have a discussion before we simply allow access to the API. So we have control over our APIs today. We have moved from open API access to personal access tokens. We know exactly exactly who's accessing our API. We know exactly what they're doing. We know how they're using the API. We know how often they're using the API. They can't get at it without our express permission. We actually evaluate every single request. You might imagine these, these AI competitors, they really like this. Actually, they hate it. They're really upset about it. Um, but what we have found is that our customers actually totally get it. They totally understand our position on this. And look, we have never said one time to any competitor, you can't access our data. But we have always said, let's have a conversation about that. Let's, let's see what makes sense for us and for the customers and for you. Now, one little tip, if you're ever negotiating with these folks, they will demand access to your data like it's their freaking moral right. But the minute you say, okay, sure, let's have a conversation about that. But of course it goes both ways, correct? We can take the AI outputs that you guys get from our data and we'll get 'em right back into our system, correct? Isn't that how it'll work? All of a sudden the shoe doesn't feel so good when it's on the other foot. Um, just a little tip when you're negotiating with these folks. So it's fine to integrate, but review the requests. But most importantly, the AI traffic is going to reveal something to you. It will review, reveal promising areas for new development. You will all of a sudden see very early on which products are gaining traction, and you should 100% copy those products. You should copy those products and build them right into your system. You have the right to do this. Do not let these folks win this battle. You have worked too hard and built an application that is worth too much to your customer base. You deserve to win this, but it's going to be really hard. All right. Price to dominate. Again, this is the more, uh, Machiavellian part of the presentation. Uh, the good news is you have built a SaaS application that likely has very high gross margins. Congratulations. We were all taught that gross margins were a really important part of building a SaaS application. Your AI-only competitors, they struggle with margins badly. They have a really hard time with margins because they have all their LLM Costs are super high. So you get to do something kind of savage. You get to sell the AI products at a lower price point than your competitors can sell them at. Why can you do this? Because you can keep a blended gross margin at a much higher rate than they can keep it at. So yes, while your investors might say, oh my gosh, like, why are we selling the products cheaper than our AI-only competitors are selling them? Your answer is, because we're gaining market share, and by the way, our blended gross is still higher than our competitors' blended gross. And here's the good news. You are way better situated to win this battle than your competitors are. And the way I know this is because the best show on television taught me how this works. I don't know if any of you have seen Curb Your Enthusiasm. Oh, here's the chart. This is how this works. Uh, you can see the, the blue line is kind of what happens to the blended gross as you drive down the blended gross on the AI gross margin versus your SaaS gross margin. For Filevine, goes from 80%, and sure, I mean, maybe it goes to 60%, right? That would be an extreme case, but that's way better than the AI competitor whose gross margin is driven all the way down to 10%. And you get to be that kind of player, and you should, because I promise you, These AI competitors, they will cede you no ground. They will be just as vicious competing with you as you should be with them. All right. So back to how I know you can win this. If you have seen Curb Your Enthusiasm, hopefully you have seen the, uh, episode where Larry David, who is independently wealthy and constantly offended at all sorts of things, gets very offended at a man named Mocha Joe. He goes into his coffee shop. Coffee shop, and Mocha Joe offends him in some way, and Larry David says, you know what, I'm going to set up a coffee shop right next door, and I'm going to drive the price of coffee down so low that you will lose all of your customers. That's exactly what Larry David did in this Curb Your Enthusiasm episode, and eventually Mocha Joe had to come to him on bended knee and finally apologize for what he had done. You get to be Latte Larry in this scenario. Lastly, you wanna build hard things. I have a couple different components of this. First of all, you're ready for a major change. You should rebrand and think about your system differently. We have created a new category. We call it the legal operating intelligence system. I think you should think in category creation. It is something entirely new. You have a SaaS product, you have AI companies. Really, if you're going to blend the two, It's an operating intelligence system. For us, it's a legal operating intelligence system. For you all, you're probably gonna call it something different, but you are creating a different category, a blended category of software. Now what's cool about Lois is we can rethink the interface from the ground up from an AI-first mentality. We assume our users are going to use AI. Think about this for just a moment. I don't think any of us in 5 years are going to be sitting in a room talking about whether or not we should use AI in our products. It's the most obvious thing in the world that we're going to use AI in our products. That would be me, that would be like me sitting here today and saying, hey, do we think we should use, uh, the cloud or should we just stay on-prem? A ridiculous argument, right? No one here is thinking about that question. But a bunch of us are wondering how fast we should go on AI. Uh, also a ridiculous question. It will seem ridiculous, of course. But also AI, just like the cloud is now assumed today, AI is just gonna be assumed. It will fade into the background. We will talk less about AI. We will simply talk about technology. At the end of the day, that's all we have ever been is a technology company solving problems for customers. And that means that the way we price has to change. Filevine is moving from a subscription user-based pricing to usage-based pricing. On a, for us, it's kind of a hybrid model. We charge on a, on what we call a matter or a project, but this allows us to kind of charge each customer for how much they're actually using the product. We have found this is working really well. We already do it in a major way at Filevine, and that has worked incredibly well. We get We make far more revenue from our usage-based pricing customers than we do from our traditional subscription-based customers. But again, the product isn't just an AI plus a SaaS product, at least not for us. We have made a very important decision. We no longer sell to customers who won't buy the AI products. We no longer sell to customers who won't buy them. Why is that? Because we have to assume that AI is implicit in everything we build. We don't want to be making a distinction. Oh geez, like how do we sell this to the non-AI customers? That's crazy. There should be no non-AI customers. What are your customers doing if they're not buying AI? So for us, we have to assume in the way we build the product that AI will be part of the build. And if we're going to assume that, we have to assume we're selling one product. You build one tool and you sell to a customer who's going to come with you on this journey. Might we lose some customers? Maybe, maybe. Uh, uh, show me the lawyer that doesn't wanna use AI and I will show you the lawyer that's about to get, uh, his butt kicked. 'Cause like, I don't know where you go if you're not gonna use AI and you're a lawyer in 2025. But we are willing to do that because we cannot build in any other way. And by the way, it's way better messaging for your team. How do you tell all the technicians that are working on the SaaS product, okay, here's what we're gonna do. We're gonna work on the old stuff and we're gonna let the old, the new team, the new AI team, the ML team, the hot AI natives, they get to go work on all the cool kid AI stuff. No way. That doesn't work. We want one company and one product where we're all building on the coolest stuff together. Because at the end of the day, it has always just been about a customer with a problem. That's what animates us. That's what animates me. That's what makes me want to get up and go to work every single day. Can you solve my problem? That's what our customers want to know, and we can solve it with technology. So hopefully this has been helpful. I'll quickly review what we talked about. Nothing is sacred. You're gonna have to have some really ruthless conversations with your team. This is very hard. Be prepared. You will see you're gonna see some human emotion and reaction when you tell your coders something that you built that's working is gonna have to go away. You're gonna shift from content to context systems. You're gonna have to change your architecture. This is a real technical change. It's not just, this is not window dressing. This is more than cultural. You will actually have to change how your product works. Hire AI natives, consider maybe just acquiring a company, rebrand with intent, obsess over usage, never release an AI product, any product, If you don't know how it's being used, leverage your data, price to dominate, be savage on this, and build hard things that customers want. Pay less attention to the technology. It's all in the end solutions for customers. Thanks everybody. Happy to take any questions if there are any. I don't know exactly what the setup is like for that, but yeah, go ahead. Please. I don't— oh, it looks like there's a micro there. Yeah. Thanks for the presentation. It was really, uh, helpful to see how you're positioning and pivoting, or already have. Um, you talk about owning a lot of the real estate data. I'm a regtech platform vertical size in a very public utilities energy. The infosec getting through to even ingest that data It's a real challenge. So how did you bridge that, and how are you getting people to trust you and AI to give them all the information? What's the secret sauce? Well, so first of all, I don't— look, I mean, it's not easy. You know, in our case, we've become trusted, I think, over a long period of time by kind of proving that. So we do a lot. I mean, we're FedRAMP compliant, which is like a really hard thing to do. It shows that a deep investment investment in security. You know, I generally, I think you have to show some trust and some action on a small thing first. So even a little bit of data, once they see, once they've created a relationship with you, I think over time it becomes a lot easier for them to do more things together. Look, it is hard. I mean, it's kind of crazy when I think about it. I mean, we have every email, every text message, every note, every document, we have it all. We have it all. And all the medical medical information for all these clients. It doesn't matter. I mean, we, Filevine does like prosecutors' offices. So like, you know, state AG's offices. We have some really sensitive data. I think you win those relationships over a long period of time and be really diligent about it. I think it is hard. You're not gonna get it easily. I wish I had a better answer for you, but I think it takes time.

Thank you.

Agreed, great presentation. You mentioned a little bit about your pricing, that you've got a hybrid model. Yeah. A couple of things. Can you provide us some advice on how to think about a hybrid model? I like that approach. And also, what did you, how did you get your go-to-market teams up to speed on new pricing, new value conversations? Yeah, great question. So I'll talk about the pricing for a second. So we charge on a per matter basis. That's a little bit of a negotiation. One nice thing is we do have a lot of visibility into what our clients are doing. So we know a lot. We know like how much they're using the system. We know what kind of uses they have in the system, and we do have different AI products, and some of them we know our margins are a lot worse than others. And so we can say, okay, if you're gonna buy these AI products from us and we know what you're doing in our system, we're gonna charge you this much per matter. Flat rate. As of today, and this is changing, my mind might change, but as of today, I think customers still want some clarity on like what their price is going to be going forward. It's really hard to say to a customer, buy this product, I don't know how much it's going to cost you, but you're going to love it. That's a tough conversation. And I think it's really hard for a sales team to have that conversation, as you mentioned. So, so right now it's a flat rate per matter. We basically say to ourselves, So let's, we're gonna make pretty good margin on what we think is almost every single, whether you call it a project or an opportunity, you know, whatever your sort of unit of, of, of metric is at your company. But what I would say is if it's flat, it's fine. As long as you know you're making money almost all the time on even kind of the, where you're gonna use more AI, where your margins are gonna be bad because you'll make it up on the others, right? So like 90% of the time you're gonna have great margins on a per, on a per kind of matter basis or on a per project basis. And then maybe 10% of the time the margins are challenged, but it should blend out over time. We haven't had a problem there. It's been fine, but we do watch it really closely. And then on the go-to-market side, just real quickly, look, it's hard. I do think there's a customer education period. They like the user-based pricing for sure, because it provides a lot of clarity to them and it's cheaper. So, I mean, it's kind of crazy. You know, we have customers that have paid us. I mean, we have one customer in the southern, in the south of the United States. They paid us $500,000 a year for the software. They now pay us $4 million for AI. We have another customer, fingers crossed, we're gonna sign this week, that pays us $100,000 today. That's it, $100,000. And they're about to sign a $5 million AI contract, annual. Annual AI contract for 3 years. So, and by the way, like these stories are everywhere where we're going from, you know, single digit $200,000-$300,000 contracts to, you know, $1 million, $2 million, $3 million annual contracts because we're now replacing labor and customers are willing to pay so much more for that. Definitely an education period. It takes them a little bit of time to understand it, but I think you do what salespeople have always done, learn to drive value and show, you know, why it's worth them paying. The good news is, like, a lot of the work is being done because these AI companies are so loud about replacing labor. A lot of people are sort of bought into this idea already. Any other questions?

Yeah, Ryan, thank you very much. You've shared generously all the steps that we can follow to build AI based on SAS. And the thing I liked that you don't make distinction between SAS AI, you can see that as the overall. And we are building blockchain, decentralized blockchain, and we now understand that AI can handle simple task, linear task, and we need nonlinear and complex tasks. And we know that quantum computer computing will do that one. How can you see that's your perspective, how they can work in the future together? SaaS plus AI and plus quantum computing.

And I'm so sorry, I'm having a little bit of a hard time understanding with the accent. I apologize. Is the question sort of how to combine the products into one? Is that sort of the question?

Yeah. Yeah. To have 3 of them is— to have 2 of them is not enough currently. Okay.

I'm going to do my best, and I'm so sorry about the accent. I wish I— Yeah, yeah. Sorry about my accent. —could hear it a little bit better. No, no, it's my fault. I'm just a— I'm an American from the West. There's only so much I can do. But you're doing good. Yeah. I am who I am. But so I have a lot of thoughts on building one product. Number one, it helps your engineering team. Culturally, I think it's really important so that your teams aren't selling more than one product, they're not thinking about more than one product. From a go-to-market perspective, it's way easier. My CMO sitting right in front here, and I'm looking at her because she would much rather just say, this is one product where she's been begging me for years to kind of bundle up all our products and sell them all as one. And I've always resisted and said, no, it's nice 'cause it helps our net dollar retention, and we can kind of sell all these other products, and we can kind of bump up the price of the customer over time. but it makes way less sense today because customers just want this one integrated experience. So I don't know that I have a perfect answer for you, except generally my, my general view on like leadership inside a company that gets to any kind of scale is you just have to be really firm about the vision. And there will be a million people who come to you and say, it won't work for this reason, this reason, and this reason. But if you know logically that this is where it has to go, it's actually fine to have your team be pretty upset about it. In fact, sometimes I think I'm not even making the right decision. If everybody agrees with me and it's really easy, I don't even think I'm like doing the right thing because the hardest decisions are always the most important. Um, as far as like technically how to do it, um, I think you're going to want just really, um, it's going to take some time. We've, we've had several kind of like tech summits where we'll bring in leaders from across the company, technical leaders, and they'll spend 2 or 3 days a week together. Our team's dispersed throughout the world. So we have teams in Prague, Slovakia, Chicago, New York, and Salt Lake City. And we'll bring them all together, usually in Salt Lake City, and say, okay, let's figure out what this massive roadmap looks like. And it gets messy. I think it's contentious, like seriously contentious, but the good news is the prize is so big and I think the vision on, on kind of where we're going as a company and what's, what's possible, uh, is so obvious to everybody that they're all on the same page and we work through it. So, um, hopefully that's helpful. Um, hopefully that kind of got, got you the answer you wanted. Go ahead. And I think I've got maybe less, this is probably the last question.

Sorry. Uh, hi, thank you again. Very insightful.

Um, question is about data security.

How do you containerize your data so that the LLMs don't end up advising one customer on another customer's set of data?

Oh, especially highly secure, highly critical data.

Yeah, it's a, it's a really good question. So first of all, there are much more technical people than me at my company who could give you give you a better, more technical answer. But I do know that the embeddings database respects our permission sets. Today it's a little crude to be honest. Like we kind of, cheat's not the right word, but we do something really easy. We basically say if you are permissioned on this particular, we call it a project, then we're gonna, then you can access all the embeddings layer data for that project. Meaning sort of all the, the, entire vectorizing. We vectorize everything. So the minute it comes in, whether it's a, you know, a note, email, task, document, it doesn't matter. We, Five9 processes 20 million pages of documents a day. So there's a ton of data coming into the system. But yes, so, so we just know if you are permissioned on this project, then you can access our embeddings layer that way. And the embeddings layer we use, we use Zillas. You know, it's, it's maybe not one of the newer providers, It does have better security. Zillow is the same company that Apple uses for iCloud, so they have figured out security. So we take that really seriously. There are definitely people at my company who can give you a better answer. Feel free to shoot me an email, just ryan@filevine.com, and I'm happy to point you to the security folks who've designed this. We do take that extremely seriously. Any other— oh, I think that's— I'm actually done. I've been done for a minute. Thanks, everybody.

Okay, hi everybody, I'm Roz, CMO and VP Product Strategy at Wiz, and I am very excited to be talking here today. I actually I actually spent a lot of time thinking about this talk and what I'm gonna say because I think we've never really done anything like that in Wiz. We've never stopped to think back and share about how we look at the growth we've had and what are the principles we use internally to sustain that type of growth. So it took me some time to formalize it, but when Saastr reached out, I knew we couldn't say no to Saastr because Saastr has actually been part of our journey. Uh, myself, the executive team, over the years we've shared content, sent each other podcast episodes of some of the best SaaS companies in the world that shared their story of growth. So it's very, uh, humbling and exciting to be sharing our story on this stage. And Wiz is a cybersecurity company. We help organizations protect everything they build and run in the cloud. You might have heard our name earlier this year when Google announced its intent to acquire Wiz for $32 billion. Obviously, uh, a huge milestone for us and one that we are still processing. Wiz was founded in 2020, and over the past 5 years we've really experienced very rapid growth. And I forgot to take my clicker. Okay. Now when I say rapid growth, um, I mean it. Wiz was, if we look for example at this chart, we were one of the fastest companies ever to reach $100 million in ARR. And the humbling thing for me is that it wasn't just fast compared to cybersecurity or enterprise products, but fast compared to some of the best SaaS companies in the world, like Slack, like, like, um, like Shopify. I, I do have to admit that this slide is, I think, 2 years, maybe 3 years old. And this year, Lovable, Cognition, you know, all of the AI companies have done it even faster than our 18 months. But I felt like it's fair to keep AI companies off the charts. And today I will share a bit about the principles that enabled us to sustain and build for this type of growth. Now, when we think of growth, what, what do we even mean first? Right? I think in SaaS-land, VC-land, we always think of those two numbers of ARR, revenue and valuation. But the reality of anybody who's been part of building a company is that those numbers are almost the outcomes of how do you scale and build the things that matter. With speed and intensity and with, with as few things breaking along the way. And those things are scaling your team, scaling your product engineering, and scaling your go-to-market. Now, in Wiz, it wasn't like we were trying to hack for hypergrowth from day one, but we were trying to build something durable that could handle that type of growth. We actually started upmarket. Our first paying customer was a Fortune 10 company, which is very unusual, but it wasn't an accident. It's actually something that I think goes to a core part of our DNA and how we think and how we build for scale. Wiz was founded by four of the most incredible and talented and smart people in this world: Asaf Rapoport, our CEO, Rory Resnick, our VP of Engineering, Inan Kosika, our VP of Product, and Ami Lutzbach, our CTO. And Wiz is actually their second company that they've built together. Before Wiz, they sold their first company to Microsoft. And instead of leaving Microsoft after a couple of years like most founders do, they actually stayed for 6 years and they ended leading the entire Azure security product stack. In Microsoft. That's also when I personally met the team. I led Azure Sentinel, which is one of Microsoft's security products. Now, I think Microsoft is one of the best schools in the world for enterprise sales and for scale. So part of our DNA was that from day one when we built those things, we did have in mind building them in a way that could scale and support to support the world's biggest enterprises. And I'll share a bit about, about that journey today. Now, personally, for me, I've been lucky enough to be part of scaling both the product and the go-to-market at Twiz. So I started at Twiz as one of the first 5 employees and as the first product manager. And for the first 2 and a half years in the company, I led the product team. And also, my entire career, my background was product, engineering, cyber. And it was one of the incredible things was we had such strong product-market fit that we actually sold for a couple of million dollars before we even hired our first salesperson. So even being on the product side, I ended up doing things I never thought I'd do, like negotiations and legal and P&Cs, and actually kind of loved that. And we did have super strong product-market market fit. And about 2.5 years into the company, we were at this point where we already had a small sales team, a small but mighty sales team. We were scaling fast and growing that sales team. But the number one challenge for us started to become that, simply put, nobody heard of us. We would meet a customer and we knew that we would win every single POV we were in, but we'd meet a customer and the customer would be like, Wow, this technology looks great. I wish I've heard of you 3 months ago when I signed a 4-year contract with one of your competitors. So the number one complaint from sales team and the number one challenge we started feeling was just that nobody heard of us and that was starting to hold us back. And at that point, Asaf, our CEO, um, asked me to try and lead the marketing team too. Now I have zero experience in marketing. And I also jokingly call that when I moved to the dark side because it was my first role ever that was part of the go-to-market org in general. I spent my entire career in the engineering org. I think luckily a lot of the principles that helped us build and scale the product were the same ones we applied to building and scaling our go-to-market and marketing approach. And those are the principles I'm gonna talk about today. But one last thing. Caveat before that is I think, you know, we often look at successful companies and we kind of analyze back and we look at all the things that they've done to be successful, but it's easy to forget that there's also a huge element of luck and timing and the market you're in. Wizz was founded in 2020 during the pandemic, and although at the beginning it seemed like a really bad time to build a company. It ended up being really good for a cloud company, right? Organizations moved faster than ever to cloud, and cloud in general grows 30 to 40% year over year. And now with AI, organizations are just moving more and more into cloud. So we've also been incredibly lucky to be in the right market in the right time. And I have to be fair and say that I think if you don't have luck and timing, not sure the rest of the talk matters, but we can only control what we can. Control. So what are the core principles for us? It's value, it's clarity, and it's velocity. And they build on top of one another and they compound. And I'm gonna start with value. And value for me is first and foremost the core of the company is the product value. And product value is the same thing as product market fit. It's not about a product having a specific set of features. It's about does the purpose of the product really change something significant for your customers? Do they have a before and an after moment? I promise this is the only slide where I talk a bit about what Wiz does in cybersecurity. So, so bear with me. I'll keep it, um, minimal. So in Wiz, we really had that. Why did we have that? We had such strong product market fit because we were solving a big problem. For our customers. We were solving the challenge of cloud security. So before cloud, if you think about how organizations would build and ship software, it was slower, it was centralized. You had, you know, releases of software, right? Security teams could kind of keep up and things would kind of funnel through them. Then suddenly cloud came and every developer in your organization could turn on a service with a click of of a button, and security teams had no visibility into all of that. They were kind of left behind. They didn't even know what were teams doing in the organization, let alone have a good grasp of securing them. Wiz really flipped that around. Uh, we gave security teams a tool that within minutes of deployments gives them that visibility, shows them everything they have in cloud and all of the interconnections, and helps them understand what are the critical risks risks in their cloud environments. And more importantly, we did not just give that to security teams. We gave the exact same thing to the development, to the DevOps teams in their organizations. So now they both have this shared context, shared understanding of the risks they have in their cloud, and a way to start resolving them together. So for organizations using Wiz, for the security teams, for the development teams, it was a clear before Wiz and after Wiz. And the value that the product gave them was very, very clear and strong. We learned the importance of having strong product value the hard way early on. So not many people know that, but Wiz was actually not founded as Wiz. We started being called Beyond Networks because our original idea was a very cool technical brilliant idea around cloud networking. And in the early days, you know, we, we were talking to security leaders, to security teams, and we would explain to them about our idea and our solution. And the conversations were kind of good. I mean, it was a good idea and they would be like, oh wow, yes, this sounds super interesting. We'd love to talk again in a couple of weeks. Or yeah, let's talk again in a couple of months once you build some more. But after a while of that, we started feeling like, The actual next step of deploying something, of getting something connected, of starting an actual pilot, it was going slow. There was a lot of like friction. It was like we were pushing hard, but the pull wasn't hard. So we had to be honest with ourselves and, and do the humble thing of stopping for a second and going back and maybe admitting that although we were very excited about the idea, that maybe the strong value wasn't there for customers. They were not pulling. And we started listening differently and we understood that we were talking to customers about the solution for their cloud networks and they didn't even know what they had in their cloud. And so we pivoted all the way back to the core problem of cloud security, cloud visibility. And ever since we did that, we started hearing, real signals of value. So it wasn't, oh, this sounds like a good idea. It was, wait, how much does this cost? Because I'm not budgeted for this, but I wanna make sure I can start a pilot and deploying this in my environment right away. Or wait, let me connect you to that team who can start taking us through the deployment process. And ever since everything, it always went like this. The minute we hit the real, real, real core value, Now, the beautiful thing that happens when you have really strong value for your customers is that your customers actually become your strongest engine. Those are screenshots from a Slack conversation I had about a year and a half after Wiz was founded with, um, a person who leads the security team at, at a Fortune 50 company. And at that time, it was before I even made the move to marketing, I worked with him as as part of my product role. And he suddenly sent me those things and he told me, hey, you know, I had this really cool idea for you. Maybe you can share this with your marketing team. I thought, you know, people are starting to come back to conferences. I had this idea that it would be really cool if Wiz hands out those like things as swag. It's a mask holder so the mask doesn't hurt your ears. And I also have a 3D printer at home and I made a small model for you and I printed one. And if you want, they can send you the files and you can share it with your marketing team. This to me is like, it's the best proof that the customer really feels part of your team. And that's what it feels like when you deliver such a strong value to customers that they're happy to share your story and they feel part of your team and they give you that strong pull. Now another example is today Wiz supports more of 50% of the Fortune 100 are Wiz customers, and we have a lot of case studies on our website, including some video case studies. People sometimes ask me, like from other marketing orgs or enterprise companies, how do you get Fortune 100 companies to give you permissions to use their logos or even do a video case study with you? And they think maybe I have some secret about customer marketing. And I don't truly— in Wiz, customers reach out to their account teams after they've mitigated a, a risk using Wiz or seen the transformation it brings to their teams and offer to do a case study with us. So the amazing thing that happens is once you have super strong value for your customers, they really start pulling you forward and they pull you into new teams, new customers, new use cases. They become them a very strong engine and they were always Wizzr's strongest engine. The second thing that happened was that when it came time to think about our brand, we didn't even have a color or anything. So I didn't know anything about marketing, but I knew how customers felt when they used the product. I knew that they felt empowered. They felt like they suddenly had visibility to what's going on in their cloud environments, customers would even tell us, wow, this works like magic. How did you get all this information here? And so when I looked around at how marketing was done for most cyber companies, it kind of looks like this. It looks red and black and scary. And the concept is kind of scaring you from something bad that might happen to your environment. But I knew that was not what Wizz was about at all. So because the value was clear, it was clear to us that our brand has to reflect the same thing. Be optimistic, be magical, and also going a bit to first principles, we were trying to solve the problem where nobody's heard of us in the industry. So standing out also made sense, but it was very easy to make those decisions because the value of the product was so clear. We never, you know, we never brought a branding agency. We didn't have those discussions of values and colors and rebranding. We just decided that that is what reflects the brand and we went with it. And I think that also shows something that in Wizz, we've always been open to doing things the non-traditional way. Like the fact that every company brings a brand agency and that's what all the VCs told us to do. Didn't mean that we're gonna do it, right? We were always open to doing things in a way that worked for us. Now to summarize what happens when value is not clear. When value is not clear, it becomes hard to scale. You feel like you're pushing hard at something uphill. Now I gave the example of, you know, the dramatic pivot we had at the beginning of the company, but I think the reality is it's also happened in many, many other places along along the way where there's even something small in the organization that feels like it's uphill, you're pushing, there's all the friction. And despite the fact that sometimes it's very second nature, I think especially to people who build companies to push harder, sometimes the right thing is actually to stop for a second and admit that maybe you have a core problem, a value problem. And in many cases where we felt like something might be a go-to-market problem or a leadership problem, Um, it wasn't, and it was a core value problem. Now, once you have value, the next thing that this drives is clarity. Once the value you bring to customers is clear, that clarity is very powerful. Now, it also manifests itself in the product itself. The product in Wiz is simple, it's clear. At the product itself, when we show a security risk, it looks like this because we need people who are not from security team, we need the engineering team to be able to understand it. So the product itself is clear, but more importantly, the North Star, the value we bring to customers, the problem we solve for customers was clear from that day one and it never changed. That entire 5 years, it never changed. So we don't have those long discussions about what to build, those long roadmap and strategy discussions. In fact, I can't remember, a single meeting we've had over the past 5 years that lasted more than an hour because the north star and the value is super, super clear. So product and engineering teams can just continue in executing. And even when something dramatic happens in around us or in the market, like the AI transformation, which obviously changed a lot about what we do because now the organizations we are protecting build AI applications in cloud. And now a lot of what we do is help protect AI applications being built on cloud. Even that didn't cause a long strategy discussion. It was pretty clear what we had to do because it was still about serving the same north star. And we've been able to continue executing just, just as fast and go through that transformation. Today, Wiz also protects the environments of most of the frontier labs building models. So once you have a clear north, north star, your product and your engineering teams can just run fast and they don't need a lot of discussions. It also means that there's not a lot of one-offs, right? There's not a lot of things you build that are outside that north star because it's constantly clear. And we've really never had to refactor anything because we've always stuck with the clear value and the clear strategy on getting there. Now that clarity also helps with when it comes to approaches to how to grow your go-to-market. We didn't have those discussions on, you know, do we invent a category? Do we not invent a category? What words are we gonna use? Because in Wiz, the value we brought to customers was clear. So the, the task at hand became, how do we explain that in the simplest terms possible in simple language to to every single person who might be our potential customer. And with those clear boundaries, it's much more easier to be different, to innovate, to be— to— you have a clear box so you can be very creative inside it. And we've always taken that approach. We've said, okay, we can have fun with it. We know what audience we're trying to reach, and we just want them to know that we are here to help with cloud security. And so we constantly have fun with it. For example, for April's World School Day, we did like a Sisso musical with songs for security people. And for Women's Day, we had young girls interviewing executive women leaders in the security space. And both of those things are just fun and speak to our audience and show our brand in the best way and stand out. Now, clarity can also simplify things that are historically very complex motions, like post-sales for an enterprise product. That clarity in the value the product gives you enables us to define even simple things and make even complex things— it makes them simple and fun. For example, the post-sales process in Wiz, we kind of pivoted it to be around something that we call the Zero Criticals Club, meaning you bought Wiz, now let's Let's focus your team's effort on resolving the critical issues that our system has found. And once teams resolve it, they get like cool swag and we call it part of the Zero Criticals Club where today most of our customers are. So even something as complex and sometimes long as a post-sales motion to an enterprise product can become simple and fun. Now when clarity is missing, suddenly you have things that start to go slow. Suddenly you have those long strategy discussions and suddenly teams have a lot of meeting and meetings and people start, you know, aligning instead of executing. And you have people that are, their role is like managing that alignment, like project managers. And those are all signs for us that clarity might actually be missing. And what we do to get back to clarity is first of all, always allow ourselves to go back to first principles. Principle. Like, not how is this typically done, but wait, what is the problem at hand and what is the best way to get to that solution in our opinions? The second rule sounds stupid, but it really isn't. And it's the simplicity rule. It's if something is complex and you start getting to this point where there's many discussions on how to do it, whether it's something big in product or engineering, postpone it. It means it's just not ready. We always have a lot of things to do that are clear to us, so let's do them first and come back to that later. And in many cases in a fast-growing company, when you come back to that, everything around it has changed already. And sometimes now it's super clear what to do and you don't even understand what you were confused about. And sometimes it's still complex and then you can postpone it again. So it's a simple rule that works well for us. The next thing here is having also clarity in the roles that people play. Every person in Wiz knows his role and runs on and executes on it. And there's such clarity. And this starts all the way from the top. I know it sounds almost insane. And every time a new executive has joined our team, he also felt at the beginning like this is a bit insane. But even as an executive team, We hardly talk to each other. Every person has a very clear task and runs forward. So to give even more examples, when Asaf, our CEO, would lead the round, then I would sometimes be involved because we liked working on it together. And of course our general counsel was involved, but that's basically it. He would finish leading a round and he would literally text our WhatsApp group of execs and be like, hey, Hey, this is what happened. This is around. Closed. Next. Nobody thought it's weird. Each is super focused on his own task and there's complete trust. I've also never heard Roy, our co-founder and VP of engineering, and Inan, our co-founder and VP of product, ever arguing about what to build or how to build it because it's very clear. Inan decides what and Roy knows how. There's complete trust so people can just run fast and each have his own task at hand. And that translates to the rest of the company. And so it also builds a company of people that are just executors that run fast on their own task. And not a lot of people from the type that are like, you know, middle managers, product managers, all the things around it. So it keeps a lean and running team. And when you have both of those things, when you have such clear value that can drive that clarity, then really the company starts unlocking its natural speed because we don't have people waiting around for instructions what to do. We just have people that run fast forward. And again, I love how everything in Wiz also shows itself in the product. But one of the things that really differentiated Wiz was that most enterprise security solutions take months, sometimes years to show value. But we believe that time to value is value on its own, and that was a very clear priority for us from the get-go. So we use cloud APIs in a unique way that really enabled us to give time to value in 15 minutes. Like for Wiz, even a large enterprise, a Fortune 10 enterprise, can connect Wiz to their environment and in 15 minutes get that visibility and get that understanding of their critical issues they need to resolve. And that has allowed us to build the go-to-market motion to be super fast. I mean, a, a POV, a proof of value for us is not months. It's like connect, see the value, let's go. I think more enterprise products should think about time to value as like a core part of their value. And then also the product itself supported it. It's easy to understand, it's clear. It's well documented, it's self-service. So that also allowed the rest of the motion to be fast. We don't need consultants or deployment engineers. So having time to value be fast enables the go-to-market motion to be super, super fast. And we put a lot of product effort to make it that way. Also, the way that we literally work inside our engineering org is meant to support that type of velocity. So in Wiz, if we have, let's say we have an area in the product that is focused on network security. So that would be almost like a mini company. Like you'd have engineers, product managers, product marketing analysts, and they would be kind of running fast on their own, but they're supported by a very strong infrastructure of first of all backend, meaning everything in Wiz gets layered on top of one single graph database. So everything connects at the end. And also a strong infrastructure of design system and UI. So every team can run on its own, but they're all building based on like one infrastructure, which ends up in one cohesive product for our customers. Now that velocity is also really important on the go-to-market side. In our industry, in security, you can wake up and the entire industry wakes up to something that's happened, like let's say a vulnerability in DeepL. Me seek. Now you wake up at the exact same moment like the rest of the industry, every customer and every vendor, and you want to become the best and most helpful voice in the room, and you wanna be there first, be the most helpful voice in the room and be there first. And so you need to build a go-to-market engine that knows to respond to those things in minutes, in hours, definitely not in days. And I'm sure this today translates to every domain. I mean, news cycles are fast, attention span is fast. You need to have an engine that can respond fast. So what did that mean for us? It means we can't rely on things like slow approval processes or slow design processes or going to external agencies or having long trainings for people when something happened, but we needed to build an engine that responds and executes in minutes. Now it is Saastr AI and you have to talk about AI, but also I feel like I mentioned before that we've been very lucky on timing in Wiz, and I think AI was also a big piece of that. The AI transformation met us exactly at a point where it enabled us to continue scaling without having to scale necessarily in people and have big and slow teams. I wanted to, to give a couple of super clear and specific examples to bring it down to earth because I feel we're always talking about AI, but I, I, I specifically picked a couple of simple examples and visual examples, mostly from the marketing team to kind of show how that enabled us to do that. Because now AI really helps teams, even if they're small, be able to execute projects at a scale of, you know, a world-class company. Large team. So for example, this is a tool we use in, in my, um, branding design team where a designer used to spend a couple of hours, half a day maybe designing and illustrating our blog covers. Now we just have a custom chat that takes a sketch just like you see. The output is the design. That's it. Like you saved from hours, you switched over a task to minutes. Another thing AI enables you to do is to take on projects with a level of scale and detail you wouldn't even consider before. We, we recently released, just as a funny thing, an emoji pack for security people with a lot of emoji that are funny for security people. Now it's hundreds of emojis. It's a lot of detail and a lot of design. But with AI, we don't have to think about it. You can design hundreds of things using, again, one custom chatbot, and it takes minutes. So AI also allows us to just deliver on projects at a new level of scale and detail than we used to be able to do before, or even consider to do with a small team. The next thing AI enables is a lot of self-service across the org. Now when an employee in Wiz or somebody in Wiz wants to build a deck or something, and he wants illustrations or icons that are designed according to our brand guidelines, he doesn't even have to approach my team. We have chatbots in Slack where they can just ask an icon, get an icon in our brand guidelines. The whole loop of somebody talking to my team is just not there. People can self-serve themselves. So again, it allows you to keep your team small and agile. And then last but definitely not least, um, it's also completely transformed how we think of enablement of growing our go-to-market team, of growing a global team. We have built in Wiz, um, an internal chatbot. It's basically a multi-agent system that is connected to our internal documentation. So all of your compete knowledge, positioning, et cetera, the product technical documentation, to our internal Slack channels, to the wide internet, to, uh, telemetry from the product. And what this is basically, it's like your PMM best friend for every person in our go-to-market team. So it helps them with any question about how to position, what to say to this customer. It even helps them analyze their customers and tell them, hey, we see this sudden drop in usage. Maybe you can use this article or this case study. It's like their go-to-market best friend. It's a new approach where we don't have to enable everybody on everything all the all the time, but you just get what you need in your context of your customers. As somebody in our go-to-market team, it also helped us scale globally. We opened, um, our Tokyo offices earlier this year, and suddenly we have, you know, a go-to-market team that speaks Japanese. Obviously all of our content internally is in English, but we never even had to think about it. They just talk to the chatbot, the chatbot answers in Japanese. I mean, I can't even imagine imagine approaching this without having something like this on our side. Like, I genuinely don't know, like, what, we'd have to hire people to localize everything, all of the internal knowledge in Slack, you know? AI enables us to do things in a speed we've never been able to do before with small and agile teams. Now, I wanna summarize this part about velocity because I think what I've realized over time is velocity is by itself a moat. When you run fast and you run faster than your competitors, then every customer you acquire only grows that gap and makes it bigger. Every customer becomes part of your moat. And so when you run faster, those things compound. And building a strong brand is also something that just helps it become much faster. Customers trust you faster. You grow faster. It's like an accelerator to this growth, especially in very competitive and noisy markets like the one we are in, um, cloud and AI. So over time, velocity has become this moat by itself. Those advantages compound and they result in, in really what becomes nonlinear growth that's very hard to catch up with. Okay, I wanna summarize. I think for us, if we, think back to those principles, it's about, okay, is the truth there, right? Do we really have clarity on the value something is bringing? Then once we have that, how do we explain that in the simplest terms possible and simplify everything around it? The go-to-market motion, the messaging, everything, and continuously focus on executing on it and executing fast. Um, I hope today's hope this talk was helpful. I know of course every company is so different and there's like a gazillion nuances to every single thing I've said. There's also a gazillion mistakes we've done along the way. But hopefully if even you can apply this to like one small thing where suddenly you're like, wait, maybe this area, this team, this project, which is going slowly, maybe I don't have to push harder, but maybe I have to stop for a second and rethink a bit. The value and the simplicity, and then I hope it was helpful. Thank you so much, everybody.

[Speaker:JASON M. LEMKIN] I'm going to talk to you about the transformation of our our company. ChatGPT just had a birthday, it's 3 years old, and if you look at Intercom 3 years ago versus Intercom today, they are entirely different companies. They're, um, it's unrecognizable. And this is really important because a lot of SaaS companies, I think, are saying they're AI companies, but they're not really AI companies. They're not true deep AI companies. We've gone through this transformation. I'll show you what it's been like. I'll show you the lessons we've learned. It's been brutal. Like, brutal. It's really hard. It's really hard. And if you're a SaaS company who thinks you're an AI company and you've not gone through brutal transformation, uh, you're not there yet. And so I'm going to show you what we did, and hopefully it'll help you kind of think about things. So Intercom was founded in 2011. I've been there since the very start. It's been a long journey, a fun and energizing one. It's been a bit of a roller coaster. We were a breakout SaaS company. We went from $1 million to $50 million in ARR in 3 years, which back then was really impressive. Now it's not so much. Things have changed. We really, really had these great years and they ran for quite some time. The company was going really well. We'd built a great product. We had really strong product market fit. We've lots of customers who love the product, who would tell other people about the product. But things changed. We then had 5 quarters of declining revenue growth. That arrow is obviously supposed to be going along the lines. It felt like that. We had 5 quarters of declining revenue growth. So revenue still growing, but the rate at which was growing was declining. And this was not a good period for the company. We had gone down the IPO journey. We, I think, had changed how we think, changed how we work. If anyone's has done the IPO journey, it changes the company. And we were in a kind of a bad spot. And Eoin, our co-founder and CEO, came back into the CEO role. And then ChatGPT showed up 3 months later. So Eoin had just come back. We were in this kind of tricky spot, and ChatGPT showed up. We had a machine learning team at Intercom since kind of 2015, 2016. And a combination of the machine learning team, an excellent team, telling us this was— this is the moment. This is a one-way door for humanity. This is the real deal, the iPhone moment, the next technology cycle. We had very knowledgeable people telling us that. We also had personal experience doing it. A lot of us had started our careers in the early internet. We kind of knew what disruption felt like and looked like. A bunch of us had worked in different scenarios. I'd worked in the mobile team at Google when the iPhone came out, which is an interesting place to be. So you could see it, we were pattern matching and we could see that this was a really big deal. So we bet the entire company on AI. We bet the entire company. It was a risky thing to do. We ripped up our strategy, we ripped up our roadmap. We made these decisions in a very small number of days. Like you're talking 1 to 2 weeks after ChatGPT. We decided, you know, we're in this tricky spot that we had nothing to lose or felt like we had nothing to lose. We had like hundreds of millions of error, a lot to lose. We felt like we had to do it. So we built Finn and launched it in March, uh, 2023. So we went very, very quickly from betting the company on AI to launching Finn. And Finn was, um, an AI agent for customer service, and people really liked it and found a lot of value in it, and it worked. If you fast forward to today, Finn has over a million resolutions a week. Week. So every single week Finn is resolving over 1 million customer problems. We've over 6,000 Finn customers, and Finn's average resolution rate is over 65%. So the resolution rate over those entire 6,000+ customers is 65%. So in other words, Finn resolves 65% of the queries that it's asked about. Amazing. And Intercom is on this, like, entirely different journey now. We have a very bright future. Finished this groundbreaking product, transformational product. But the journey from Owen coming back and ChatGPT 3 years ago to today was really hard. It was really hard. And we learned a ton of things along the way. So I'm gonna talk about a bunch of the, the biggest things for me. I'm gonna talk about building a culture of change and having no sacred cows. I'm gonna talk about why does your business exist in a post-AI world? How do you build software? How we build software at Intercom is completely different to how we built software 3 years ago. And actually building software, I think, was our strongest thing. Some companies are marketing-led, sales-led, product-led. Different companies have different strengths. For sure, building great products that customers loved based on deep customer insight was our strength. We built processes around it, and that's all gone. We've completely changed how we build software. And if you're successful at these things, you end up with kind of two companies. You end up with Intercom and you end up with Fin. And they look very different. I'm just gonna talk about what we're doing with that. And lastly, I'll finish up by talking to you about the biggest mistakes I think you'll make, because it's certainly mistakes that we made at times and had to try and avoid other times. Okay, so let's talk about culture. This kind of sounds like ridiculously obvious. To survive AI, you need a culture. Don't resist AI. And no one says that they're like gonna resist AI. It's not a thing. However, look at the list of things that you have to change. You have to change everything. You have to re-found your company. If you have a SaaS company and it predates AI and it's not an AI-native company, in other words, you started the company designing an organization for the AI world, you have to change everything. Your org, your product, your roadmap, how you build, the metrics, how you measure success, how your sales team works, how you go to market, pricing, how you price, what does successful pricing look like, what does successful revenue growth look like, seats, outcomes, those two things together. You have to change everything. Everything. And it's really hard to do it. And culture is deep. I'm gonna talk about it in a sec. It's deep. It's really hard to to get people to change things that they do. Humans hate change. Humans, we all hate change. We're not fans of change, so we're averse to it. And it shows up in loads of different ways. If you're a deep AI company, your product looks like this. If you're truly a deep AI company, of course you have an AI layer, but you also have an app layer and a model layer. So this AI layer, this diagram is our RAG system, and this is a gross simplification of the RAG system. We believe we've one of the best RAG systems in the world. It's been 3 years in the making from a very talented team, and it's complicated. I do not understand the depths of that RAG system at all. It's complicated, but it's necessarily complicated. But you also build an app on top of that. There's how do people access the RAG system? Access access your AI, and you need a model layer. We have started building custom models, training custom models. I'll kind of show you how they work in a second. But if you're a true AI company, you're doing all three of these things. You're building an AI layer and it's complicated. You're building a model layer, you're looking at how you train your own models, building custom models, looking at different parts of the kind of space that you solve for in building models, target bits and pieces of it. May or may not work. You're experimenting a lot. And even app layer on top of that. So to get to this place, you need to change the culture. And the problem was that no one's gonna say no, but they'll say all these other things. They'll say, "Not now." They'll say, "Well, you know, we should start that next quarter. We've a lot going on. Our customers aren't ready for that. Our customers don't really— the AI thing, they're not sure about it. Last week I updated the board, I gave them these new forecasts. If we change, if we do the AI thing, it's risky. Maybe we don't meet our number. They're gonna give me pressure. We'll look bad. It feels risky. Do you know what we'll do? We'll dip our toes. We'll like do a bit over here and a bit over here and a project or two, and we'll prove it out and see if it works, and then we'll change. Then we'll transform. So step one on this journey is you have to tell the company that it's happening and it's not a choice. You're changing the company. You're refounding the company. You're changing everything about the company, and it's not a choice. And the reason is because AI is inevitable. To say like AI is the next technology cycle isn't even a new thing to say. We're— it's age— that happened ages ago. We're like 3 years in now. AI is inevitable. Whatever you loved about SaaS and the last decade of mobile and social and and the internet before that, it's gone. It's gone already. So you have to change. You've no choice. And you can do, you can kind of like, I feel like everyone, most SaaS companies at least, SaaS companies at scale are at this crossroads. They have two paths to go down. Opportunity or whatever the opposite of that is, death. If you nail it, if you truly change, you'll become a category leader like Finn is. You will grow to like $100 million in ARR very quickly. And Finn's on that path, very close to it. You'll be on the front page of Forbes if you care about that. You'll have all the series, you raise all the money that you want, no problem. But if you don't do it and if you fail, and deciding not to do it I think is failure, you'll slide into irrelevance slowly but surely. New competitors will eat your lunch. You'll have down rounds, all your best people will quit because they'll go work at the true AI companies, and ultimately the company will die. And this is a really important— we're at this crossroads, a lot of companies, and it's really important that people pick the right lane. The AI revolution cannot be negotiated. It's not something that you negotiate with people and you negotiate with staff, you negotiate with your peers. Remember all those people that will not say no but but they'll say to other things. And here's the mistake people make when they go into this journey. They say, they make the easy, fun choices, building AI features and fun things like that, trying out different models, you know, talking to customers about it. Like, that's all the fun stuff. But they'll avoid all the hard, messy ones. Like, a third of your company, you need to part ways with them. They're not fit for this new world. They don't want to build the same way you want to build. They don't like fast, crazy fast pace. They don't like chaos. It's gonna feel chaotic. They're not meeting the moment. They're just gonna, you know, slow things down, get in the way. That's the hard part. That's the real hard part of this. And then you'll convince yourself you've done enough. You'll do the fun stuff. You won't do the hard stuff, and you'll convince yourself that you've done enough. And I see this all the time. A lot of people, it's just because it's hard and we don't like change. So what we did at Intercom was we went way too far, because the only way to know if you've gone far enough is to go too far. The only way to know where there's a boundary is to cross the boundary. So for example, on paper, I'm our Chief Product Officer. I've always run the product team at Intercom. 6 months ago, I took over 2/3 of marketing, so I'm now actually running marketing much more than I'm running product, at least for the last 6 months. I'm running brand, product marketing, content marketing, advocacy. Why is a product leader running marketing is an entirely different 30-minute talk. The first thing I did is I blew the entire thing up. Everything's gone. Teams, roadmaps, marketing calendar— delete the marketing calendar. People loved Coda. There was spreadsheets, planning about plans. Like loads of stuff. The only way I knew how to build a marketing org that was fit for purpose for this age is to build it from scratch, blow it all up, and the right things will grow back again. So we have a calendar now again, but it's way simpler than the calendar we had before. And if I tried to iterate from the pace, the place we were, it wouldn't be the same. It just wouldn't be the same. So you have to go too far, and people hate this. People don't like this, and you have is to kind of upset people and explain why you're doing it and bring them on the journey. And for those people who just aren't up for that, you have to part ways with them. Smile, shake hands, it's amicable, it's all good. They'll all go on to find different things to do that they're more happy with, but you have to do that. I'm always asking myself this question: What would a brand new startup incorporated today do here? I'm like, literally last week we're doing some org design stuff and I'm asking myself, What would a new company do? Would they have product marketers? And would they have content marketers? Or is that actually kind of the same job a lot of the time? Maybe we should have a full-stack marketer. So, I'm constantly asking these questions. What do product managers do? What do product designers do? Would a new company have both of those roles too? So, product designer, product manager, product marketer, content marketer, all that like chained together. I don't think they'd do that. So, what would they do? So, we're asking this question all the time. And I mentioned that culture runs really deep. Culture runs really deep. So you have to be wary of the people you need to keep convincing because they're gonna be a liability. You have to be wary of those people. Just not everyone's up for this. It's a fight to transform from a SaaS company to an AI company. To see the success we've had with Finn, you know, by the way, it gets better. You know, the hard part gets easier and then it gets exciting. Today with Finn, it's, it's, we're in a really good place. But to get to that place was really hard and it was tricky. You have to delete all the parts of the process that are not part of the future. And there's a ton of these, I'm gonna show you some in a second. You have to— there's loads of processes at Intercom around building software that I designed and I was a part of designing and I was very proud of them and loved them. I've given talks about them. They don't exist anymore. You have to delete these things. They're not a part of the future. And you need to ask yourself, of why you exist in a post-AI world. Most SaaS companies sold seats. These are seats that humans sit in. Those humans use some kind of GUI, some CRUD-type tasks to achieve some outcome. None of these things make sense anymore, right? None of these things make sense. There's no seats in a post-AI world, or at least the way seats are orchestrated is very, very different. People don't use GUIs necessarily. A lot of the product invisible and a lot of the quality of the product is invisible. And ultimately the company with the best outcome built on the best AI, built on the best RAG system, built using custom models, they're the ones that are gonna win. And you gotta do that really fast. You gotta build this world really fast. That's why we ripped it up, took the pain because you have to do it fast. We could see new competitors entering the market. We were worried and we were paranoid about it and we knew we had to do it really quickly. One thing you gotta do early is understand how ambitious you're gonna be. So this is like really kind of simple. Value goes up, but all, you know, ambition and risk go up too. So you can decide that what you're gonna do with your SaaS company is change it a bit. We're gonna help humans do a single task in their job. That's pretty straightforward to do. Kind of low risk, but kind of low ambition. The value is like, you know, reasonably good, not bad. Or you might decide we're going to tackle a whole workflow. That's way harder. Now you've got all the steps in the workflow, but the ambition is greater. Maybe you can charge more for it. Maybe the company will be more successful in the long run. Or you can decide we're going to do the whole thing. We're going to be extremely ambitious. We're going to disrupt our entire business that we run today. We're going to do the work of the entire team. Today we have all these people who who pay for us with seats to use our SaaS product, and we're gonna disrupt the entire thing, risk all our revenue, or at least risk the revenue growth by trying to change it into a new AI outcomes-first world. So you gotta decide where you're gonna be on this spectrum. And to do that, you really have to understand the job you're solving and like deeply, deeply understand it. So for example, these are like two examples of kind of classic SaaS products. We've like a meetings and project management product. We've an expense management product. These again are like GUIs that, you know, CRUD tasks. People are like reading, reading things and writing back to systems. And it's very workflow focused. Like a lot of SaaS products look like this. But if you wanna solve this problem in an AI-first way really well, you have to break down every single bit of these. And understand them really deeply because what you're gonna end up doing is pointing AI at all of these discrete things. 'Cause getting AI and getting a RAG system, uh, to accurately do these things is, um, easier when it's discrete. You can experiment much faster, you can solve bits of the problem independently and then kind of chain them together. So you need to understand, uh, each of these steps and you need to understand if you can reliably solve it. And the reason this matters is because all of these things compound. If you're really, really good at each of these steps, really good, like here we're really good, very reliable, you still have to multiply all these together. And by the time you're finished the entire workflow, especially the workflow of an entire team, you're down into like the 90% success rate. And if you're not as good at it, you're down at the 80% success rate. And it's just way harder to build a product unless you really understand it. So that's what we did with Finn. This is again, like a, an illustration of some of the system, some of our AI system with Finn, where you can kind of see we have these like custom models, um, that pointed directly at steps along the journey, pointed directly at very specific things customer service and customer experience teams do. And that's how we're able to get such incredible results with Finn. This is like a lot of experimentation, days, weeks, and months of experiments. Constantly trying to find a new edge, constantly trying to find the, like, tiny little increases because the reliability compounds. Each single tiny incremental improvement in each of these steps adds up to something way better, adds up to the highest performing product, adds up to something people can trust, adds up to something that people will replace their humans in places with Fin. But they'll only do that if they can trust it. They'll only do that if it's, like, credibly the highest performing. So you got to attack these things very systematically, each step along the journey, and to do that you really have to understand them. Another thing you got to think about is marketing the product. There's a thing that we call the marketing overhang, and this is like very, very, very common. I think one of the hardest— I think marketing AI products right now is extremely difficult. Evaluating is difficult, but marketing is really, really hard. Marketing Fin is really hard. Because a lot of companies claim that they have a thing too, and their thing works just as well as our thing, and they have demos. And so you can kind of see here, like, they're building things and they work, and it kind of works a bit, and they can demo it. But a demo isn't a product. A demo isn't a product that works at scale. And so what happens is people market, go to market, with this amazing product idea that they've real demos for. They'll do product launches of the demos. We have a rule at Intercom that we will not do a product launch that is not real. We're not gonna do a product launch that isn't a real demonstration of the product, and we know that it's gonna work at scale, which kind of inhibits us a bit and it slows down marketing sometimes. The reason is the product might not work. And when you build AI products, you really, really don't know if they're gonna work. You need to have them in production, experimentation all the time, trying to find edges, trying to find improvements in performance. And so you get this problem where you've promised the world— maybe you've already signed up customers— you've promised the world this incredible product, and you've got one really good demo, but you've got a huge delta between what's actually possible in reality when it really goes live and the promise you delivered. This is a great example. Of this. Apple Intelligence. Amazing. In June 2024, Apple are finally here. They finally entered the AI race. They're going to have this incredible product, and maybe they're going to win. They've got your device, like, you know, they've got the OS layer, they've got loads of advantages here, customer data, all sorts of stuff. It's now next spring 2026 where we'll see Apple Intelligence finally, and then we'll see, right? This is the marketing overhang. You have to be really careful that what you're building, demoing, is real, and it really works in production at scale. OK, the next thing is that how you build software is very different. So how you market it is different, and how you build it's very different. Today's software is built with empirical evaluation. So running these evals, running, having a robust evaluation framework, and tweaking the architecture, yeah, using these experiments experiments constantly to find these, like, little edges in performance, comparing the old, comparing the new. We've run hundreds and hundreds and hundreds of experiments with Finn, and, like, obviously not all of them work. Uh, I don't know, do most fail? But certainly many of them fail. Uh, or we see an improvement in one part of the system, but back to that lit— like, kind of line of the workflows, another part of the system isn't as good, and we've kind of degraded performance there. It's gonna have to think about this thing systematically. And then the majority of product improvements are invisible and unpredictable. So, in the SaaS world, this is what building great SaaS products look like. Most of it you can see, you're building like really great user experiences, really great UI to help people do those jobs well. And the CRUD backend, you know, tiny, kind of invisible, not a big deal. There's no kind of tech technology problem here. You know, you can do it if you have like dedication, resources, you know you can build the thing you want to build. And actually, the hardest parts are in the user experience— making it easy to use, easy to learn, all that kind of stuff. AI products don't look like this at all. It's actually the inverse. Building the UI is the easy part. That's the small part. The hard part is building great AI infrastructure, building products that actually truly work at scale. This is an entirely different world, and my background's in design, originally as a designer, and the left-hand side here was like my happy place. You know, this is an awesome world. And I had a reckoning, a personal reckoning, weeks of anxiety. What happens in the new world? You know, the thing I'm good at is like the top part, and that's really small now. And things like infrastructure, I do not, I was like, I'm way out of my wheelhouse if we get into like infrastructural layers. But I had to learn it. I had to get better at it. I had to kind of face reality and say, "Okay, well, I've got to get better at this kind of stuff." We had these principles at Intercom that evolved over years, and they made us really good at building SaaS products. They made us really good at it. We had like kind of 6 core principles. We had like design principles. And what we would do is we would teach new people these principles. So, a designer joins a company, an engineer joins the company, we would say, "This is how we build software." to come. This is why we're good at it, because we follow these principles. We have a little process for each principle, like think big, start small, ship early, and there's like ways in which we would do that systematically, and we got really good at it. But these principles were designed for the SaaS era. They were designed to build a SaaS product. We had to abandon all this shit. It pained me. It still pains me. Looking at this. It pained me to do it. You know, Dez, the product leaders, the engineering leaders, it was really, really hard. People did not like it. People like me didn't like it. I love those principles. You know, I helped make them. They're part of— I like looking at them on the wall. Doesn't matter. You know, looking at principles on the wall isn't gonna build a great business. And people were disempowered. They're like, this is the way I know how to build software. I'm good at building software this way. I don't know how to build software the new way, and it sounds way too technical and way too hard for me. It's not how I think about things. So there's an old way and a new way to build software. The old way, you pick a job to be done, you listen to customers. If you're very good at it, you get really good at customer insight, getting true, true customer problems down into the depths. You design a solution, you build, and you ship. And in this world, execution is certain. The technology stack is pretty stable. You know, design is one of the hardest parts of the process, not the engineering side. Users could talk about their problems. They were very easy to talk about, like all the ways in which their workflows didn't work and their company and their org design. It was a stable world. Technology was stable, but organizations were stable. How organizations are structured, how they think, all stable. There was never a question of like, Can this thing actually work? Is this even possible? There's an entirely new way to build software, and this is what we've changed. Again, it's painful to go through this cuz you've gotta delete the old way and it's risky. It feels very risky. There's an entirely new way to build software. You start by asking, what does AI even make possible? Like, AI is a magical technology. I think people like, we're kind of, we've kind of lost ourselves in it. There's so much hype, so much news, No one can keep up. There's all sorts of mad stuff happening. Is it like hyped, underhyped, overhyped? Is there a bubble? Is there not? Like no one knows. But if you step back, we have this like crazy magic technology. Like it's magical when you actually step back and think about AI and think about the fact it's only 3 years old. It's kind of wondrous. You know, I think if you went back 5 years ago and looked at us today and the things we can do with ChatGPT, with Cloud, like with all the coding tools. I cannot code. I never could code in my life, but I can build things now. Like, it's an entirely— it's a magical thing, and we lose that. So there's a new way to build it. You start with, what does it make possible? What might the product be? And can we build it reliably? I don't know. Let's find out. Let's prototype. Let's try stuff. You then later build the kind of CRUD part, the UX, the UI. You build that later, you ship, and then you look at scale. You look at, like, does it work in the real world at scale for thousands and thousands of companies? So there's way more possibilities, but it's way more ambiguous. It's a way— it's more chaotic. Execution's uncertain. You got to throw stuff away sometimes, start again, and you learn the most when it's live. There's also, like, these new muscles to grow. These are blog posts. We have a research blog. Again, like, our machine learning team has been around at Intercom for a long time. It's now way bigger than it was. It's now called the AI group. We rebranded it, and the amazing team, and I've learned so much from these things, but this is like way out of my wheelhouse, way out of our product team's wheelhouse. I think way out of some of the engineers' wheelhouse too. There's like empirical evaluation. It's way more scientific. There's like scientific rigor required. You need to really understand causality and how things change other things. There's a whole new world to learn. And design has changed as well. Design, when we used to build software, design was kind of the hardest part. Design was often the bottleneck. Design was often the difference between a great product and a good product. It's because it was better designed. This is what we'd do. We would sketch out the wireframe. We'd sketch some ideas. We'd pick one. We'd iterate that a bit, and then we'd ship. And the reason we work this way is because design is expensive. This is the new world. You don't work like that anymore. Design is now cheap, and you can do loads of stuff quickly. PMs can design stuff. Engineers can design stuff. People can prototype stuff. You can give these systems, like, principles and rules and guidelines and product design rules of thumb, and you can just build stuff really, really quickly. You can explore really, really fast. And so the possibilities are much, much more open. And that, of course, then leads to changes within your company. Well, if design was expensive and now it's cheap, what does that mean for designers? If PMs can design, do PMs also PM still? Like, how does this— how should all this work? So there's changes we've made, big changes we've made. The old way was, should designers code? If you're a designer like me, Oh my God, this question. Like, for years and years, existential, like, navel-gazing about whether designers should code. Yes, designers should code. In fact, they do, and I'm gonna show you that in a second. Designers should code, of course. Now they can, and they should. What about engineering velocity? Like, eking out incremental improvements. We are 2x-ing our engineering velocity. Hackathons, playing around with stuff, versus saying, AI engineering is now the default. Here's examples of this. This is Domingo from our product design team. Just talking about him kind of shipping some frontend changes. Every single designer in Intercom has shipped code to production. Like a year ago, maybe 18 months ago, no designer at Intercom had shipped code to production. We went from zero to every single designer shipping code to production because we said we're gonna do it. Remember, AI is not negotiable. We said, every designer is now shipping code to production. It is now part of your job. If you don't like that, we had a great run, but we should part ways, and you should go somewhere where designers don't have to ship code. And we'll hire designers who love the idea of shipping code. This is really empowering for a lot of people, because what it means is that all of these designers can now fix all the frontend bugs that they hated for so long. Right? The old world was like annoying engineers, please fix that stupid 2-pixel thing, right? Now designers can change all of this stuff. So we started small, little changes like that, like little UI bugs, and it's gonna grow. And the design team is loving this. If you go on LinkedIn, you'll see all our product design team shipping these code changes to code and loving it. We've also doubled engineering productivity, or on the path to doubling engineering productivity. And the only way we did it is the same as design. We said, "It's happening. We're doing it. It's not negotiable." If you like that, you're gonna love this. And if you don't like it, we're probably not the right match for each other anymore. It's, you gotta pick a hard-to-hack metric. So, 2x-ing productivity, you have to enable the team internally. You gotta explain why we're doing it. You gotta persuade people. You gotta get people on side. Explain why this is valuable to the company. Explain why this is amazing for their careers. You know, since Finn's success, one of the hard parts being a leader at a company that's now succeeding is everyone wants to hire all your team. So it's really good for your career if you're a designer who can code. It's really good for your career if you're an engineer who says, "We 2x productivity. I can tell you how to do it." So we're publishing all this stuff externally too. We have this Ideas blog, and you'll see a lot of the design team on the Ideas blog posting kind of frameworks for how they're doing this stuff. Okay, next is GTM. So how do you do go-to-market with an AI product? Again, there's a ton in this. I'm gonna just cover some small things. Customers do not know how to buy AI, especially if your customers are not technologists like everyone in this room is. They don't know what, how it works. It's a black box, it's a mystery. It's also scary. You know, we living and breathing this thing can't keep up. So like, God help anyone outside of technology. And most of our customers are people outside of technology for the most part. So you need to help people understand AI. You need to really invest in education. Customers don't know how to evaluate AI products. They don't know how to run good evaluations. Oftentimes we look at Finn, we lose a deal. There was a head-to-head. We look at the eval and you're like, shit, they didn't set it up right. You know, like they just, if they'd learned a bit more, maybe we still would've lost, but I think we would've done way better if they'd set up the evaluation properly. You gotta market your strongest differentiator, which might not be like this big screenshot UI. And then you gotta really invest and happy customers and successful customers. The people are just looking out, people are crying out for education. They're looking for like, show me, I don't know what to do. Show me people like me who've done it already and I'll learn from them. So customer advocacy is just enormously important. The product marketing has changed too. Like all these products look the same. A lot of AI products do look the same. A lot of them are now AI agents. They all kind of look like some kind of chat interface. They all do different things, but they all kind of look the same. It's hard to tell them apart. Look at 5 competitors, they all kind of look the same. So actually, the pitch is actually much more about infrastructure than it is about features. You know, we talk at length about Finn's RAG system, at length about this empirical evidence, our scientific rigor, why Finn is better at scale across these like hundreds and thousands of customers. That's like a different way to market to market a product. That's not how typically SaaS products are marketed. And the buyer has changed. So, in the old world, this is our example from Finn, but I think this is like fairly universally true. In the old world, what happened with us was we had a customer service product. So, we have a customer service leader buying the product. We have an ops leader, but they're not the buyer. They have influence. We need to kind of get them on side, talk to them, market to them. And the team, the team are like good vibes, you know, good for loyalty. They'll tell the boss, which one to pick, but ultimately there's a single buyer. Doesn't work like that anymore. There's a, there isn't a single buyer anymore. So for example, with Finn, we have, yes, the customer service leader and they're still heavily, you know, heavily influential. They have a lot of clout, but they don't make the decision. And the reason is because there's two other people involved. There's typically a C-level executive 'cause their job is AI transformation. They've been told, Hey, my job as a C-level executive is AI transformation in our company. We're changing how we work. We're gonna start with customer service. So I'm sorry, but I'm gonna pick the tool actually. I'm gonna decide, you know, you can help me inform me, but you know, and then there's the AI person. There's like, you know, IT buyer here. It's often an AI kind of someone who's very fluent in AI. Maybe they run an AI team and they're gonna evaluate if the thing is actually good, right? Again, these AI products are invisible. It's hard to, remember the iceberg, hard to know if it's good or not good. So we have a whole different kind of setup here. It needs to get good at marketing to these people, selling to these people. A CEO and a customer service leader live in different universes. They don't go to the same events, they don't like the same things, they don't read the same things, they operate at different Zoom levels. So I'm doing like dinners with like CEOs, I'm going to like trade shows for the customer service leader. We're doing all sorts of types of activities because they're not like the Venn diagram here isn't even overlapping at times. So if you do this really well and you're really successful at it, you get a different problem. And that's where we're at today. We now have two companies. We have Intercom, the SaaS company, the SaaS product, uh, seats. And we have Fin, the AI product, uh, outcomes And with Intercom, we have a really easy product domain to understand, easy to roadmap, easy to talk to customers about it. They've done this thing for 10 years. They know how to talk to us about it. They know what features are missing. We've got predictable metrics. We've got history for these metrics for a decade. We've got like a really clear differentiator. We know what not to break. We can kind of look at like different types of activities. But Fin is a different beast entirely, a new product domain. Customer service teams, customer experience teams, if you sell to sales teams, if you sell to marketing teams, everything's changing in their world and it's chaotic. They don't know how to even talk to you about what they need. They don't know what they need. Everything is changing so fast, no one really knows what the future looks like, so it's really hard to articulate those jobs to be done. People actually don't really know what the job to be done is, so it's really important to help them The metrics are really unpredictable. What will this outcome's revenue do to the seats revenue? And actually, we think about that. What's it— is it— what is this trajectory? Good? Is it not good? Jeez, I'm looking at like all these Lovables of the world, you know. Waze were just here, like, holy shit, look at their growth chart. Should we have a growth chart like that, or is ours okay? Ours is really good, but is it, you know— Finn's growing 300% year over year. Is that bad, right? So everything changes. The differentiator's often performance, evaluations, head-to-heads, and you gotta do startup stuff. We kind of say, it's a bit of a kind of cliché maybe to say like you're a startup reborn, and we're trying to run Intercom like a startup. Intercom's 1,200 people, but we are trying to run it like a startup, which means it's a bit chaotic. But I think a bit of chaos is actually good. It means we move fast, we ship things. And then we end up in this situation, which is like Intercom, Fin, we're trying to work out the brand, what's the brand relationship to these two things? But I'll take it. It's, you know, a good— it's a good problem to have. Okay, so let me finish up with just mistakes I think you might make along the way. We've made some of these. I've seen other people that I talk to make them, and I'm telling you them, so I hope you don't make them. One is you won't reimagine your product. You'll add AI to your product. That is not reimagining your product. Fin is totally different to Intercom. It's a completely different product. The two of them are like, they've almost nothing in common in some ways. It's a completely different product. And like, yes, Fin is, Fin usage is eating Intercom usage at times. They're totally different products. You design them totally differently. Think about them totally differently. So you won't reimagine your product. And again, you do the easy stuff. You add a bit of AI to it. It's not the same thing. You won't make self-harming decisions to win. You'll protect revenue. You don't want to piss off the board. You know, you don't piss off your sales leader. Like, there's loads of reasons and loads of places where you won't say, do you know what, we're gonna take a 10% revenue hit to do the right thing for the long term, or even the medium term. These self-harming decisions are critical. I remember I said at the start, it's brutal. It's a really hard transition. It's really hard. It doesn't feel really painful painful, you're not deep enough. You're not going for it enough. You gotta refind your company to fight properly. You need people who are gonna fight for success, fight for these amazing outcomes. They're there to be had, by the way. If we can do it, any SaaS company can do it. There's nothing special about us. There's nothing unique about us. We just decided to do it. We decided to refind the company and we decided to fight for it. And it was hard and there was tension. There was a lot of disagreements and people fighting fighting with each other. Positively fighting, you know, not actually fighting. So you'll avoid that, 'cause people hate change. And lastly, these are the things you will do. You'll dilute down the vision. You know, you're like, "Oh, you know, we'll just kind of dial it down a little bit. We have a vision, it's kind of new stuff. We'll kind of dilute it down." Or we'll delay. "Hey, you know what? Q1, we did have a great Q1, so we'll do it in Q2." You'll convince yourself you've done enough 'cause you did the easy stuff, not the hard stuff. Big company habits and slowness will creep in. You listen too much to customers who say no to AI. We had lots of this. We had loads of people saying to us, we're not ready for AI. We don't wanna do AI. Actually, we're gonna differentiate ourselves on human service forevermore. And now they use Fin, right? So you have to really think about who you're listening to and what they're saying and why they're saying it. 'Cause all of those businesses that you sell to, they don't like change either. And they don't wanna transform either 'cause it's really hard to do. And then you'll make these mistakes and deny it to yourself or deny it to your team. You gotta be really honest. We've had the most honest soul-searching conversations in our exec team. We still do today. You know, every day in our exec team, there's like some version of some existential question that might mean we do or don't succeed greatly. Or, you know, it's, you really have to like look at each other in the eye, build these like deep relations with your peers and colleagues. So that's how we did it. That's what we did. You can learn more on our Ideas blog. You can learn more about our research, our AI research, on our Research blog. If you want these slides, I'm very happy to share them with people. Just email me and then we'll send you on the slides. And we have 5 minutes here. I think I'm happy to take questions from the audience, but thanks.

Amazing presentation. I'm an Intercom customer. Love it. And I also use Fenn. How do you price Fenn? I don't know how to— how did you evolve it? What are best practices? How should we be thinking of pricing agents?

Yeah, this is a fascinating question. I'll tell you what we did initially and then what we're wrestling with now. So initially, so I think we were the first to market resolution-based, outcome-based pricing. So Finn was originally, I'm trying to remember the details. It was basically a dollar. We priced it at cost. So we said, hey, the wrong way to price Finn is seats. That's just, that doesn't make any sense. Like again, does rethink, doesn't make any sense. So what does make sense? Well, if Finn resolves a customer question, that's great great for the company, and they'll definitely pay for that outcome, and it's great for us. And if they're not paying, it's 'cause it's not working, and if it's not working, we've a shit product, right? So, like, there's just this really nice symbiotic, hey, it's good for you, it's good for us, it's kind of win-win. And we went out and priced it at cost. We were like, hey, how will we, what price should it be? I don't know, no one's ever done this before. Is it $1? Is it $10? Is it $100? Like, I don't know what it is. A customer support rep costs like $20 or $40 an hour. Like, how do you think about it? So anyway, we end up with 99 cents, and that's been like generally really successful. So like a lot of outcome-based pricing, it really works. What gets tricky is we're now in a situation where we've much more than Finn answering questions. We now, for example, have an AI Insights product, and this AI Insights product looks at like the thousands and sometimes millions of Finn conversations and drives insights, like product issues people are having. People are looking for like, what's the root cause of these questions? We can resolve the questions by fixing the product. So we've this like AI Insights product, it's given us customer insight, product insight, et cetera. How do you price that? And the, like the calls we're making to build those insights cost a shit ton more than 99 cents, right? Yeah, so we don't know what the answer is, honestly. I don't know where it'll end up, but I think we're gonna have for sure outcome-based pricing. It's just a really nice thing. It's just a really obvious, easy, nice thing. And I think we're gonna have something else. Is it add-ons? Is it like bulk calls? Do you just price on the kind of like LLM usage you have? Do you try and like, Take the token usage and move it up the stack to the customer. Like, I don't know the answer. We're literally trying to work that stuff right, right now. But, um, for sure, I think we're not gonna revert back to seats. I, I don't think, you know, it doesn't make sense. I don't think so. Thanks. Yeah, no problem. Uh, anyone else have a question? Yep. Is the AI model behind Finn a homegrown model or an And, and if it is, is it, why is it better than ChatGPT, for example? Yeah, so, um, originally Fin used, originally Fin, we partnered with OpenAI. We then moved to Claude. Um, and for a lot of things, uh, especially the more complicated queries, uh, Claude is behind them. Uh, it's really good. Now the differences between these models is very small, like You know, Claude tests, we do so much testing. Like Claude, Claude is better than, than, you know, OpenAI ChatGPT, like, you know, by small margins. But the open source models are really close too. You know, the open source models are really good. There's very small delta between these two things. And so they're obviously, obviously doing the same thing we are, trying to find this edge. But these are general purpose models. Like Claude wasn't trained for customer service. You know, the open source models weren't trained for customer service. Same for any, same for any domain like Harvey, you know, a product for lawyers, legal AI product. You know, Claude wasn't trained for law. These are general purpose models. And so we use them as a foundation and then we build this RAG system. We build a lot of sophistication on top. You know, training Finn very specifically for customer service and we've huge gains. Huge, huge gains built up literally week by week. There's a chart I didn't show which is Finn's resolution rate. Finn's average resolution rate is now at about 65%. We have this crazy chart. I've never seen a chart like it in my whole career. It goes up 1% every month, or is it every week? Every week, I think. Anyway, every single 1%, every single time, and that's because Finn's It's a constant experimentation we're doing, constant A/B testing, you know, changing little parts of the models and how they work. So I think specialized models outperform general purpose models. And then we started to build our own custom models. We said, actually, what if we take an open source model, just the basics, and then train it ourselves on customer service, train it on, we've like hundreds of millions of customer conversations. What if we trained it on that? Does it perform better? So we started doing internal testing there. So I think specialized models directed at very specific things will outperform general purpose models. And so that's, that's basically what we've been doing. Uh, any more questions? I think we're outta time. I, what, maybe one more question?

Yep. A quick one. How do you balance your investments between this revolutionary path where you reinvent the whole thing, make Finn, and then between the AI that you put into Intercom? Do you still stay competitive on Intercom? To come and try to, to invent new things on that, or are you keeping it completely as a legacy model, not putting AI into that?

Yeah, there, there's so much. I, there's, but I'll stand at the bottom after this. I'll go to the back room. People have questions, they can kind of ask me down there. Um, there's so much to this, like I haven't even talked about. Um, for example, we moved all of R&D onto Fin. Like we started at 70/30. We said, okay, first of all, it's like 100% of people work on Intercom. And we're like, hey, if we're gonna bet the company on Finn, we moved 70% of R&D onto Finn. That's hundreds and hundreds of people, like 300, 250, 300 people. Again, like, if you work in the company, people think this is batshit crazy. What? Like, that's insane. So now Intercom's on maintenance mode, effectively. We're like building features for customers that are important, little gaps here and there, but for the most part, we're just kind of a little bit on maintenance mode. Then we went to 80/20. Says, hey, with efficiencies, we can get the Intercom part down to 20 and have 80% working on Fin. And now, and so that's what helped us accelerate Fin's trajectory, get it to market, get this like deep, big, broad product, very successful, only because we did that. We fast-tracked everything. But now we're moving back a little bit. Actually, we're moving back by hiring more people. People. So we're not really, you know, but we're now, we're hiring more people on to work on Fin, our Intercom. Loads of our customers have loads of Intercom feature requests still. Our crazy thing is that when we started selling Fin by itself, that was another huge risky thing we did. We decided we're going to build Fin to work on Zendesk. Loads of customers were like, we'd love Fin, there's no way we can swap Zendesk because like we have a 10-year instance of Zendesk, there's no way we can swap that out. And we said, okay, well like we'll just bin Fin to work on Zendesk. So Finn works in Zendesk, Finn works in Salesforce, Finn works in our competitors' platforms. Again, it's like a not a normal thing to do, and I should go into kind of bet it all. But now we've loads of Finn-only customers who've decided actually we want the helpdesk too. We will switch. And so now we need to like reinvest there a little bit. But we bet the entire product engineering team on Finn. It's the only way we could do it. All right. I think I have to leave. I think we're done. But yeah, thank you.