I think there's an unspoken covenant that as a founder, you go down with the ship for better or for worse. It's changed a bit over the last year, and I think it's a bit disappointing to be honest. I think there's some retreats to your point that often, there are actually a lot of really valuable pieces that get left behind. In AI code, to be truly honest, if there was zero progress, the world would still be entirely different. - This is 20VC with me, Harry Steppings. Now you wanna know how real time this episode is? It was recorded about 12 hours before you're listening to it. It was recorded late, late on Thursday night, and yes, 20VC team worked through the night to edit it and turn it around. That is commitment for you. The only thing that anyone will talk about this week is cognition by windsurf. What happened? Did the team desk root over? What happens from here? How does it fit into cognition's grandmaster plan? What does it mean for the space in terms of who wins? Curse of versus cognition versus Claude Code. We unpack it all today with Scott Wu, co-founder and CEO at cognition. But before we dive into the show today, Harvard Management Company is constantly seeking out the next generation of great investors and entrepreneurs. HMC has managed Harvard University's endowment for nearly 50 years and was one of the first institutional investors in venture capital. Their experience and long-term investment horizon makes them ideal partners to get world-changing ideas on a path to viability and success. They work as a true partner, providing insightful perspectives to help managers succeed. I personally have had the pleasure of working with the HMC team and can say that they are truly exceptional partners and savvy investors. So whether you're launching your first fund or your fifth, HMC welcomes the opportunity to partner with both developing and established managers. Have an idea you wanna share with the team? Just send it over to
[email protected]. And after securing top tier partnerships, like those with Harvard, the next step is building trust with customers and Secureframe makes that simple and seamless. Secureframe empowers businesses to build trust with customers by simplifying information security and compliance through AI and automation. Thousands of fast-grown businesses including NASDAQ, Angel List, Doodle and Coda, trust Secureframe to expedite their compliance journey for global security and privacy standards, such as SOC2 and ISO27001, CMMC, NIST standards and more. Bat by top tier investors and corporations like Google and Cliner Perkins, the company is among Forbes' list of the top 100 startup employers for 2024. G2's best software awards for higher satisfaction products and a recipient of the 2024 Cybersecurity Excellence Awards, something I definitely never got in school myself. Learn more today at Secureframe.com. - You have now arrived at your destination. - Scott, I am so excited for this dude. I was literally just telling you, it is a cool prep when you get to cool up like Vinod Kostler, Joe Lonstale and you call him and you're like, "Hey, what should I ask?" And they're just like, "Ah, I love this guy." So thank you so much for joining me. - Yeah, thanks for having me. It's been a crazy few days for us all. - It's been pretty insane. I just want to start present day and work our way around. How did you learn about the potential ability for you to acquire Windsorff? And how did that opportunity arise? - Yeah, we found out honestly on Friday, at the same time everyone else did that all of this was happening and that this was the split and that it was Google and here's what happens next and here's what happens with the team. And we were kind of talking about it, which is on the dev inside, I think in a few ways, it seems to be a pretty natural fit. One, because I think obviously there was an incredible team left behind and if anything, I think, well, we have at cognition, I would say, is especially focused on the core engineering and product team. Whereas obviously, I think Windsorff has built an amazing go-to-market team, marketing team, finance operations. And then similarly in terms of the products, we found that actually there was a very naturally complimentary lean. And so we reached out cold Friday evening and our first conversation was Friday night. - Just take me to it then you reach out to Jeff and you're like, "Hey, figure something's going down. Should we jump on a cool?" - Yeah, yeah. And so I mean, we basically said, look, it seems like there's enough here that it's worth having the conversation and talking about it. We got on the call together. We were talking about the ways to partner and what things would look like. I think what we came to pretty quickly, I always think about these things in terms of what is the correct way that things should happen here? And I think from Windsorff's perspective, the way that we saw it was, you could talk about the product of the team and the business, right? And what is the right thing for each of those? And I think for the business, the thing that you have to call it is sooner or the better. You could do this and you could have like, months of diligence and everything. And that has apparently happened multiple times already in Windsorff's company history. And I think in this particular case, I think everyone is scrambling. All the customers want to know what's going on. The whole team is wants to know what's going on. Everyone in Silicon Valley is reaching out to them, saying, "Hey, I heard the news. You want to come interview here." And so we kind of just came to this cushion. Look, a month long is not the way to do this. I mean, even like a week long, it is Friday night now. We want to have something ready to go by Monday. - Oh, I'll pause 'cause it sounds like, yeah, but-- - And just like, entry, when you look at scale, it's like, you know, there's a husk left behind which kind of deteriorates. And you're seeing that deteriorate in real time. Why did you believe that this was inherently different? And that the husk was not a husk. It was actually a treasured test of value in a way that was within the scale. - Yeah, so there is definitely this conversation online, right about all the best researchers left and there's just nothing left behind. And I don't think so. It's basically how we felt about that. And we looked at it and I mean, it's an amazing product that a lot of people use. It's the entirety of the customer book as well. It's all of the code, a lot of the data and the proprietary IP. And then obviously it's a really incredible team. I think it's an interesting state in the sense that you could call it perhaps like a somewhat incomplete piece. I mean, as it turns out for us, I mean, we had the very nice compliment of what were the pieces that were needed, right? - Those pieces of the rationale that lead to the decision, how do you actually execute it? How does that work in process? - Yeah, yeah, for sure. So we came to them and we said, look, I think there's a very natural partnership here. I think the team pieces are very complimentary. I think there's a lot we can do on the product. The thing that just seems absolutely clear off from both of our sides is if we're gonna do something, we wanna be ready to do this and announce it by Monday morning. It kind of reminds me of like, the bank goes into receivership Friday night. You gotta have an answer by Sunday and by Monday morning, right? And I think we thought of this as a similar situation, which is, look, there's a ton of value here, but everyone's wondering what's going on. And we need to be able to give them a clear answer as soon as possible. We are gonna do this. We're gonna support the product. We're gonna make sure things, everything's going smoothly and running smoothly and so on. And so that's what we came to. And we gave them our honest take of like, here's what we think we can do together. We are not gonna have time to get to go like as deep into the diligence and details as we would like to, but I think at a high level, we at least are in a position to understand the business very well. Get to a verbal agreement on Saturday, hash out all the details and the terms and the legal on Sunday, get everything signed and done on Monday morning, and then we're ready to announce like, that's, there's no other way to make this work out. And Jeff and Graham and Kevin and the whole crew, honestly, like, I mean, it's a huge shout out to them because they really worked around the clock for their team. - Were they left in the lurch by this deal? - Yeah, so I think they had known about, I think they had known about the previous thing, maybe a few days in advance. It was not a time of advance notice. And I think they were honestly really thoughtful with how they went about it, which is, look, here are the options, we can operate as an independent company. We can go and raise a new round of venture capital that they're actually no investors, or we can find someone who we think we're gonna be the right partners to work with. But either way, it's gonna be, we're gonna have to do this quickly and we're gonna have to figure out what we think is the right long-term fit for it. I think most people see this as a really epic deal make from you, respectfully. And I mean, total heart tip and respect to you for it. The question that I have is, and actually, I think it was Ruchi at South Park, who asked this, but she said, was this a Google Blunder by leaving such a valuable asset up for grabs and not seeing the true value also in the asset? - Yeah, I think there's some retreats to your point that often there are actually a lot of really valuable pieces that get left behind. Given that you've been through this, deal structure, it's different and novel, do you think this becomes a new norm for how companies are acquired and how people join companies today? - I think there's an unspoken covenant that, as a founder, you go down with the ship for a better or for worse, it's changed a bit over the last year, and I think it's a bit disappointing, to be honest. - Is the talent war getting a little bit out of hand? When you look at matters hiring, Spree, and just how much might, is it getting out of hand? - Yeah, maybe crazy opinion, controversial opinion, but I actually think it's quite reasonable, (laughs) is my view. And the reason I say that is because AI, I think we are truly on the cuts, but the greatest technology shift in our lives, that's already clear. Even if you froze all the capabilities today, and then the question was just, if you said there were no new research breakthroughs, no new discoveries, you don't even have to bet on that. And you just said, all right, we're gonna grow products and figure out how to build the right experiences, and get those out to users, and meet users where they are, and all of these different verticals all over the world. I think it would still be the biggest thing. I think the biggest thing, in my mind, the biggest technology shift in our lives has been the internet. I think there are some others, which are obviously pretty massive as well, the mobile phone, the personal computer, and so on. But yeah, I think even if you froze all the capabilities today, I think AI already would be bigger than that. And the thing that's crazy about AI, and by the way,
I don't think it's good for you, I think it's gonna keep moving even faster. But the crazy thing about AI is, there really are so few people that are really just determining the trajectory of AI. But I think in aggregate, I think the view is right, which is, there's so few people that have to. - Can I interrupt you? - I'm so sorry, I'm totally agree in terms of what. Is it like a hundred? Is it like 10,000? Just what quantum is it? - It's a great question. Somewhere in between those two is right. There are at least a hundred folks that really matter in making a lot of difference. There's certainly less than 10,000, and probably a good bit less than that, if I had to guess. - Okay, so we've got like a hundred to 1,000 is kind of a range there. Do you need those people, unless you're working on the most cutting edge frontier models? If you're in the application layer, which is very hard, I'm not done it, but like do you need them if you're in the application layer? - I think the level of talent and the fierceness of competition is, in many of these is actually like greater even in the application layer, which is crazy to say because I think the competition in the foundation lab is extremely strong. - When you think like holy shit, anything can happen style, which it is today. One of the like, oh, shits that could happen, that's very obvious to I think me and everyone is like, the dependency now around anthropic. You know, obviously, when I had Varun on the show, they were very much reliant on them then, and then they got cut off when the open air ideal was done. Is there reliance on anthropic greater than it's ever been? - Yeah, look, we work very closely with anthropic. I think anthropic is a great company, and I think the foundation labs in general are great companies. People always ask this question about, where does the value accruing AI? Is it chips? Is it semiconductors? Is it foundation models? Is the application layer is the infrared? I think the boring but true answer is, it accrues wherever you are able to establish real differentiation in your space. And I think there will be a lot of spaces for which that's the case. I think the foundation labs as businesses will do extremely well. I think the thing that we own is, and that we spend all of our time on, I think it's actually a nicely complimentary piece, which is in some ways really thinking about how to best optimize for very particular capabilities within software engineering, and how to deliver and ship like a really great product experience around that, right? - You do not want the developer products for these foundation models to commoditize because their new game leverage. If anthropics in Claude continues to be so much better than it is, their leverage is gone. Do you not want that commoditization? - Yeah, look, I mean, I think it's, to be honest, I mean, I think whether we want it or not, it seems like folks are pushing ahead in the foundation model space and making a ton of progress. We've seen a lot of launches even in the last week or two, right? I mean, there's Chimie and Grock, and a ton of great progress that's been happening. And I think that is the natural way of things, which is there will be competition in that layer. And I think some folks will, especially in particular, verticals and use cases be able to establish differentiation and there will be competition in our layer. And then some folks will be able to establish differentiation. As this equilibrium develops, I think it is just naturally the case that folks and these layers want to collaborate and figure out the right ways to work together as long as that holds. - Why would anthropic wanted collaborate when they could just own it? Sorry, I'm a VC, so I'm a VC. - No, no, of course, of course, I love it. They all hidden me with all the hard questions. Look, I think the answer at the end of the day, what we focus on is very, very different. The kinds of questions that we think about, for example, on both the Devon or the Winstorf side, by the way, is how should humans and AI work together to produce code? It's one thing to just kind of like solve for generally smarter and smarter based models. It's another thing to teach your specific model, all right? Here's how to go to data dog and pull up logs for this thing and here's how you debug a front end live. And here's what the, here's our representation of the code base, which we're learning and iterating over time and so on. There are a lot of different verticals to work in and to solve for. I think we have parts of this equation and I think the foundation labs have parts of this equation, but I think the truth is you really need both. - Be preface that one, where you can ask the hard questions. So fuck it, it's the end of the day in the UK, so I'm like using that as an excuse. What percent of your revenue goes to anthropic, do you think? - Bullpark. - I have to pass on that one. (laughing) It's like with children, you know what, I'll push and I'll see how far they can go. (laughing) When you speak about that progress and the incredible progress that we've seen, some think that it will take the same route as self-driving cars, where there is this kind of plateauing and then we'll see again another inflection. What gives you such confidence that we will see the continual progression in models that we've seen over the last 12 months, over the next 36 months? - Yeah, so I'll give you two dots there. The first thought is there's very strong signs of this continued progression. Largely because by the way, there's a lot of specific work which is you can call it research, you can call it engineering, you can call it infero or whatever you like, but there's a lot, which is roughly that we know the techniques and we have a lot more to do to go and scale this. It's RL, for example, is I would say the biggest breakthrough of the last year and a half call it. It's crazy to think that you roughly can solve any benchmark. It sounds insane to say, but that's really what RL is converging on, which is basically if you have a clean enough set of here exactly the behaviors that I want, here are the environments that you need to be able to operate in, here's what it means to succeed or fail, you can just train a model that does that. And naturally that is just such a powerful capability, which I think is going to be applied to more and more spaces. And so I think we will see that progression. The other thought I would give though, and I mean this truly sincerely, is there are a lot of spaces, and a lot of different things going on in AI. I think there are a lot of spaces that are early today, that you see AI making progress, it's getting better and better, and you're going to the next step and the next step. In code, you are just slower as it's offered to you if you're not using AI. That is the truth, and it is a no-brainer already today. And look, I think it will be more of a no-brainer, and I think we'll be able to make engineers even more efficient, you know, be able to do even more with code. But I think that is already the case. So many things to unpack there, dude. Do you think tools like Devon and others make 1X engineers, 10X engineers, or 10X engineers, 100 engineers, 100X engineers, if I were to put you in one camp. - Interesting. It actually really does depend tool by tool. I do think that the product experiences for the 1X to 10X and the 10X to 100X, or even the 0X to 1X or whatever you want to call it, are some of the different project experiences. - I've interviewed Benny offer Salesforce, saying she blad at Robinhood, both very recently, and they both said that 50% of their net new code is created by AI. Would you agree with that in what you see with your customers and payers, one? And then what will that be in three years time? - Yeah, absolutely. People often talk about the percentive code that's written, I think the obvious thing to call out, as well, it's a little odd because for one, you have to factor in how important each line of code it is, right? If it's just like a ton of protocols, then that's one thing versus a lot of the core business logics and other thing. And then two, of course, is written with the help of AI is one thing, but how much. And so for those reasons, we often like to think about in terms of how much faster is an engineer using all these tools? And an engineering using the best AI tools, and who really understands how to get value out of them, how much are they doing in one hour versus how much they would do in one hour with no access to AI? And I would say that something in that range of 1.5 to 2x feels right to me today, in aggregate. Look, I think in three years, there's no reason that shouldn't be a 10x. The thing that's really fun, by the way, is I think we're going to have more than 10x more code. It's something I've been saying recently, is I keep track of every time in my daily life where software fails me. We've become conditioned to be okay with it, but the truth is it happens all the time. You know, you think about products out there and the way I like to say it is, there are, I say, this top tier of products, which are in some sense like, I'll call them, the best made products in the world. And I'm thinking of like YouTube, TikTok, Instagram, and so on. And you can really feel every little detail that they've done with a ton of care. It's, they're streaming in tons and tons of data and it's always super efficient. It never goes down and it's super reliable. The algorithm basically knows you better than you know yourself. And that is like, call it like hundreds of millions of hours of engineering time. You know, that went into building that piece of software, right? And then you kind of go down to the next layer of, you know, the next order magnitude of software that has tens of millions of hours spent. And then single digit millions and so on. You know, and you see the differences really quickly. I mean, when you're logging into your bank or when you're dealing with your healthcare, you know, you're working with your insurance and trying to get things going. Or when you're trying to go and like, you know, navigate your customer list or things like that. And the truth is, all the software can be 10x better. You know, and I think there actually is 10x more of it to, right? And I think that's one of the fun things in code, right? Is that it really does have this Jeff and Sparadox? - When you think about that, develop proficiency, but what are agents not able to stay? That they really need to be able to do? - So I think there is a real point at which agents are truly able to take over ownership. I guess is how I would describe it of the work that is done. And the way I kind of want to say it is like, "Okay, I think in the future we will get to a point where you're not looking at your code." You know, you're looking at your product, right? And you're looking at your product. And at the end of the day, code software, engineering, this whole thing, is just telling your computer what to do. And code is the language that your computer happens to speak. And that's why we all have to learn how to write code in order to do it, right? But I think over time, yeah, you get to the point where you can just say, "Yeah, this website, let's add a new tab here and let's put this in that information." And maybe let's go collect this info from the user and we'll save it in the database this way. And this button should be a little bit rounder. And you're just able to make those calls and make all those decisions. And I think we'll still call it programming, but I think it's going to transition into being more of a technical architect.
a technical product manager, someone who's really owning these decisions. Agents have to get to the point where they are basically that high touch and that understanding of context that you can give them that level of instruction and they'll just go and do that. In a world like that, what skills become more valuable? More skills become less valuable. Everything that you do is going to be about essentially this kind of core thing of deciding what is the solution you're going to build. What is the problem that we're facing? What is the solution that we want to build? How exactly do we want to architect that solution? I think that's going to be the most important skill. There's so many things I want to ask. I do just want to give up before I forget it. You said about the one and a half, that is more efficient and more productive. Do you think we are struggling now in a value-casm gap? What I mean by that is how much do you charge for Devon today? Percy, average. It's all usage-based. It's essentially by the hour. We try to make things so that they're about 10x cheaper than basically the value of your tide. I think about, say, a software engineer being 300 grand a year, you give it a take. 150K then would be that.5x that you are adding. All these tools are going to be sufficiently paid for the value creation that they are enacting. Yeah. Value creation is a beautiful thing. To this point, by the way, of where does the value of a crew in AI that we were talking about earlier? There are 30 million software engineers in the world. We're going to make them all 10x more efficient over these coming years. We're going to be writing 10x more code. We're going to be doing a lot here. I think that we could talk about whether it's 5% or 10% or 20% or 30% of the value that actually gets collected by the company is doing this. Honestly, I think the highest order bit is actually less that and more about getting the technology and building the products to a point where everyone is going a lot faster with them. When you look at something that no one sees that everyone should see, when you think about the future of AI code and the future of software engineering, what does no one talk about? Do you think more people should be talking about? Yeah. A focus on deep context. It's already better than us, honestly, at these sandbox problems. A lot of the tough questions that we answer today are things like, well, there's this project that we're trying to use today and it's very similar to what somebody else actually asked Devon a month ago. How do we use that knowledge and kind of improve on that in order to make Devon smarter? Or there's these little things of like, you want to go test the front end for your code base and make sure everything looks as expected. You should be able to understand what is that supposed to look like or why is this different from what, you know, if you find a bug, you should be able to understand how you found the bug. That's a lot of the little detail of, honestly, I'll call it the difference between code and software engineering, which is basically working in a large complex code base, building some intuition and some representation of all the different pieces and how they interact with each other, learning how to use all the various tools at your disposal to actually understand what's going on into it, to debug and diagnose. And I think that is actually the big problem, honestly, in AI coding. I think that's the big thing that folks will work on next. The thing that's fun, by the way, it's a very practical problem. You can call it a research problem, obviously in many ways it is a research problem in terms of how to push for these capabilities and make these things better. But it's not the kind of thing that you can solve in a sandbox. It's the kind of thing that you actually really just need to think about the practicality of software engineering to get into. You said there about kind of, hey, coding agents was what you saw before other people saw what no one was discussing. It's a hard one, and I don't know how blunt I can be. But it felt like Devon fell out of the zeitgeist a bit, if we're honest. El Cursor and Windsorth owned consumer attention and owned the brand. Do you think that was the case that Devon just wasn't very good at marketing? Or do you think that was like a product misstep? Yeah, so it's funny because I think there's always an external perception and then there's an internal what's actually going on with the numbers. Without going too much into detail, what I can say is that in the last six months between January and now, even aside from this, obviously, the latest deal this week with Windsorth, the usage of Devon has actually grown something like five to 10X, and that's been the case in both self-serve and enterprise. I think the pattern for Ford is worth is it's really like real engineering teams that bring on Devon and they tag Devon all the time in Slack, they tag Devon in linear and so on, and they kind of use it and grow it and share it that way. In many ways, it is not the same kind of single non-engineer just going and signing up for an account can immediately just build something really cool with it. People do use it that way, but it's not the majority of our usage. I think the majority is really just kind of like real teams using it. But I think the point that you're saying I think does get to, I think, an important thing, which is that, like, IDEs and the IDE experience are obviously, they came and they really started working, I think, about a year before the Asian experience really did. I kind of think of agents as having taken off in the last six months or so and I think of IDEs last year when they really started to become kind of like no brainer obvious, let's say, in terms of the value that they provided. I think what we're kind of seeing is an artifact of that, which is that yeah, like folks are more familiar with IDEs because they've been around for a while. You fast forward six to 12 months from now and I think people will be familiar with agents to the same level that they're familiar with IDEs today. I think it's a 510X is mega, it's fantastic. The hard thing is if you compare it to like a rap leader or a lovable, the growth is just fucking nuts. Like, I'm in love with the ball, Scott and I just look at the numbers and I'm like, what? What? How does that work? Is that not a fair light for like comparison if you're thinking about growth rates? The reason I would say that it's not, also actually, but actually I don't even know the number, what are their numbers of the last six months or what are their, they've gone from basically zero to 80 million an hour. For what it's worth? Well, yeah, look, I don't want to get into exact numbers, but even with the wind surf deal aside, like not including that at all, we've done roughly that as well the last six to eight months. I stand corrected on my consumers' like guys come out. Which is, I think it's an important point to consider, by the way, I think these are great companies and you know, obviously I think that's a good thing. But then, but then is a more clean problem, dude. I'm saying this to yours now, you know, I hope we can be friends. I'm projecting for this relationship as a friend, I would say. You guys should be better at selling yourself. That's great. Well, perhaps we should be and you know, the great news is we've just now inherited a great marketing team. And we get to do that. But you know, the only thing I'd call out, I know I'm dreadful from roughly, I've met a lot of these folks before. I think it has actually quite different products and businesses and so on. Which to the point that we were saying earlier is, you know, there are a lot of things, a lot of different products experiences that you can solve for and code, right? There's like bringing a 10x engineer to 100x, there's a 1x engineer to 10x, there's, you know, someone who doesn't know how to code and bringing them to 1x for better or for worse. Repelated, lovable, to my understanding, at least, are a much more consummary lean, right? And perhaps that's why you hear about them more on Twitter or on YouTube or things like that. Whereas Devon is really, you know, it can be anywhere from startups to some of the biggest companies in the world. But it's really focused on specifically engineers on engineering teams trying to do their work and to go faster that way, right? And so, you know, you can see this and how this is all set up, right? Like most Devon sessions are started through Slack or through linear and Devon makes these pull requests and GitHub that you go and review and merge and Devon works with your whole development system, right? And learns how to get onboarded to your repo and so on. And I think that's kind of one difference there. But with that said, you know, the point taken, I take the feedback. Yeah, I think it's a very fair point. Dude, I'm always here to show you how to do a good tweet. I'll pretend you now know how to do a good tweet. That's all I know. You're the champion. I mean, you're the master at it. I was going to say so I, you know, after this, we'll talk. I want to get your tips. Dude, honestly, I have so many for you with those numbers. I'm like, wow, dude, you need, yeah, anyway, can I ask you, on my job as an investor also is to think about like market makeup at an end state because that's where value kind of also kind of recruits in my mind when you think about that and you think about like developer, where the market shakes out. Is it that cursor when bottoms up developer minds and windsurf, wind and cognition now, when top down, large and price, super solid blue chip clients? Is that how you think that developer market shakes out? I think the honest answer I would say is, I think it's far too early to call on any of these. I'll give you a heartache, which is none of us are that close to the future of software sharing. You know, and I think the future of software sharing is, look at this. This is going to take place over the next couple of years or so, but it really is this version where we call it an IDE or a coding agent or this or that. These are the terms that we use. I think if we're being real about what we're building here and what this is all going towards, I think of this as the next generation of human computer interface is the problem that's being solved here, which is basically, as we said, code software sharing, the whole point of that is just telling your computer what to do. At some point, telling your computer what to do is not going to take place with code. It is going to take place with you just expressing your intent. There are a lot of things, don't get me wrong, there are a lot of capabilities problems to solve to get there. There are a lot of interface problems to get there, but I think that is essentially what we get to is a simple thing I would say is Tony Stark does not pull up his laptop. Tony Stark goes and talks to Jarvis, right? I think there is a point at which you just have a very clean connection, the ability to express your intent and do these things. People talk about generative UI, people talk about single use software and so on. At the end of the day, what it all really boils down to is you to have a very clean connection and just being able to tell your computer what to do and it will do that for you. I think if we think about where everyone is today, I think there are 10 or 20 levels of what the product experience looks like until we get there. The fun thing is every level is two or three months and so if you kind of just multiply that out, it means we're going to be there in like a few years, if this phase keeps going. I believe a common enemy is a very important thing within a company. If I were to push you, Scott, and say, "Hey, who is the capacitor that you most look to think about?" Respect, even. Who would that be? Jensen said this one. It's always stuck with me.
When you have figured out a way for your company to win that means that no one else has to lose then You will know that you have found your path You think he's done that. I think that was a huge my You know I had videos an incredible business in the Jensen obviously has To freaking monopoly like they go well and I think his point right is is that I think at the end of the day There are so many different verticals to serve There's so many different niches to own and there are a lot of businesses which of course we'll work in adjacent spaces And we'll see things and you know you'll you'll run into folks on a deal or whatever it is But at the end of the day people have the things that they specialize in and and I firmly believe that like I think The thing that's been fun I think throughout the history of cognition is for better for worse Maybe it's because we're insane. I don't know But we've always had a pretty unique approach and a pretty unique view about like here's here's what we think of the future It's gonna be here's what we want to build for and that's different from the cursor It's different from a you know open-air and throttic themselves and how they think about these things and so on Right and I think like the fun thing about that is is there is so much to build on code And and I actually truly do think there will be more than one winner. What have you not built that you would most like to build? Yeah, the answer I would give today which is which is I mean the biggest thing that we've been thinking about in the last few days since the steal is Really and figuring out what is that combined experience of IDE and agent because I think there really is something here to what we said earlier You know, I think years from now all these systems will look so different You know and we made it even have the same terms for them But I think and in the immediate future I think both IDE's and agents will be an important part of it developers workflow And you can imagine all sorts of things of hey, I want to go and plan out a task in the IDE I want to be able to use the intelligence and the retrieval and you know dev and search and you know read on the wiki to understand exactly what What decisions I need to make or what parts of the code this is going to touch and and what things I need to plan out And then I want to be able to hand that off to an agent have the agent go do the bulk of the work And then I'm going to go and review the code and naturally it'd be great to review the code locally in my IDE If there's any touch up that I need to do I can use that you know You I can use my in IDE tooling to go and do that But but basically you know figuring out what is that combined experience where you can go from synchronous to asynchronous to synchronous For you know and be there for whatever parts need you and be able to go and parallelize and do more for the parts that don't need you I think that's going to be a really fun question to figure out in terms of you know what happens with the combination of windsurf and Devon I think in the immediate short term obviously like I think that there's a lot of Devon to run and a lot of windsurf to run and we plan on kind of like We certainly plan on on maintaining the philosophy of both of those products But I think kind of finding that intersection in between of how do you make it a really smooth experience to go between for the folks who use both Is going to be the fun one dude. I've so enjoyed this it was yeah, it's just been a really fun conversation I do want to do a quick fire around if that's okay, so I'm gonna say a short stop. Okay. I have to say I'm a bit under slept And so it's please pardon me if I need like an extra you know one or two seconds for the fire How have you been sleeping? Next question No, look at I be it. It's it we had some crazy we had some we've had some crazy days and I mean Saturday Friday Saturday Sunday Obviously was figuring out you know if there was a deal to be done and making that happen and Monday Tuesday Wednesday has been figuring out how we bring the teams together and how we build the really great thing and obviously like the The immediate thing is going out to the customers and letting them know look like we're here You know we've got the the engineering firepower and the work to be able to support things We're gonna make sure you guys have a super solid experience and if anything we're gonna be able to make it better even faster And figure out the right way you even up leveled the engineering team, hey And Long story short not not a lot of sleep this week Okay, dude was one widely held belief about AI that you think is completely wrong. Yeah, I'll give you a take Sam all men had this post 10 years ago called bubble theory. He was basically saying this was in his YC days And he said you know everyone here says that all these companies are valued and we're in a bubble and it's obviously fun to talk about bubbles and all But I don't believe that and I'm gonna give you a bet and he said it looks hilarious by the way when you read it in Resrespect which is like here are the top, you know YC companies today and it's literally like Uber and Airbnb and like you know What I and it's like today these are worth this much I predict that five years from now they're gonna be worth you know three acts more You know and then it's like here are these kind of like emerging, you know mid-mid-level companies and then that list was like you know Stripe and some of these others and it's like today this is what this much and I predict these would be up at least three acts You know and he basically said everyone says we're in a bubble if you'd like to go and bet on that and put your own money down I am happy to put you know money on this I have to say in AI I really feel this I feel this I think now more Strongly than ever which is over the last two years people kind of think of it as this this gen AI wave again Or like we're saying like lots of incremental jumps, but the truth is RL is the master like the big story I'd say over the last year two capabilities and I think people have really underappreciated how much is possible with RL And I would see will not see with RL that they should see because I kind of think I don't yeah We've had a few years of I would call imitation learning before RL Which is that's what when it's a GPT 3 and 3.5 and so on which is basically you take the entire internet and you read the whole internet And you get a model that sounds like somebody on red It's crazy. I mean it was obviously chat GPT like shocked the world that it was a huge moment for everybody But but yeah, you could kind of see why well, you know talking like the average is not necessarily the thing that you need to do to go build a great You know medicine specialist or a great you know software engineer or any of these things and I actually got a lot of the way there To be fair more than people might have guessed But I think that the next big thing with RL is as we say you can take any benchmark and solve it It's crazy to say but but I think the next step of that is Honestly an open question for everyone in every application which is hey, what is the benchmark? What if you're you're you're an accountant, you know the benchmark I mean in real world terms the benchmark is roughly like you're right You submit returns on behalf the client and you know if you get audited by the IRS and something comes up Then then that was a fail for the RL and if you don't Then you did a good job, but obviously, you know you want to be able to constrain it to shorter feedback cycles than that and think about What are the ways that you can determine if an agent's work was a success or a failure? But what what once you have that it turns out that you can just train agents to do all sorts of things and that's what we've seen By the way, I mean we've seen like folks getting gold medals and these these international math competitions and things like that with AI And and the truth is that this is really hard stuff, right? And and obviously you have to go and apply it to every vertical, but it's gonna happen So I guess on that point I guess I was just gonna say it's like you think about the companies today There's the foundation layer companies, right? There's the the application layer and there's a lot of the tooling. I would happily bet take all the foundation labs today Right, it's it's the private ones right there's open a eye. There's a thropic. There's X, you know Let's put an SSI we can include thinking machines in that list as well if you add all their evaluations today What is that like 500 B something like that? I think that's gonna go way up in the next five years and then if you take all the application layer companies, you know It's the top ones that come to mind. You know, there's perplexity there's Sierra there's Decagon There's Harvey there's us there's cursor there's you know and you take that list of companies and today that's the list of Companies is probably worth I don't know 5,000 to 100 billion and I think that's gonna go way up in aggregate And obviously it doesn't mean every single company is gonna go up But I think that the value that we're gonna produce is is so massive here that you can invest in open AI at 300 bid in Or anthropic at 60 billion which one do you choose to invest in? Honestly, and I mean this sincerely. I think both are great investments amazing. You can put my name one Oh, man, anthropic has quadruple their revenues since they were valued at 60 billion So it seems like a fair one to it to pick do you think we will have you mentioned that like six in the foundation model there? Yeah, do you think we will have six? Do you not think the consolidation already begun? I think the consolidation happens and obviously you know, there's there's Google and there's there's meta I think the consolidation happens and I think in total we probably end up with something like two to six players Three to five seems pretty reasonable, but two you think just two you got chat GBT in open AI on consumer Then you've got anthropic on enterprise really you think it's just those two you might have like really tertiary ones It's just like search your like mine ones that you duck duck goes of the world, but like The duck duck go of AI I'm sure someone's looking forward to winning that elusive title. I think the thing is funny on consumer I totally agree which is chat GPT. I mean that everyone knows just chat GPT as the term I think there's an interesting effect which is obviously you know companies want choice and and I think this is gonna happen in a few spaces as well Which is you know, it doesn't necessarily it is always a power law distribution where number one is like 75% and number two is 20% And number three is like 4% and everyone else combined is 1% or things like that But like I think in terms of folks in terms of like market share Let's say but but in terms of people getting there on capabilities and being able to offer something which is at least competitive enough that 4% or 20% of people would consider it I think there is reason to believe that there will be at least a few folks that that get there Yeah, for the acquisition what was the percentage between stock versus cash? Is it make sure of both? I probably can't comment on the exact split I'm so enjoying this constant game of like push push I know I'm very happy to you you can ask all your I'll push back what I would I need to push back so can I be honest I think you missed a trick with your acquisition video. I'm just gonna be I totally I didn't I didn't I didn't know you from shit Okay, obviously and I watched it and I was like no offense It's a bit personalityless and meeting you you're like I'd love to hang out with you And you're like amazing personality charisma You're awesome. I think the
I think an acquisition video could have been more like, "Personable." - That's fair. - I agree. (laughing) - Dude, we were in between, like the lawyer is all pulled in all nighter as well, going and getting this because it was like, yeah, I mean, we need to get this ready to go, and there's just all the various little things of this turn, this turn. But yeah, it was in between that and in between sorting out all the questions with the team, and then we basically had like 20, 30 minutes to go just film something real quick. - Oh yeah, dude, you're right. - I got GBT right as script for an acquisition. - I was gonna say, I was pretty happy with the output relative to the amount of effort that was put in, but I fully agree. (laughing) - But there's more to it. - I'm 100% with you. - My favorite bit was, funny enough, we did have one competitor who we thought really highly of. - Condition. - Okay, hit me with your hard question again. - So what did you believe in the world, AI, that you have subsequently changed your mind on the loss 12 to 24 months? - I think one thing, which is, it just really started to change in the space. 24 months ago, actually the story was all about data and just more and more and more data, right? And quantity of data and just figuring out how you get even more. I think that has changed a lot actually towards figuring out a small set of highly curated data for exactly the use case that you care about. And that's been a fun one. You know, I'll just give an example. We, a lot of our research work, we obviously, we obviously can't talk about publicly, but we did share one project, which is a model that we released called Kevin. And so basically, Colonel Bench is, you know, a bench product that involves your ability to rate CUDA kernels. - I'm gonna be honest, whoever names your products really needs to be fired. Daven and Kevin. Like, no offense to you, like that's not very creative. (laughing) - Well, we'll work on that, we'll work on that. But, yeah, so we rolled out Kevin 32B. And the whole idea is basically with RL on agent trajectories, on very specifically CUDA kernel, you know, agent trajectories, you can make it way better than the soda based models on this stuff. And this is kind of the, this is a story that we're seeing in a lot of these, which is rather than just this mass volume of data, if you have a very particular use case and very particular behavior that you're looking for, honestly, it turns out that like a small amount of data in exactly that vertical, and you set up the environments correctly, and you set up the feedback loop correctly, is more what you need. So, it's a lot of compute, not a lot of data, I think is what we're seeing. And then it's like quality of data over quantity. - Would you advise a new young student to study CS today? - Yes, absolutely. - If anything, I would say, I'll give you a hot take, which is the complaint that we've had all along about schools is they, they're supposed to teach you CS and software engineering. You go to school and you learn about garbage collection and algorithms and architecture and everything. And then you go on the job, and then a lot of what you have to do is, you know, debugging JavaScript traces, right? And I think the thing that's kind of funny is, as we said, like this fundamental skill set of how you solve problems and how you think from first principles and understanding the model of a computer and understanding a lot of these kind of like architectures of yeah, how does the database work or things like that. Like those are actually gonna be the important things, I think, going forward. And so, you know, maybe one way to say it is, at the end of the day, like the degree in CS is more a degree in how to think. And I think that will always be valuable. - Scott, I wanna finish on one, which is I think really important. I've so enjoyed this. And I watched this actually take talk. And it's that if you wanna build a relationship with someone, ask them about the trait within themselves that they aren't most proud of, or they like the most. Because in future interactions, you're able to refer to it and it makes them feel good. I'm intrigued. What trait are you most proud of? What's your favorite trait of yourself? And why? - Yeah, that's a fun one. It's a. (laughing) I actually am very curious to hear your answer to this question as well, by the way. I think my answer, funnily enough, is, I actually would say it's like a certain emotional calmness, I guess. - I'm a very competitive person. I get very frustrated and things like that. But I think for better, for worse, we've been in a lot of stressful situations. As you can imagine, we've been in a lot of them in the last six days in particular, but honestly, that's been the entire story of the company. It's probably a bit more than average, you know, than an average week, but like not by that much. And I think the. - Yeah, I think for a better for a better, I'm proud that I'm just able to stay composed and in the situation. - Dude, I love you. You're great. I think you are probably the most under-discussed personality in this business. Like you need to do more. Like the Scott Wood Brown needs to be a bigger personal brown. - What do you suggest? I'm looking to you as the interim chief marketing officer at Cognition. - The thing I think is nice with your product is actually that you fall into the Nike bucket. And what do I mean by the Nike bucket? The Nike bucket is that success was, they make you feel like a superhero. They tell you that everyone is an athlete, that even if you don't have the skills, you are able to do things you never could before. And I think that you and Devon fall into that similar bucket of human enhancement. And so I think you should tell the stories of that much better. And then I think you should, it bluntly, very clearly, tell the story of your own growth much more deliberately. People want to be part of a rocket ship. And gossip is very vicious. And it happens when you don't shape the narrative yourself. If you shape the narrative, gossip doesn't really happen because I can't say you're at 20 million in revenue when you're not, you're much higher. And so don't let other people shape your narrative for you. - Yeah, no fair enough. I think it's been a real shift, which is I think a few months ago, it's like we were like, you know, it's almost better if people don't hear that agents are working. You know, and we'd rather it that way. I think in the last few months, everyone in their mother is now trying to do agents anyway. And so we might as well, we might as well be more public. - But I also think just for talent, I think for funding, I think, people also just want the brand. - Yeah, that makes sense. (air whooshing) - I mean, more to fantastic guest. He put up with some pretty tough questions there, which he was not prepared for. It was an emergency podcast. He was just a fantastic personality. If you wanna watch the episode you can find it on YouTube by searching for 20VC, that's 20VC. But before we leave you today, Harvard Management Company is constantly seeking out the next generation of great investors and entrepreneurs. I personally've had the pleasure of working with the HMC team and can say that they are truly exceptional partners and savvy investors. And after securing top tier partnerships, like those with Harvard, the next step is building trust with customers and Secure Frame makes that simple and seamless. Secure Frame empowers businesses to build trust with customers by simplifying information security and compliance through AI and automation. Thousands of fast-growing businesses, including NASDAQ, Angel List, Doodle and Coda, trust Secure Frame to expedite their compliance journey for global security and privacy standards, such as SOC2 and ISO27001, CMMC, NIST standards and more. Bad by top tier investors and corporations like Google and Cliner Perkins, the company is among Forbes's list of the top 100 startup employers for 2024. Cheat-tooth's best software awards for higher satisfaction products and a recipient of the 2024 Cybersecurity Excellence Awards. Something I definitely never got in school myself. As always, I so appreciate all your support and stay tuned for an incredible episode coming on Monday.
PodcastTranscript AI