20VC: Cognition CEO Scott Wu on Acquiring Windsurf: The Process, The Deal, The Rationale | Did Google Overlook a Goldmine in the Core Asset and Did Founders Leave a Sinking Ship | How Cursor and Cognition Deal with Ever Increasing Reliance on Anthropic

0:00I 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, 20 VC team worked through the night to edit it and turn it around.

0:38That is commitment for you. The only thing that anyone will talk about this week is cognition by windsurf. What happened? Did the team get screwed 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 vs. Cognition vs. Claude Codd. 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.

1:18Their 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 want to share with the team? Just send it over to ventureathmc .harvard .edu.

1:50And after securing top tier partnerships, chips. 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 -grown businesses, including Nasdaq, Angel List, Doodle and Coda, trust Secure Frame to expedite their compliance journey for global security and privacy standards, such as Sok2 and ISO 27001. CMMC, Miss Standards and more. Bad 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.

2:31Cheetu'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 Vinneau 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.

3:05It'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, you know, on the devon's side, I think in a few ways it seems to be a pretty natural fit one because I think obviously there was an incredible, you know, team left behind.

3:39And if anything, I think, well, we have at cognition, I would say, especially focused on the core engineering and product team, whereas obviously, like, 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.

4:15We got on the call together. We were talking about, you know, 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 WinSurf's perspective, the way that we kind of saw it was, you know, 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, you know, months of diligence and everything.

4:45And that has apparently happened multiple times already, you know, in Windsorps company history. And I think in this particular case, I think it's everyone is scrambling, you know, 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, like, I heard the news, you know, 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. Even like a week long, you know, it is Friday night now. We want to have something ready to go by Monday.

5:14I'll pause because it sounds like, yeah, but. I'm just like, entry, when you look at like scale, it's like, you know, there's a husk left behind which kind of deteriorates. And you're saying 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 we still can scale. Yeah, so there is definitely this conversation online right about all the best researchers laughs and there's just nothing left behind and I don't think so is 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.

5:48It'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, we had the very nice complement 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.

6:20I think the team pieces are very complementary. I think there's a lot we can do on the product. The thing that just seems absolutely clear from both of our size is if we're going to do something, we want to be ready to do this and announce it by Monday morning. You know, it kind of reminds me of like the bank goes into receivership Friday night. You got to have an answer, you know, 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.

6:49That we are going to do this. We're going to support, you know, the products, we're going to 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 going to 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.

7:22and Jeff and Graham and Kevin and the whole crew, honestly, 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 advanced 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 now 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, you know It's gonna be we're gonna have to do this quickly and we're gonna have to to figure out what we think is the right long -term fit for it I think most people see this is 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 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?

8:27Yeah, I think there's some rich roots 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 for worse It's 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 are in spree and just how much might is it getting out of hand?

9:03Yeah, maybe crazy opinion cut for a short opinion, but I actually think it's it's it's quite reasonable 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, you know, and then the question was just, if you said there were no new research breakthroughs, no new discoveries, you know, you don't even have to bet on that. And you just said, all right, we're going to, you know, 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.

9:40I 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 or freeze, I think it's going to keep moving even faster. But the crazy thing about AI is there really are so few people that are really determining the trajectory of AI But I think in aggregate I think the view is right, which is you know, there's so few people Is it I'm so sorry.

10:15I'm totally agree in terms of what is it like a hundred? Is it like 10 ,000 just what caught him is 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 a lot of difference is 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 a thousand 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 down here, but like do you need them if you're in the application layer?

10:50I think the level of talent and the fierceness of competition is in many of these is actually like greater even in the application there, which is crazy to say because I think the competition and the foundation labs are extremely strong. When you think like holy shit and I think can happen style, which it is today, one of the like, oh shit's that could happen, that's very obvious to I think me and everyone is like the dependency now around anthropic, you know obviously windsurf, you know 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.

11:21Is 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 of crew in AI? Is it chips? Is it semiconductors? Is it foundation models? 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.

11:55I think the thing that we own is, and that we spend all of our time on, I think is actually a nicely complimentary piece, which is in some ways really thinking about how to best optimize for very particular capabilities within software and sharing, and how to deliver and ship like a really great product experience around that. You do not want the developer products for these foundation models to commoditize because they're new -gain 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.

12:35We've seen a lot of launches, even in the last week or two, right? I mean, there's Kimi and Grock, and a ton of great progress that's been happening. And I think that is the natural way of things, which is, you know, there will be competition And in that layer, and I think some folks will, you know, 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, you know, I think it is just naturally the case that folks in these layers want to collaborate, you know, and figure out the right ways to work together as long as that holds.

13:09Why would anthropic want to 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 WinSurf 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's 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 We can ask the hard questions.

14:03So 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 bullpog? I have to pass on that one It's like with children, you know where they'll push and they'll see how far they can go. When you speak about that progress and the incredible progress that we've seen, something 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 inflation. 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?

14:41Yeah, so I'll give you two thoughts 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 is what it means to succeed or fail, you can just train a model that does that.

15:24And 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. In AI code to be truly honest, if there was zero progress, the world would still be entirely different. You know, 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.

15:55That 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 one -axe engineers, Tanaxe engineers or Tanaxe engineers, a hundred engineers, a hundred 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 one X to Tanax and the Tanax to 100X or even the zero X to 1X or whatever you want to call it are somewhat different project experiences.

16:31I've interviewed Benny Offert Sales for 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' appears? One. And then where 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 is, right? If it's just like a ton of protocols, then that's one thing versus a lot of the core business and other things.

17:03And 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.

17:39It's something I've been saying recently is I keep track of every time in my daily life where software fails me. We become conditioned to be okay with it, but the truth is it happens all the time. You think about products out there and the way I like to say it is, there 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. You know, it's, they're streaming in tons and tons of data and it's always super efficient.

18:15It 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. and that went into building that piece of software, right? And then you kind of go down to the next layer of the next order magnitude of software that has tens of millions of hours spent and then single digit millions and so on. 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're working with your insurance and trying to get things going or when you're trying to go and navigate your customer list or things like that.

18:51And the truth is all the software can be 10X better. And I think there actually is 10x more of it to write. And I think that's one of the fun things in code, right? Is that it really does have this Geven's paradox? When you think about that, develop proficiency. But what are agents not able to say 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, look, I think in the future we will get to a point where you're not looking at your code.

19:23You're looking at your product, right? And you're looking at your product. And at the end of the day, code software is doing this whole thing. It's 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.

19:51And 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 skill 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.

20:25What 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? And 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 go back before I forget it. You said about the one and a half, that is more efficient and more productive. Yeah. Do you think we are struggling now in a value -casm gap? And what I mean by that is how much do you charge for Devon today? Per seat, average? So 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 time.

20:54Because if you think about, say, a software engineer being 300 grand a year, you give a take, 150k then would be that .5x that you are adding. Yeah. All these tools going to be sufficiently paid for the value creation that they are enacting. Yeah. Value creation is a beautiful thing. And to this point, by the way, of where is the value of 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. And I think that we could talk about whether it's 5 % or 10 % or 20 % or 30 % of the value that that actually gets collected by the company is doing this.

21:35But honestly, I think the highest order bid 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 that 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.

22:11How 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 wanna go test the front end for your code base and make sure everything looks as expected. You know, you should be able to understand like 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 like how you found the bug. And that's a lot of the little detail of, honestly, I'll call it the difference between code and software engineering, right? 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 and to debug and diagnose.

22:51And I think that is actually the big problem, honestly, in AI coding. And 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 for 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.

23:20It'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 the latest deal this week with Windsor, the usage of Devon has actually grown something like five to 10X, and that's been the case in both self -serve and enterprise.