Best of the Pod: Vercel's Guillermo Rauch on AI and the Future of Coding

[00:00] I think a lot of developers are probably thinking, okay, is it even worth it to be able to code right now? I don't think I would identify today even as a coder, even though that's what I obsessed about for years. Since I'm like 10 years old, Trent has been away from the implementation detail, which is the code, and towards the end goal, which is to deliver a great product or a great experience. One of our enterprise customers for VZero is a company that has been in the cloud space since the cloud was born. [00:30] literally like in the room where the cloud was invented. And they told me we have it as a rule that we don't do code. They showed me their Slack conversations. It's just them sending these zeros back and forth. There's no GitHub. There's no infra. There is no anything. There's just building products. When things are specific skills, machines tend to take them over over time. If your skill when you were a kid was like, I can do math in my head really well. Awesome. Great asset [01:00] very well and even better than you. And so what I try to separate is what are the meta skills that are not as easily replicated by machines that you should still nurture? And I think those tend to be more around very high level conceptual thinking. What's about to become possible that I can start cooking on now?

[01:30] you [01:33] Hey, Dan here. I want to take a second to tell you a little bit about LTX Studio. There are tons of new video generation tools out there, but few really give you what professionals need. And that's what makes LTX Studio different. The platform gives you industry-grade pre-production capabilities, detailed storyboarding, and visual concept development that actually serves real creative processes, not just novelty generation. [02:03] control, and you can make changes in real time as your vision evolves. And the AI-powered precision from Flux Context gives you consistent characters with professional level control over details like lighting, objects, and angles. [02:15] LTX lets you toggle between different models, including Google's VO3 and LTX V, [02:20] the industry's fastest open source model. And all of this comes with seamless collaboration features for real-time teamwork on concepts, pitches, and final production. Whether you're storyboarding your next film, prototyping ad concepts, or actually creating final pixel-ready assets, LTX Studio empowers you to fully realize whatever it is you're imagining. Check out the link in the description below to see how it transforms your creative process. And now, back to the episode. Guillermo, welcome to the show. Thanks for having me. So for people who don't know, you are the co-founder and CEO of Vercel. [02:50] You're also the creator of Next and Socket.io, which is very close to my heart. I told you when we met that Socket.io is the reason I could build my last company. So I appreciate all you've done for the developer ecosystem. Just met some folks last night that said the same. It's nice to get recognition for open source, but I'm thankful to the people that maintain those projects. So we have a lot to talk about.

[03:20] I want to talk about V zero and what you're doing with for cell. [03:24] But where I want to start is I was... [03:29] Before this interview, I was thinking about the kinds of things that it seems like you're interested in building and the kind of taste that you have, and then think about how that might apply in this sort of AI era. And where I kind of came to, and I'll admit O1 was involved a little bit. O1 was involved in the pre-production. Yeah. What I kind of came to is... [03:53] It seems like you're very good at playing at the edges of what's possible technically, looking for where there's like something that's new and valuable, but like pragmatically valuable, but still pretty messy. And then coming up with like a very clean, opinionated, zero config solution to it. So like I would put Socket.io in that bucket. I put Next in that bucket. I think Vercel is in that bucket. So one is I'm curious if you feel like that's kind of. [04:23] over. That's pretty good. Like, I'll tell you, like, it's funny. I was having a conversation with one of my co-workers the other day and, and [04:29] I was giving them the example of Socket.io as when I started the project, WebSocket, which is the underlying technology that later become a part of web browsers. Yeah. [04:40] For context, maybe everyone doesn't know about Socket.io. It enables real-time chat, real-time communication. It powers websites like Perplexity. And when I created that developer tool, WebSocket was in its infancy. So to your point, you kind of want to identify the wave.

[04:56] when it's just a few drops of water, but you have that, you see that potential. And it's also extremely messy. So being able to go from, I wanna do something with real time [05:07] to want to create an application, I think that's the vacuum that, or the huge gap that Socket.io filled out. And with vZero, I think we saw the same thing. So... [05:20] models became good at writing code. Specifically, they were very good at writing, like being able to design with HTML and CSS, and also writing React code. [05:32] So, [05:33] Even when GPT-3 was out, we started thinking about, well, this could be used for revolutionizing how design and code gets changed. [05:41] emitted and how [05:43] the process of bringing an idea to life could be completely disrupted, but it was really early. So to your point, there's almost like a parallel between like WebSocket was a draft of a specification when I started Socket.AO. And when we first thought about V0, models could barely be coherent in outputting code. Yeah. But there was that promise. And I think... [06:05] You kind of want to ride those waves when they're early. Yeah. Yeah, and I think one of the interesting things is it feels like, so everything that you've done so far has been very developer-focused, and it feels like you have your finger on the vibe of what developers want. Well, I'll tell you something I don't think I've shared before, which is the company was called Zite, and then we renamed it to Vercel. And the reason I called it Zite was a few things. One, I was obsessed with this idea that if you really want to

[06:35] capture developers, you need to capture the zeitgeist. - Right. - The zeitgeist is this idea of, you know, the collective conscious and how people are thinking about the future and you need to really be on your toes. Because things change so quickly. The vibes change so quickly, right? Like you see this with models, like one day, one model is popular for developers, the next day is another model. And another point was zeit means time in German. [07:03] And I was obsessed with this idea of real time with my background in Socket.io. And what I thought is that the best way to build software would be in real time. And it has two meanings. One is real time in the sense of [07:16] You're just typing and you're getting feedback instantaneously from the system. I think V0 is almost like the culmination of that idea. Interesting. Because you're literally chatting with the system and getting feedback in real time. But the other one was chatting with your customers. Right. [07:31] Getting feedback from the world and adapting. I sometimes joke that for early stage founders, their job is to, you know, chat on X with our customers and prospects, get feedback, fix things, and put a ton of quality craft and taste into their products. [07:49] I want to stay with that idea of like zeitgeist or like paradigm or like having your finger on the pulse of something. Because I think one thing that's interesting is, you know, to come up with Socket.io, to come up with Next, like you have to be... [08:03] developing stuff yourself all the time. I realized that

[08:06] this is possible and that you want to make something. Yeah. Like it's, it's too hard. So you need like a tool to help you do that more easily. And I'm curious how that has changed for you. Like, I don't know how much programming you're doing. It really hasn't changed surprisingly. So, um, [08:20] Like I mentioned earlier, one of my goals is to enable Vercel. [08:24] to do a lot of what I did as an individual, like come up with those ideas, being able to bring them to life. And so one of the things that I mentioned is, you know, kind of inspiring our employees with like, well, you could build the next Socket AO if you think about these principles. So I think a lot about... [08:40] principles that I can set for the company at large. And one of them is we're always customer zero. [08:47] We dog food our tools. We put product experiences before tools as well. I was just having a conversation with our tech lead for the AISDK. [08:58] So AI SDK is a really interesting project. It's becoming the number one framework in the JavaScript world for how you interact with AI models. So it's almost like the Socket AO or Next.js of LLMs. But it didn't come out because I was... [09:15] excited about AI or I live in San Francisco and I saw a lot of billboards about AI. I was like, hmm, we should have an AI framework. What it came out of is us building AI products. [09:24] So we're building a lot of really cool demos of how to use Next.js with AI. We're building V0. And we realize there's an opportunity to extract out the infrastructure of those AI products quickly.

[09:37] and then share them with the world. And so the operating principle here is, [09:41] You want to always put the product first [09:45] not the framework first. [09:47] A framework in isolation without having that initial sort of patient zero is never going to be a good tool. And this actually came out of my fascination with how Meta put out React. So Next.js builds on the open source UI infrastructure that Meta open sourced. [10:05] And what I noticed is that I kind of reverse engineer why I liked React and why I was impressed with it. And I remember, [10:12] It first started with their product. [10:14] I would go to the Facebook newsfeed and all of the things that they were doing that were very real time, like the chat thing and the notifications badge, and it just felt snappy. I actually remember a specific moment [10:27] when they announced that comments were going to start trickling in in real time. And I was like, [10:32] "That's hard." Because I built Socket, I was like, "That's really hard at that scale." It's very impressive. And then I realized, okay, you ask yourself, "Okay, how did they build that?" And this is actually something that a lot of people do unconsciously, and it's really good to tap into that. It's almost like a primal instinct. When you see something good, you ask yourself, [10:55] How did they build that? It's like when you walk into a studio and you're like, oh, I really like the vibe here. Well, the next thing you do, I mean, maybe not everybody, but like, I guess. It's a very foundry. The creatives, people, whatever, like we can call them like, I wonder whether they got that chair. And I wonder what fabric is this curtain. And so developers do this all the time. And so start with the product.

[11:18] You almost don't need to promote the tool. Yeah. Because people are going to ask themselves, like, all right, I want a product like that. Yeah, yeah, yeah, yeah. And that's how I still do it. I try to use our products, but I also am always on the lookout for really good things and then reverse engineer how you get there. Yeah, I've definitely found there's so much there to talk about. I've definitely found with every, so we have the writing and then we build software products. And a lot of the writing is about building new technology. 100%. [11:48] its own right yeah like the honestly like the best marketing for our articles has been building great products because yeah we're like well i want to know how they think about it totally um which has been really really cool and the other interesting thing that i think is is sort of um it's your ethos and it's also i think uniquely valuable right now is like the sort of dog fooding ethos we do that too because i feel like ai sort of um [12:10] it changes the landscape where there's still, there's now a ton of low hanging fruit of things to build. Cause like there's a new technology paradigm. And so you can just like wave a stick around and just think about like, okay, what are all the things that I want? And probably someone has not built that in a good way yet. Whereas like, I don't know, three years ago at the apex of the B2B SaaS wave, like all the low. There was a saturation. And that's what you see when there's a platform shift. AI is a new platform. And when a new platform emerges, it's, [12:40] it's kind of hard to even estimate how many new applications will emerge. Yeah. I remember the early days of the iPhone. [12:48] It wasn't clear to people that

[12:50] It was the new platform. In fact, it launched and there was kind of a hint that there could be apps of your own in that home screen. Mm-hmm. [13:02] But it actually took quite a few years for people to be like, yeah, this is the platform. We have to claim a square in that home screen grid. Yeah. [13:12] Not obvious at all. And I think you can think of AI as almost like another new, it's a new home screen. [13:18] And right now there's a handful of icons on that home screen. There is chat GPT. You could have like Google AI overviews, perplexity. And the question you should be asking yourself is like, how many more of those are there going to be in it? And my guess is probably millions. And so very exciting time for especially small teams like yours to be thinking about what are my pains that AI can answer. [13:48] from [13:49] Um, maybe like in the socket IO days, you have your finger on the pulse and you're the one who's solving your own problem to, um, thinking about how do I enable an entire organization of people to think that way? So you're more like thinking about the principles underneath like that, that mode of thinking, and then trying to infuse that in the culture. Yeah. Um, tell me about how that process has been for you personally. Like, do you like that versus like actually coding a lot or like, yeah, tell me about that transition. Yeah, it's awesome. So I believe that there is [14:19] Yeah. [14:20] And in the early days, you need to be obsessed with getting a high quality product out into the world and you live and die by product market fit.

[14:29] I think at some point you start realizing that you have it and there's other things to build to operationalize that product market fit, to make it into a machine, right? [14:38] And I would say like, that's the transition between a startup and a scale up. And think when you're a scale up, [14:45] whatever that term means, but I guess like a large, more mature startup. [14:49] you realize that [14:51] you no longer live or die by one individual product market fit. [14:55] you're probably thinking about becoming a platform or you are a platform and you're thinking about multiple products. And so you start becoming more of a machine that outputs products. You're like essentially a Y Combinator in the initial like recursive sense. [15:11] And [15:13] when you're creating [15:14] a machine, you still apply a lot of the thinking that you apply when you design products. It's only the company itself is the product. [15:23] And you're still worried about all the things that you worry about when you want to create excellent products. For example, onboarding. [15:30] If you're going to be hiring a ton of people, what do you want to have? You want to have an excellent onboarding experience, right? Just like when you first start using a new app, when people join Vercel, they need to be equipped with everything they need to succeed. So you start thinking about other aspects of the company. I think principles are. [15:48] are kind of like the sort of like product specifications of your company. And so thinking about what are our values, what are our ethos, what is our mission, what are the right people to be?

[16:00] to pursue this mission. [16:01] In terms of like that recursive product generation machine, I think a lot about what I call the recursive funder mode. [16:09] So there's founder mode. I believe founder mode fundamentally doesn't scale in the sense of if your aspirations are very large, the total output and creative output of a company cannot just be limited to the founder. Right. And so especially when you start having those ambitions of like. [16:26] being a company that [16:28] can nurture new founders within it. And so I think a lot about what is the DNA that we can incorporate into the company that is almost like, [16:37] the people that would start a company, [16:38] on their own. - Yeah. - But they can actually kind of do that within Vercel. [16:42] And so we select for specific trades. We offer very unique things. So we build a lot of cool open source technologies. People know us for Next.js, but we also contribute to Svelte, which is a very popular web framework. We acquired initially, but we support a tool called Turbo Repo. [17:01] that helps companies to scale huge code bases to the size of like Google and Facebook. So people can come to Vercel to fulfill their dreams, to reach millions of developers through open source. And so those are kind of the things that I think a lot about, like, you know, how could I... [17:19] find not just the next socket.io, but how can you create the environment that creates socket.io's and next.js and the ecosystems around it? I think that's really interesting. And the reason I'm kind of like keying on that journey from like doing it yourself to making the machine that builds the products is it's sort of this journey of like moving up the layer, moving up layers of

[17:49] is sort of the, [17:52] future of developers in an AI world. And I think that there's something similar happening in some aspects of being a developer where, you know, Cora, for example, we have this email product that I demoed for you. And Kieran, who built Cora, he didn't write like 80% of that code, right? He knows how it works. Like 80 to 90% of it was like written by O1 or a quad. And I think that's a [18:22] the AI is actually typing the stuff and you're thinking about what are the principles and what's the architecture that you want it to write. Yeah. Talk to me about that. Yeah. One thing that immediately comes to mind is that it does seem like people are becoming full stack product builders. Mm-hmm. [18:40] I don't think I would identify today even as a coder, even though that's [18:45] what I obsessed about for years. Since I was like 10 years old, [18:51] My ego. [18:52] And my identity became tied up with I [18:55] code. That's my thing. And yet, I think what I was lucky to have is also that aspiration to build products that [19:03] And that's what allowed for things like Socrates to happen because it was like, [19:07] I just want to make the web more real time and more interactive. It had like a product idea in mind that wasn't just like only in the code. I think it's an important asset to have. And when I look at people at Vercel today, I've been noticing that they're just more full stack. With VZero, for example, they can do design. They can bring context data copywriting into their creations that otherwise would have required chatting with other people and crowdsourcing ideas.

[19:35] We are going into a world where coding is a specific skill. And when things are specific skills, machines tend to [19:44] take them over over time. If your skill when you were a kid was like, I can do math in my head really well, [19:51] Awesome. Great asset to have. [19:53] I love it. But also there's a specific machine that can do that skill also very well and even better than you. And so what I try to separate is what are the meta skills that, [20:04] that are not as easily [20:06] replicated by machines that you should still nurture. And I think those tend to be more around [20:13] very high level conceptual thinking. Your engineer who, you know, automated the production of that product to like an 80% degree, [20:23] he still has a very good understanding of how all the concepts relate. He can prompt the right things to the AI. It's not that he just was like, okay, 06, please build me this thing. And then you just like checked out and like went on vacation. [20:53] contact enriched and organized from the start. From there, Adio's AI goes to work. It gives you real-time intelligence during calls. [21:01] It prospects leads with research agents. And it automates your team's most complex workflows.

[21:06] industry leaders like union square ventures flat file and modal are already building the future of customer relationships on adio go to adio.com slash every and get 15 off your first year [21:18] That's A-T-T-I-O dot com slash every. And now, back to the show. [21:24] The concept of symbolic systems comes to mind a lot for me because it transcends the [21:31] language, runtime, framework, coding, and that's [21:36] extremely important to have. How things work and relate to one another, [21:41] Interestingly enough, it's a skill that VCs kind of nurture a lot. Because I've been in rooms with a lot of venture capitalists that play by ear. And you can tell that they play by ear because they have... And by the way, this is a skill in its own right. And it's very impressive. They have extreme, what I would call like... [21:57] "token breath." I'm using token in the sense of LLM. They know 30 companies for the space of continuous testing, and they know 50 for LLM pre-training models. [22:11] And that skill actually is super helpful when you're prompting because you understand how the concepts relate to one another. You can point the agent to use a specific technology that might be the right solution, but you're not actually doing the coding. And so that's, I think, the biggest step that's going to play out in the ecosystem. Yeah, I think that I love the idea of meta skills and sort of one of those being like, okay, how do things fit together? How can I be a little bit more full stack? How do I think about things from end to end?

[22:41] Another thing that I've been playing around with is like I have this whole – [22:45] idea of us moving from a knowledge economy to an allocation economy, where in a knowledge economy, you're compensated based on what you know. In an allocation economy, you're compensated based on how you allocate the resources of intelligence. And that the skills in that economy that are valuable are the skills of human managers today. So like an example, which I think is kind of interesting, because I think a lot of developers are probably thinking, okay, is it even worth it [23:15] Um, like when you think about being a manager, let's say being a technical manager, um, [23:19] uh, you always have to, there's, there's this line between, okay, am I going to, um, be in the details and know everything in micromanage, or am I going to completely check out and like, just let them do whatever they want basically. Um, and both options are bad and you have to be able to know, okay, when is it important to be in the details and when is it important to just delegate? Um, but in order to do that, you kind of have to know the underlying like skill. Uh, [23:49] I love this concept, by the way, the idea that you're allocating resources, [23:54] in delegating to these agents is already happening. One of the things I think is really, really interesting. So Vercel has two parts broadly. One is our managed infrastructure. [24:06] we basically host websites. We deliver them through a global CDN network. We make it so easy for developers to deploy and build, et cetera, and that has a usage-based model. So for example, we host UnderArmor.com and OpenAI.com. If they have more traffic,

[24:24] uh versell costs more for them and if they have zero traffic it costs nothing which is awesome [24:29] On the other hand, we have V0, which is more like a design engineer tool in the sense almost as like VS Code or Figma would be and almost like if they had a child. [24:42] And those products you typically pay by subscription, not by usage. I think AI is actually disrupting that and it goes back to that idea of allocation. So I'll tell you a great example. You have users that are draining AI tokens and running these GPUs super hot day in and day out. They're like AI powered engineers. [25:03] And there's users who use it in a more casual way. They're like, oh, I'm on my phone. I want to create a personal app. I'm going to use V0. And what happens is that at the end of the day, we're seeing the first category of apps, I think, that are... [25:18] tools [25:20] that have a consumption billing model attached to them. Because what you're doing is not, you're not simply using it as a system of records, it's not like Salesforce, you're actually, [25:30] "setting machines in motion to perform tasks." And I like that metaphor, it's almost like you're doing management. [25:38] in capital allocation, okay, how much... [25:41] computation. [25:42] "Am I gonna allocate to perform this given task?" And you even have to think about risk, right? [25:49] It's true that AI is really exciting and capable. [25:53] But sometimes it's not that capable, right? So you have to think about like, am I going to burn, you know, like lots of inference time compute

[26:00] Am I going to set this entire like 500 billion data center on fire to try to solve this task? And I actually don't know if the AI will perform. Is this on the customer side or on the like? It's on the customer side. So like the user of AI will have to think. [26:15] almost as if they were a cloud infrastructure user. I think it's a leap that is maybe the closest thing to that in today's sort of [26:24] product development world would be CICD, where a given engineer can say, okay, I'm going to run 100 machines to test my program. [26:33] But it's a lot less fluid. It's a lot less elastic than this kind of thing where you can sit down and say, okay, five agents. [26:42] Go at it. Yeah. [26:44] And it might or might not give you the right result. And so you have to think like a capital allocator and you have to think like a manager of these agents. Yeah. What that makes me think of is like, yeah, in this world, you're much more likely to, it makes much more sense to pay for tasks to be done. Yes. And for the price that you pay to reflect the level of risk that the task will get done well. 100%. Which is like hiring any other human for that matter. Exactly. I go to Upwork and I'm like, I don't know if this is the person, right? [27:14] the task for these agents. - Exactly. And then what that makes me think of is like, well, at that point, like, [27:19] If you have a market where like different agents can bid from different companies to like do any particular task, that's like actually really interesting market. Yes, 100%. Yeah. Because I think also that we're going to see specialization in agents. Some will have better context. Some will have better taste. Yeah. A lot of that deals with like the post-training process. And so there is definitely a marketplace and a competition that happens in who is the right agent for this problem. Yeah.

[27:49] Just you saying, I don't think of myself as a coder. [27:53] Um... [27:54] I think that's so interesting. And also, you serve developers, right? So where do you think that term goes over the next five or ten years? Yeah, you know what's interesting? Coder, as a word, used to be more popular in my community. [28:10] Even before AI, people were starting to realize that their job was not just to introduce keystrokes into an IDE. [28:19] There's a history of this. One example that I like to give is before [28:26] Automatic code formatting tools became popular, [28:29] People used to care a lot about what their code looked like. [28:32] on the screen. [28:34] How many columns? Are the words neatly aligned? People would have discussions about this. It was actually considered to be a productivity drain, but it was a huge subject of debate within companies. Now we look back and it's like, of course, people should not be discussing if the aesthetic arrangement of the code is right or not. And so people introduced this tool called the formatter. Mm-hmm. [28:59] I think, I don't know who the exact [29:02] project that popularized the most, but I remember when Golang came out out of Google, [29:09] It came with very simple code for automatic code formatting. [29:12] And then in our community, in JavaScript and TypeScript, Prettier came out. It's called Prettier because it makes your code pretty with no discussion, like objectively Prettier. And then that whole category of quote-unquote problem

[29:26] disappeared. And I would say, look, [29:28] format. [29:30] your code is formatted, that felt appropriate to the word coder. [29:35] It's just a person that makes their entire purpose to write code. [29:39] And then we kind of, I think developer, [29:42] engineer, [29:44] Then product design, which is a role that sort of came out of Facebook, became more important. So the trend has been away from it [29:52] the implementation detail, which is the code, and towards... [29:56] the end goal. [29:58] which is to deliver a great product or a great experience. And it's in that context that our company has become popular because one of the things that we came to market with is like start focusing more on the front end. Start focusing more on the customer experience, on the site speed. We've demonstrated to companies that you make your website faster [30:18] You just sell more things. You make more money. So we started trying to like [30:21] divorce ourselves from the code and put the focus on the experience. Okay. Yeah, that makes sense. How do you [30:28] Talk to me about how you see V0 fitting into that and just generally where you see V0 sitting in the whole ecosystem of, there's like Replit Agent, there's like Cloud Artifacts, there's all these different, there's Cursor Composer, there's all these different tools. Talk to me about that. Yeah, it's exactly in the context of that idea, right? We look at V0 as code last rather than code first. When you come to V0, you prompt with your idea or your intention.

[30:58] So it can-- [30:59] bring your designs alive. That's for companies that already kind of relate to that mode of working. VZero will make a lot of sense that way. But you can also just come with an idea. Build me this thing. You can... [31:12] We're about to introduce templates. You can already kind of start with, for example, an AI chatbot. Like you can use AI to like produce AI products. And so the biggest difference with those is that those started with the code. They look at themselves more like IDEs. And we look at this more as a product development environment that merges what you would expect from a design tool because the things that VZero produces, we aim to make them great. [31:42] All of the things that I've been trying to teach our customers in a less scalable way by means of customer support, I would say, or professional services, you could call it, or enterprise agreements where a big company will come to us and say, wow. [31:59] look, you're the experts in building storefronts. Help me have a killer storefront for Nintendo. And [32:07] a lot of that process is just manual. Like, ultimately, we would have to... Of course, we had the framework. ConnectJS tries to set up the right foundation. But on top of that right foundation, there is a ton of business logic that needs to live. So the idea of vZero is... [32:22] Could I partner you with an agent that is an expert in web development, has taste, and produces things that we're legitimately proud of? Which is hard to do in the AI world because AI has become a little synonymous with, well, you can get a lot of slop from different products. So we're very focused on the outputs have to be great. One of the things that – so I had Nabeel from Spark on the show last week.

[32:52] is when he looks at these sort of agent products, there's every agent has to make a decision about how long the leash is between the user prompt and when the agent like gives a response back. Yeah. So like Copilot's a very short leash, right? Devin is like a very, very long leash. Totally. Her cursor is like sort of somewhere in the middle maybe. Yeah. How did you think about that? And how do you think about, yeah, like when I put something into V0, right? Like what my [33:22] something like immediately and like give me the result. How do you think about that versus other ways that you can, you know, maybe check in halfway through or whatever? Yeah, it very much depends on the task. I believe that the future of AI will be sort of the classification of an estimation of the difficulty of the task and then taking different strategies, but also getting feedback from the real world. So a concrete one is when you give us a prompt, you're going to give us a [33:47] We have to give you feedback right away because you're almost in the taste making part of the part of the job. It's it's like you hired an interior designer and like they're checking for your vibes of like, yeah, OK, what do you like? Let's get deeper into this. But then you have very specific functional requirements. I've sometimes told the zero make this faster. Yeah. And it does. And sometimes actually it's taken me a few prompts. No, no, make it faster again. Mm hmm. [34:15] Like it's amazing how good it is at optimizing React code. So I had a very intensive animation that it rendered. And it was just expensive in how much it was recomputing. So I literally just told it, make it faster. And then I told it, make it faster again. And it was all vibes from my side. This is why I was telling you the coder thing is interesting. It's strictly less important because I didn't care how. It just made it faster.

[34:39] And so I would expect that for that type of task, we can give Visero a longer leash. And I think [34:46] there's a precise technical distinction is one feels like more in the one shot world where you're getting knowledge plus rack data out of the LLM. And the other one is more the feedback loop of thinking that can have an arbitrary length. Yeah. [35:04] And that to me is... [35:06] One, the thing that has allowed us to increase the quality of the product, vZero actually fixes a huge amount of error that comes out of the models. [35:16] It has error correction loops because it looks at feedback behind the scenes. Another careful balance that we need to strike is how much do we expose of that process to the end user? You kind of want to learn, but you don't want to be overwhelmed by everything that the agent is trying. So we just... [35:35] We have an, I mentioned earlier, we have a product called AISDK, which is kind of like the infrastructure underneath VZero and it's open source. And companies can use it to build their own agents and their own products. And we have a playground for it. And we just added the DeepSeq R1 model. [35:53] which people are raving about right now. And you can see when you use that model, [35:58] that you can see every thinking token. And it's actually completely overwhelming. It's so overwhelming that it's, and the classic demo that people are doing is they count the number of Rs in a strawberry. And then you see it think through that process and it's overwhelming that,

[36:17] is nauseating. It's almost like peeking into the mind of a mad person. And so that, and now extrapolate and bring that into your product. Like you don't want to emit every thinking token into the user, but you kind of want to give them an overview of why the leash is longer and how the agent is using your time and resources. One of the things you made me think of in this sort of like transition from coder to something else. I don't know what word would you use? I like to use [36:45] Product engineer or product developer. Another role that has emerged a lot is design engineer. I love design engineer because... [36:53] I care a lot about design and I want to make things work in the real medium. So those are the two roles that are sort of emerging. Okay, so let's say like product engineer and design engineer, from coder to product engineer, design engineer. One of the things that I think that transition reflects is if you have machines that can build things really cheaply. In a world where that was really expensive, you care a lot about how all of that works. And you want to do that by hand because you want to like optimize that really, really well. [37:23] really cheap, you go from caring about how it's done to how the end result feels, which is why you want that rapid prototyping loop. And it occurs to me that that's actually also like a really long running process. [37:38] I don't code an assembly or I don't do garbage collection. Totally. It can even – it's so funny. Garbage collection is such a good example because when people –

[37:49] naively try to do memory allocation to make things faster, sometimes they actually underperform garbage collection. It takes a lot of skill and knowledge and a domain-specific task to get the right ROI out of a manual memory management. Yeah. And sometimes that's useful, but more often than not, it's not today. But 10 years ago, if you use Python or JavaScript, people were like, I don't know, that's not real programming. And I think today probably something similar is [38:19] Python languages. I have a lot of thoughts on this. One is... [38:24] AIs need to, today, let's call it this product development AIs, need to sit on top of significant infrastructure for them to be useful and productive. They need to be on top of a platform. That's why it makes sense for Vizier to sit on top of Vercel and Next.js. Because think of it just from an economics point of view. [38:44] the AI either [38:45] re-emits all of the code necessary to render the thing or it says I'm going to use Next.js and I'm going to render the thing with it. One would take you [38:55] millions of lines of code of tokens to output if you needed to reinvent the framework [39:00] The other one might take you like 100 lines of code to produce a really good result. And so you have to think about AIs as a collaboration with infrastructure and platforms that already exist, most of which are human made. Very much like a Waymo self-driving car needs to interoperate with the real world.

[39:20] We couldn't modify the streets and say like, we're going to build new streets for self-driving. No, we needed to put the cars on top of that infrastructure. And so... [39:29] Another thing that your comment made me think of is, [39:32] Thank you. [39:33] It is true that [39:35] there are certain skills that will yield, for example, better performance or better safety, such as using Rust instead of using JavaScript. Now, my question is, is optimizing something in those languages, something that AIs will be able to do in the future? Everything seems to indicate that that would be the case. Because when you use these languages, you're also respecting very [40:05] and I think it's going to be the case that AIs can do a lot of that optimization work for us better than a human could do. It's going to take a while to get there, but I'll give you another concrete example. There's a project out of Google called Project Zero that seeks to unveil security vulnerabilities in the entire world, not just Google's products. And it's some of the most cracked security engineers on the planet. And there's a category of bug that they always find that is in C code. There's [40:33] memory corruption bugs that lead it to [40:35] catastrophic security vulnerability. Sometimes it's one line of code that you get wrong. And in our global encryption infrastructure, OpenSSL, such a line of code was found that did a very unsuspecting MCPY, one line of code that basically made the entire universe vulnerable to a really, really, really bad vulnerability.

[40:59] And now the problem is we know that C is not the right language for that kind of infrastructure. It's just so easy for human beings to sneak in, [41:08] those vulnerabilities, either by mistake or tinfoil hat on, some like dark secret agent. It's just six one line to create a massive backdoor. And so the problem that we have is how do we rewrite [41:20] all of that C code in Rust, and I know that this is a meme of like rewrite everything in Rust or like any other memory safe language for that matter. Will AI agents do it? [41:29] My bet is yes. Will AI agents help us uncover those vulnerabilities? Because [41:36] Google has maxed out on how many cracked engineers that can sort of like read all of this world code and identify those specific bugs. So it's going to be exciting to see that AI will also have a very profound perspective. [41:47] Impact on the underlying infra, but it'll be a different kind of task. I think another way I think to say what you're saying and you tell me if you think this is right is for any task or [41:57] where you can quantify whether the task is done well along one dimension. A fitness function. A one well-defined fitness function where the steps are verifiable. Yes. That is going to be very easy to automate. Yes. And then anything else that's much higher dimensional, which I usually call taste. Yes. Sort of like this aesthetic thing that's like, this actually just feels good. Yes, and it has social effects. Yeah. Right? That's going to be a little bit more... [42:27] Yeah, you just need to figure that out. You need to be the vibes guy above the machine. Yeah. How do you cultivate that in yourself?

[42:36] Yeah. [42:37] A lot of practice, I think. I remember one of the best engineers I ever worked with, he was also a very good designer. That's how I kind of became infatuated with the idea of like, we can do more. Common wisdom in Silicon Valley was like, you're either like a knee deep in the data center room and writing back in code, or you're the... [42:57] designer, like super polished, like mustache, you know, like you can do both. And, um, uh, this guy, TJ kind of taught me that. And I remember having a conversation with him and he was like, well, look, I just, [43:09] train my taste a lot. [43:11] I look at a lot of things. I look at what people like. I seek a lot of feedback. So I do think that you can work on your ability to discern what people are going to like. There's obviously some rules underneath. And this is what we're trying to sort of uncover with VZero. It does turn out that, you know, how you distribute spacing. [43:30] on a page does lead for more pleasant outcomes on the viewer's eyes. And we try to sort of like yield those out of the box. But I do think taste can be worked on and sort of like the higher dimension, like the metaphor you use of like bringing in inspiration from multiple sources and multiple things, like thinking like radically outside the box can be super helpful. And then the [43:53] understanding what's possible and what you can sort of like start to push the boundaries on so [43:58] It's kind of funny, but like being terminally online, kind of like answers for this, for these two things. One is you want to consume this stream of consciousness of like what good design looks like and what the avant-garde is. But you also want to be at the forefront of like.

[44:16] What's about to become possible that I can start cooking on now? Kind of goes back to the beginning of the conversation with Sock.io. I became aware that a real-time communication channel API was about to become possible in every single web browser on the planet. [44:33] and I could start cooking on it [44:35] before it was real. - Right. Like, tell me about that now. Like, what are you, what are you, what's buzzing in your ear right now that's about to become-- - Yeah, we kinda touched on it in a few, it's like this idea of like, [44:46] domain-specific agents that... [44:49] are infused with taste and tools and knowledge for [44:53] Even tasks that you would think, okay, maybe I could solve that with chat GPT, [44:58] but it's a ton of work and maybe the result is not that great. And it can like infuse a lot of quality into that vertical. This idea of agents that have the long horizon like work, but also can give you rapid feedback when necessary. I think there's a big opportunity there. Broadly, I continue to think of AI as this like iPhone home screen on top of which we can deliver lots and lots of applications. [45:22] And, [45:23] Yeah, I think the the idea of [45:26] ChatGPT4X continues to surprise me. I just heard from a, I believe it was like, [45:32] cardiac surgeon that said, [45:36] I basically spend a lot of time during my professional life using not JGPD, but this tool called Open Evidence, which is... [45:46] infused with all the health

[45:48] care data used by 250,000 medical professionals in the U.S., [45:53] And it kind of makes you think, right? Like if you're a medical professional, are you going to go to the general purpose AI that is sort of like the lowest common denominator for everything? Or are you going to go to the thing that is constantly improving for your domain and field? So encourage entrepreneurs to think about it that way. And also... [46:11] find the opportunities to bring creativity into [46:16] tools that are otherwise pretty plain. Of course, Copilot could autocomplete code in your IDE, but V0 took that way further, as you mentioned earlier. Like it's not just about the little bit of autocompletion, [46:28] bring a lot more ambition and creativity into what the AI agent can do. Yeah, I think there's all this talk right now, or it's probably a little bit less so now, but especially like a year ago where everyone was like, well... [46:39] incumbents are just going to win. Like if you have access to AGI, like distribution is all that matters and blah, blah, blah, blah. And I don't, [46:47] think that that has panned out and I don't think that will is not that will pan continue to pan out because like part of what you're saying is like for example the doctor issue is like it's not just that maybe oh one or chat GPT is like not tailored for doctors it's that it has to be a general purpose tool so it has to say like I'm not a medical professional right and like you probably and I and it has to have guards against like what it there's almost like a principle they're like every refusal of chat GPT is [47:14] can be turned into a tool that doesn't have to refuse, because it's actually going the extra mile

[47:20] Whether it's with better data, with, you know, maybe even human in the loop, whatever it is, every refusal is my opportunity. Exactly. Yeah, I love that. And I think another way to say that is startups get to take risk and the intelligence that is available is not necessarily a technical limitation anymore. [47:50] And so they have to take lower risks. And in terms of disruption, there's a very concrete one that I think a lot about. [47:57] I [47:58] The history of SaaS has been, I started out creating a product that's really good in its category. And then I expand its feature set to beat out my competitors in enterprise bake-offs. And so you become a competition of checklists. And there is a precise correlation between product maturity, enterprise maturity, and amount of UI. Like you can actually quantify the number of buttons that's, [48:25] The number of dropdowns, the number of nested dropdowns. And so all of that UI can vanish without a trade-off in functionality if you're thinking from first AI principles. [48:38] And that gives the startup a very concrete wedge to attack the incumbent. I'm going to... [48:45] Under do you. [48:47] I'm going to underwork you. I'm going to under UI you. Right? And I actually will beat you on functionality. Because if you're solving problems with classical engineering alone,

[49:00] And if else branches, this is, I love the Carpathie, software 1.0 to software 2.0, [49:05] Software 1.0 is if else invert binary tree and algorithms and data structures and AI is like, well... [49:16] Maybe that computation ends up happening, but we don't know why. [49:20] And so we have that agility that comes. We're searching through the space of possible programs to solve the problem instead of like having a recipe. And that's a concrete way of underdoing the competition. Yeah. Because you're just having the AI produce better or equal outcomes with a lot less manual code. Yeah. And I think that that's such a, it's also an important like thing that I think startups sometimes get wrong in this era, which is like trying to do, let's say like an email product or like, I don't know, any classic like productivity software. [49:50] like PowerPoint or something like that, where what you're trying, what you end up, some products end up doing is doing the classic version plus AI, right? [50:01] like plus a chatbot or whatever, which is exactly what incumbents do too, where I think like what's what you actually want to do is figure out, okay, what's the totally new norm factor? And how do I be worse on all of the checkbox dimensions, but so good on the thing that matters that- 100%. Actually, I would say there's two models that I think a lot about. One is actually you can be better at eventually every checkbox, but in the meantime, you're overwhelmingly better and

[50:31] Mm-hmm. [50:32] do more because companies just want to sit like another thing I've learned with Vercel and [50:39] when you go to enterprises and larger companies they want to buy only a handful of software platforms this is why microsoft and in amazon and others are so prolific it's like they provide like broad solution spaces yeah so eventually you need to get there but i like your idea of like in the meantime like you're just like a killer yeah at a specific dimension you're way simpler yeah and perhaps you'll never have to have the ui complexity of those incumbents [51:04] How does that sit with you? Because you're just stylistically like a minimalist simplicity guy. And Vercel like kind of has to be a little, if you want to be enterprise, you have to have all the checkboxes or something like that. Like, how are you thinking about that? Yeah, you can think of V0 as almost like the AI native interface into the Vercel world. And I expect the next billion developers to sort of come in through that door. And then there is the eject button into Vercel. [51:34] - It's so much like Vercel, [51:36] kind of did that to AWS and people still need AWS. And we have great integrations and many more coming where we, [51:44] create a bridge to the lower level. I like your concept of like, you're basically like raising the abstraction bar very much like we did with manual memory management and garbage collection. We're saying, look, at the very, very bottom, there is the rock solid foundation. There's a bedrock, literally, of hyperscaler clouds like AWS.

[52:02] And then you have the developer infrastructure on top. I believe most of the world will want to interact with developer infrastructure because you don't want to go to the raw materials and do like alchemy there. And the success of ourselves speaks to that. But now there's another level. [52:17] It's almost like a self disruption that we're doing with VZero, which is like, look, most people are going to want to interact with these agents. They're going to make it more accessible, more design oriented, less infrastructure oriented. And what's nice for us is that you still have the two. [52:32] sort of like levers to pull. Some more mature organizations say like, look, [52:38] I-- we're not ready for AI. [52:40] I've talked to folks in like super regulated markets where we're running a lot of their web projects. And they were like, please don't say the word in my office, like literally like during an onsite. [52:53] And then there is the people that are like, don't say the code. I actually, fascinating. One of our enterprise customers for VZero. [53:02] is a company that, without giving away too many details, has been in the cloud space since the cloud was born. Like literally like in the room where the cloud was invented. [53:13] And they told me, "Our new organization is just a VZERO organization." [53:18] We have it as a rule that we don't do code. Most, they showed me their Slack conversations, it's just them sending these zeros back and forth. There's no GitHub, there's no infra, there's no anything. There's just building products, going back and forth, and having shared channels with their clients, where they also send them these zeros. And so they're trying to even innovate in how,

[53:38] company altogether works. And so there's people that are going to live in that level of abstraction. Just like there's so many companies that only use Vercel. There's companies that combine both. A lot of enterprises use Vercel. [53:51] It can connect into AWS to a product called Secure Compute, and it kind of lives in two worlds. [53:55] Most of the startups live only in the Vercel world. And so it's gonna be interesting to see like, [54:01] And perhaps is there another level of abstraction coming where like, [54:06] the result comes to you in a much more... [54:11] sort of, [54:11] objective way in your [54:14] giving less feedback yet to be seen. That's interesting. Yeah, one of the things that that sparks for me is, um, I watched I watched a talk by Eugene way, that he did it, we ran a conference called thesis, and he, he spoke at it, and it was an amazing talk. And he was talking about, um, [54:31] the different kinds of company cultures and how that affects the kinds of products you can build. And he made a distinction between a written culture and a prototype or demo culture. [54:42] And in a written culture, which he would say like Amazon is a written culture. I think Stripe is a written culture. Oh, interesting. [54:49] um you're uh like you're building products that can be expressed um cleanly in in writing right yeah and what that lends itself to is if you look at like aws for example it has all the specs like all of the all of the like logic and rigor is there but like the amazon phone like just feels like shit because it like like it has the like specs but like it doesn't feel good right i think

[55:19] highly available, secure infrastructure and what goes into making that sausage. Exactly. And that's an output of like being able to have everything written down versus he makes the example of Apple, where like the early iPhone, like Apple is a prototype culture. Like every time you do something, you go and give a demo to Steve and Steve like rips you and then you like make it better. Right. Yeah. And I think V0 and this company you just mentioned, like it allows us to like move. To live in that Apple world. 100%. It's so interesting. I think to each their own, [55:49] like for each [55:50] role that you play in the ecosystem, you need to like lean more towards one or the other. So for example, AWS, not only do they have a written culture, they have an automatic reasoning team that writes TLA plus proofs of correctness of all their cryptography and durability. I can trust them with my clinical records. I can trust them with like anything. And I know that those who have like written mathematical proofs of the certainty of that code and the distributed [56:20] Thank you. [56:20] you need tasks that are so creative that imagine writing a mathematical proof before you sit down to like cook on a product yeah and so [56:30] What I like about Vercel is that I think we've blended both. [56:34] We have so much respect for that work that sometimes a lot of people ask me, like, do you want to build your own data center? I said, like, do you even realize what goes into getting to that level of quality? My job is to procure the best possible components, just like the iPhone bundles a lot of incredible hardware that you don't know about, but you know that it's the best at doing that task. And then I want to enable you to live in that world of, like,

[57:01] vibes almost right like prototyping creativity velocity performance etc yeah that's great um i think that's a that's a good place to leave it um this is an incredible conversation thanks so much for doing it that was fun we'd love to have you back will do cool [57:25] Oh my gosh, folks. You absolutely, positively have to smash that like button and subscribe to AI&I. Why? Because this show is the epitome of awesomeness. It's like finding a treasure chest in your backyard. But instead of gold, it's filled with pure, unadulterated knowledge bombs about chat GPT. [57:48] on the edge of your seat. [57:49] craving for more. It's not just a show. It's a journey into the future with Dan Shipper as the captain of the spaceship. [57:57] So do yourself a favor, hit like, smash subscribe, and strap in for the ride of your life. [58:02] And now, without any further ado, let me just say, Dan, I'm absolutely hopelessly in love with you.