Elad Gil: Silicon Valley’s Most Dangerous Startup Advice

Transcript

There's a bunch of conventional wisdom in Silicon Valley that I either think is always wrong or mostly wrong. An example of that is you always need a co-founder. Michael Dell didn't have a co-founder and Jeff Bezos didn't have a co-founder. What is the most common mistake that you see AI startups making nowadays? Your thing isn't working and you stick with it too long. There's radical openness to trying things that didn't exist three, four years ago. And that's really important in a way that I think few people really understand. Very excited to welcome a close friend, a close Polygon Ventures friend and a close friend of SBC here for a minus one five side chat. Ilan, I think you are the first guest who's coming here for the third time. Wow, they keep inviting me back. I don't know why. Thanks for you, Frank. Keep raising this big fund so we don't need to invite you back. So Ilan, we've been here three times. I guess to start off, you have a bunch of folks in the room at various stages of ideation, figuring out what they want to work on. And you know about the minus one phase, which I would say is a stage you've kind of started companies. You've been in and out of companies, started a new venture fund. But the first set of problems I want to talk about is how do you do minus one in today's day and age that might be pretty different than five years ago? When the capabilities for building seem a lot more pronounced, but also what you build is in some ways a lot more commoditized. So how would you go about doing minus one today if you were starting a company? [SPEAKER_01] Yeah, I think there are multiple ways to start a company. [SPEAKER_01] Nothing goes like a single prescriptive approach or a single right thing. I think fundamentally there are two or three different ways that you can approach things that are still pragmatically true. One is obviously being very customer centric. Sometimes it's building for yourself as a customer. That's a brain trust. You know, Ankur Goyal's company where it's an eval and prompt the ground and tool. And he built it for himself effectively as a potential user. That's something we kept beating over and over. You can also do that for specific customers. So one approach is doing that and being very customer centric or honestly self-led centric. Second approach, which I think is really interesting in today's day and age, which I think is very unique, is I'm starting to see people do these AI driven rollups or buyouts where they're effectively buying companies and then expanding margin dramatically using AI. And there you need three skill sets. You need the ability to buy things. You need the ability to operationally change them, which is often the marketing part, the change in management. And then you do the AI stuff on top of that so you can actually expand margins and make it a better business and make it more software-like. And then there's a third which I know most of you is a grab bag, which is you iterate on different things. You invent things. You run it by customers. You try customers. That's almost a traditional approach and it's a little more of a winding road in some cases where you come up with an idea and then you try and embed it and see what people think. [SPEAKER_00] So, okay. So let's perhaps say our goal page would take a slightly different approach. [SPEAKER_01] And by the way, sorry to interrupt. I do think things are faster now. I was going to say that. And so I think that part is dramatically faster and better. And then I'd also say it's faster in two ways. One is you can build things much faster, which makes a big difference. But the second way is customers are very interested in trying stuff in a way that I haven't seen in my lifetime. And because of AI, every CEO has the edict of, you need to do something in AI. What are you going to tell your board about AI and how are you going to do AI? And so there's radical openness to trying things that didn't exist three, four years ago. And that's really important in a way that I think few people really understand. So I remember when I was diligencing, I invested in Harvey, the AI legal company, really early in their first round. And then I led their series B. And for Harvey, I remember calling their customers as due diligence. And I called these big law firms and I asked them. And usually law firms are really bad adopters of technology. They're very slow. They're hesitant to change. There are big security reviews. Many legal laptops are completely locked down, right? You can't download things onto them without CIO permission at the law firm. So I call these people running these law firms, the ten biggest law firms in the world. And I asked them, what is the biggest change you expect to the party and why are you adopting it and all the rest? And even two, three years ago, the insight was so interesting and unexpected where they said, we think this will augment associates to the point where we'll have fewer associates for the same book of business. We can grow our book of business quite a bit. And we grow partners from associates. So who are going to be our future partners? Are we going to have enough associates per partner to actually have the right number of partners? Like it's part of the training. It's part of choosing people. It's a selective criteria. It's the filter. Whatever you want to call it. And I thought that was fascinating as an insight. And this is an insight like three years ago or whenever it was two years ago. And so I do think there are other aspects that are really overlaid on this. They're super fascinating. [SPEAKER_00] Would you say that as an extension of that observation that there is more openness to buying today than has probably been in our generation, that if your stuff is not taking off today, if you're having a hard time selling your thing today, then that's a pretty bad spot to be. [SPEAKER_01] A hundred percent. [SPEAKER_01] And I take it a step further. I think there's a bunch of conventional wisdom in Silicon Valley that I either think is always wrong or mostly wrong. And an example of that is you always need a co-founder. And you look at the biggest market caps in the world. And Michael Dell didn't have a co-founder. And Jeff Bezos didn't have a co-founder. And you go through company by company. [SPEAKER_00] If your stuff is not taking off today, if you're having a hard time selling your thing today, then that's a pretty bad spot to be. A hundred percent. And I take it a step further, which is even a couple of years ago. So I think there's a bunch of conventional wisdom in Silicon Valley that I either think is always wrong or mostly wrong. And an example of that is you always need a co-founder. And you look at the biggest market caps in the world. And Michael Dell didn't have a co-founder. And Jeff Bezos didn't have a co-founder. [SPEAKER_01] And you go through company by company. Steve Jobs was a dominant founder. There were two founders, but he was really the main one. Although it wasn't built everything initially. Initially, but your company, my company, either it's unequal in most cases until YC, right? YC and Google was roughly equal to Larry had a little bit more stock even early on. [SPEAKER_00] Larry Ellison, Mark, a local founder. No co-founder, right. So you look at many of the biggest things: Bill Gates co-founded with somebody, but then he really became the sole founder when Paul Allen left. So over and over it's either unequal or solo for many of the biggest things. [SPEAKER_01] Now there's lots of people who are founders who have been incredibly successful. [SPEAKER_01] In some cases, equal. In some cases, unequal. If you actually look at cap tables as companies go public, you actually realize that most things were unequal. And it was really YC that normalizes equality. So anyway, it was a lot of conventional wisdom that I just think is wrong. [SPEAKER_00] I mean, we don't love equality. [SPEAKER_00] We love winning, right? Yeah, I think winning is important for startups. People often ask me, what is the single biggest determinant of culture in a startup? And I say, winning. It's not the kombucha, it's not the ping pong table. It's not the number of GPUs you have. [SPEAKER_01] That is exciting. [SPEAKER_01] It's winning. [SPEAKER_01] But anyway, back to the original question. [SPEAKER_01] The companies that we said I've been involved with that have tended to work, have tended to work early. [SPEAKER_01] And it's not that we're going to grind for 70 years and then the thing will work. [SPEAKER_01] It's usually something starts working really fast. Yeah. It's actually interesting. You bring up that particular state. And one of the precise lines I remember from when you spoke last was actually that for startups, they can often be wandering around for a while. But when stuff works, it tends to work very quickly. [SPEAKER_00] And there are some cases where success happens on a slow curve. [SPEAKER_00] But the best companies, when it starts turning the flywheel, it turns even quicker. [SPEAKER_00] So I think I'm combining the two things where I'm really hearing is that if in today's day and age, you think you're building something or you're having a hard time selling it, that's a pretty bad signal. [SPEAKER_00] Because I agree that there's propensity to buy across almost every vertical, including old school regulated verticals that we haven't seen in a while. [SPEAKER_00] And also, if it's working, it should be working quickly. [SPEAKER_00] Right. Yeah, this actually ties into competing schools of thought in Silicon Valley around whether there are enough founders in the world. So the YC school of thought is if you just have more founders, more magical things will happen. And the alternative viewpoint, and I think both are partially correct, is there's only so many markets that are open in a given moment in time. [SPEAKER_01] And therefore, there's only so many markets in which you could build a big company. [SPEAKER_01] They're open because of a regulatory change. [SPEAKER_01] They're open because of customer behavior. [SPEAKER_01] They're open because of technology shift, right? [SPEAKER_01] Right now, we're going through a simultaneous massive shift in technology and a massive shift in buying behavior. [SPEAKER_01] So many more markets are open right now than I've ever seen in my lifetime. One of the frameworks we talk a lot about at SPC is modality one or two. Modality one is that it doesn't actually matter if it's good or bad. The only thing that matters is how pretty it's drawing, right? And often in those kinds of domains, there's often not just one player. There's often two or three. It's often pretty neck and neck in the early days. But really, the only thing that matters is actually winning. And modality two is you're actually pressing something heretical or non-consensus. But by definition, that should mean that you should find it hard to get funded. You probably shouldn't have most people laugh you out of the room, right? Or they should say that this actually doesn't make any sense. [SPEAKER_00] And I think a lot of people want to do modality two. [SPEAKER_00] But they're still seeking validation, whether it be customers or employees or investors. [SPEAKER_00] And I think modality two is actually very lonely. [SPEAKER_00] And I think there's not that many companies that I think are able to pull that off. [SPEAKER_00] I'm curious. [SPEAKER_00] Have you seen any good modality two companies? There's a few. And there's also a lot of stuff that took off in ways that people didn't anticipate, right? OpenAI would be an example of that. But also ChatGPT. It was launched Thanksgiving week. [SPEAKER_01] It's a buried launch. [SPEAKER_01] November 2022, right? [SPEAKER_01] It was, oh, nobody's going to pay attention to this thing. [SPEAKER_01] So we launched it during Thanksgiving. [SPEAKER_01] So what does that say about our ability to predict what's going to be compelling? So today we have three to four players. And let's call it three players who are dominating the frontier. I will say four. Because obviously, OpenAI is Anthropic. There's Google. [SPEAKER_00] And then we'll put all the Chinese open source models into their own bucket. [SPEAKER_00] So let's say there are four big categories. Do you think that this is a stable ecosystem? Or do you anticipate this shifting, call it in a five-year timeframe? [SPEAKER_01] Yeah. [SPEAKER_01] And then I probably had XAI as an ongoing wild card. Actually, sorry. You're right. I mean, if you want to discuss that, you should go raise. Because obviously, OpenAI is Anthropic. [SPEAKER_00] There's Google. [SPEAKER_00] And then we'll put all the Chinese open source models into their own bucket. [SPEAKER_00] So let's say there are four big categories. [SPEAKER_00] Do you think that this is a stable oligopoly? [SPEAKER_00] Or do you anticipate this shifting, call it in a five-year timeframe? [SPEAKER_00] Yeah. Yeah. And then I probably had XAI as an ongoing wild card. [SPEAKER_00] Actually, sorry. [SPEAKER_00] You're right. [SPEAKER_00] I mean, if you want the beans that was happening like that, you should go raise. [SPEAKER_00] Yeah. [SPEAKER_00] As in that, Grok is the best of that. [SPEAKER_00] Yeah. You know, so I wrote a post about this maybe two or four years ago where I tried to predict the foundation model market. And at the time, what I thought would happen was that you'd end up with three players, each of which would be aligned with a hyperscaler. And it'd be a captive hyperscaler model relationship where the hyperscaler would fund it and then the key on that platform. And so I thought it'd be Google for itself with GCP and then OpenAI and Microsoft, because there'd be a relationship at the time. Then I thought Anthropic maybe partners with Amazon or something. And then I thought NVIDIA would be the primary funder of open source. And if you look at every technology wave, there's usually a big company that funds and monetizes the open source. So during the nineties for Linux, IBM spent billions of dollars effectively on Linux development and really monetized it by selling it through services. And I got it wrong. I got it partially right and partially wrong. I got it right in terms of thinking it'd be in an oligopoly market. And the reason I thought it would be in an oligopoly because of the scale of capital needed to do it. So eventually you just don't have new entrants where you're able to compete, but I got it wrong in a few ways. One is Meta and the US-based funder for the time being. And China, the Chinese government became the other major source of economic funding for open source models. And then I got it wrong in terms of the models partnering more aggressively across right in traffic as part of the GCP and Amazon. And Microsoft is working with in-board for models. So Google is working with Anthropic. So it's much more of an overlapping world than what I anticipated. And everybody kind of cross and busted. It's actually honestly closer to what happened in China across a few different waves of the big Chinese companies effectively funding the versions of Uber and other next-gen applications. So I do think the natural structure, assuming one of the companies doesn't truly lift off in terms of model capabilities ahead of everybody else's oligopoly, that seems at the natural state for now. I do think one interesting thing that's emerged is the importance of the harness relative to some of these applications like Claude. I think a big shift there that I observed was when 5.2 came out for OpenAI, there was less switching than I anticipated because some people argued it was a bit of a better model, but people had already been using Claude in a harness that was really optimized for using Claude. And so fewer people switched than I expected would. It could be the model's only incredibly better. It could be a variety of thinking. So an open question to my mind is how much does the harness matter for sticking this now versus the core model? And if that transition happens, then that has really deep implications in terms of how you think about things. And then the other piece of it that's the interesting question is when do the certain types of models asymptote and functionality? And if you hit the asymptote, then other aspects become more important in terms of switching or not. If you don't hit the asymptote and it continues to be what's the best model, again, with this Uber of Leo, is it significantly different enough that you'll abandon whatever harness you're using? And then the question is like, what's the enterprise version of the harness? [SPEAKER_00] I don't think anybody in this room could probably confidently predict whether we are getting the asymptote because it seems as though we've been in a three to four month cycle of someone having a leap for the last almost three and a half years. [SPEAKER_00] So I tweet a lot about thought code, which means that I have people who go to my friends that OpenAI Codex being like, yo, can you switch and use it? [SPEAKER_00] And I'm like, no, I have a bunch of contacts and I know how to do these things. [SPEAKER_00] It's hard for me to switch. [SPEAKER_00] So I think the harness thing is real. [SPEAKER_00] And it has also surprised me that there's hadn't been as much shadow about trying out 5.2 or some of the more recent models. [SPEAKER_00] And maybe you're right. [SPEAKER_00] Maybe it's incrementally better. [SPEAKER_00] Maybe we're not hitting the asymptote. I think Opus 4.6 came out pretty close to 5.2. So maybe one of that is contributing to it. But I guess the question in here is, do you think that we are running out of model iterations that are significant? I don't personally think so what I created today. [SPEAKER_01] Yeah, no, I think there's tons of room. [SPEAKER_01] I think the question is more, what is the closeness between the models and what is the utility you get through things at ancillary? [SPEAKER_01] And the ancillary things include the harness, it includes brand, actually. [SPEAKER_01] Yeah. [SPEAKER_01] Right. [SPEAKER_01] What you could tell your friends you're using. [SPEAKER_01] Yeah. [SPEAKER_01] No, seriously. [SPEAKER_01] I don't know. [SPEAKER_01] It matters. [SPEAKER_01] So I think it's a couple of different things. [SPEAKER_01] You just get used to having things configured your way, just for types of prompts or everything is kind of set up for you. [SPEAKER_01] And I think it's a recent phenomenon. [SPEAKER_01] I think it's very new. [SPEAKER_01] And that's why I think it's not discussed very much as a thing. [SPEAKER_00] As you kind of, from your vantage point, you know, obviously we've had, we're still continuing to innovate along. [SPEAKER_00] Okay, let's do, you know, longer chains of thought. I don't know. It matters. So I think it's a couple of different things. You just get used to having things configured your way, just for types of prompts or everything is set up for you. And I think it's a recent phenomenon. It's very new. And that's why I think it's not discussed very much as a thing. [SPEAKER_00] From your vantage point, obviously we've had, we're still continuing to innovate along. [SPEAKER_00] Okay, let's do longer chains of thought. [SPEAKER_00] Let's do better RL. [SPEAKER_00] Let's do longer context. [SPEAKER_00] Let's do multimodal. [SPEAKER_00] Let's do omnimodal. [SPEAKER_00] Let's do world models and so on. [SPEAKER_00] Right. So when you think about the vectors along which these models improve over the 8 to 24 month horizon, how do you think about it? At this stage, you might not always work for a foundation model company? [SPEAKER_01] Yeah, we all will be in the long run. [SPEAKER_01] Yeah, we all will. [SPEAKER_01] I'll get some sort of thing. [SPEAKER_01] Yeah, when they work on a railroad. [SPEAKER_01] Yeah. [SPEAKER_01] It's very durable. [SPEAKER_01] So I think there's a few different types of models, right? [SPEAKER_01] There's the core language foundation model, like the LLMs, et cetera. [SPEAKER_01] That's mainly what we talk about. [SPEAKER_01] And I think two, three years ago, everybody thought this is going to go agentic over time. And for agentic things, you need some form of persistent memory. And you need to have these actions. You need to be able to integrate with these different things. I think it was clear what to do. And it's just a matter of time to do it. Right? We knew reasoning was coming because we'd seen reasoning in other contexts. And so I think a lot of it has been reasonably predictable. I remember talking to Eric Steinberger, who runs Magic. Three years ago, he was talking about super long context windows being incredibly important for code because then you could put the whole code base in and they could work across it. And so again, I think the smartest people had all these ideas that awaited attention. And similarly, I think there's a very clear roadmap in terms of what's to come. For other areas, I think there's lots of really cool stuff to do. What are the best models for physics simulation? What are the best models for materials? You go through a whole wide vertical, area by area, and there's super interesting stuff to do. Now, that may be severable for economic value. Right? And that's my big question about a subset of the bio models. I think there's broader things people are doing. But if you look at the cost of developing a drug, for example, if it's one and a half billion and it takes 15 years, tens of millions of that is preclinical work. That's the molecule. And the other 1.45 billion is clinical trials. Right? Now, you may be able to impact the success of a clinical trial based on the molecule. You look at AdMed and other aspects of drug development, et cetera. But fundamentally, those companies, in my opinion, many of them will end up being drug development companies. And the question is, do they have real systemic advantages? And so those feel like tougher categories than maybe physics or language or image gen or other areas. And so I think part of it is what are you doing? How advanced is it? How interesting is it? How much utility has it shown? The overlay is what is the economic model that you're dealing with? [SPEAKER_00] Maybe two questions rolled up into one here. [SPEAKER_00] I think that over the last six weeks, there's probably more encroachment from the foundation models into a number of different domains, like legal and finance. [SPEAKER_00] And obviously all the coding-related things. [SPEAKER_00] So I guess the two questions here are: How do you as a startup think about what is in the effective blast radius of what these models can do? [SPEAKER_00] And then secondly, what are the durable modes that an application or a product can have? [SPEAKER_00] And I think one of the things you've said in the past, which actually resonated with me a lot, was that data is overrated as a mode. [SPEAKER_00] Everybody talks about having proprietary data and being able to do better with context window stopping. [SPEAKER_00] And that feels a little off to me. So if you look at the history of technology, every time you have a new platform, it forward integrates into the most valuable application on the platform. For example, Microsoft OS forward integrated what became Word and Excel and PowerPoint and Access. And they basically killed and bought companies that were doing that. And then they forward integrated in the browser. And that was a famous fight with Netscape in the 90s. Google in the 2000s forward integrated into the biggest vertical search categories, right? They had local and they did travel and they did finance. And this always happens. And so it's not odd that the foundation model companies or labs will forward integrate into the biggest applications, starting with code. Code in particular also has this interesting attribute: if you have a very good coding model, it helps you generate the next model faster, right? [SPEAKER_01] And that's because it helps you write code for it. [SPEAKER_01] It helps you with data labeling. [SPEAKER_01] It helps you with a bunch of stuff. [SPEAKER_01] And so one of the hypotheses in terms of how we get true liftoff into a singularity is you have models, build models, build models, and then you just take off, right? [SPEAKER_01] And so it's not a surprise that the labs are focused on that. [SPEAKER_01] The biggest application starting with code. [SPEAKER_01] Code in particular, though, also has this interesting attribute, which if you have a very good coding model, it helps you generate the next model faster, right? [SPEAKER_01] And that's because it helps you write code for it. [SPEAKER_01] It helps you with data labeling. [SPEAKER_01] It helps you with a bunch of stuff. [SPEAKER_01] And so one of the hypotheses in terms of how we get true liftoff into a singularity is you have models, build models, build models, and then you just tick off, right? [SPEAKER_01] And so it's not a surprise that the labs are focused on that. [SPEAKER_01] It seems like the single clearest specter of speed of change. [SPEAKER_01] And eventually we have evolving models and evolutionary trees and you spawn 50 versions of the same thing and there's some utility function run against all sorts of crazy shifter, seems likely to happen at some point. So what is the one? It's great. We will be all, we'll be all at the edge of the same. [SPEAKER_00] Sorry. [SPEAKER_00] If we get, I got a little bit distracted. [SPEAKER_00] I was going to be like, you're a set day. Yeah. You're happy. They should not see questions like this is a letter question. And I'm by the way they're going to afford an interim. Yeah, exactly. So I think if you look at durability, there's three ways you can be durable. And by the way, if you look at the whole wrapper around a model thing, I don't really buy it because if you look at a fast company, is this a wrapper around a SQL database and all those things just buy it, right? I think usually what you need is you need to build a multi-product company. So there's three theories. One is your system of record. You're basically where core data about a thing is stored in all its attributes. And so that's a work day for all the people working at your company or whatever. And so then you keep all the dozen applications around it and it's very durable. That's one hypothesis in terms of what creates durability. I think the single biggest thing is just can you build a dozen different products that you're cross selling roughly to the same customer or user, they're deeply integrated and therefore you have multiple workflows. And I think then you're very defensible. So for example, if Harbi has two dozen different workflows for different legal applications and use cases, that's defensible. If they're just bust a child with the legal document, they're not very defensible. [SPEAKER_00] Super observation. [SPEAKER_00] I actually haven't. [SPEAKER_00] Maybe I've said this publicly before, but that's interesting because it's a very look at what Parker always talks about with the tech startup. It's Rippling. [SPEAKER_01] It's HubSpot. [SPEAKER_01] It's actually Microsoft. It's a lot more doable today because you can just crack that stuff out. [SPEAKER_01] Yeah. [SPEAKER_01] It's what's the surface area of your product. [SPEAKER_01] And I think it's very under discussed and it's often discussed as a revenue driver, which it is. And you have different wedges in that you can cross sell against. You can cross sell to the same account. It's hugely defensible, right? Because usually what happens is a founder will start a company that builds one product and they'll win because the product is 10x better and that'll get in distribution. And then they keep going and they forget the fact that now that they're the incumbent and they have all these customers that can build a product that's 80% as good as the best thing and still win because they're just cross selling. They worry about the security review. They worry about the purchasing. They worry about they've gone through procurement. They've gone through everything legal. And so it's really easy to start cross selling, in which case it becomes very defensible, right? It's really hard to rip out a dozen things instead of just one thing. [SPEAKER_00] I was going to say this, why we don't usually like to sub themes. [SPEAKER_00] Well, teams is such an interesting example of that, right? [SPEAKER_00] I see. [SPEAKER_00] Yeah, it's actually an interesting example of what you said. [SPEAKER_00] Yeah. Yeah. It's fascinating if you look at it. So the early days it was started. The primary thing you're competing with is other startups. And then incumbents often used to have five to seven years to react literally because they could just cross sell to everybody. And so that's Slack and Zoom versus Teams, right? They were both like this and then they flattened out and then Slack had to sell. [SPEAKER_01] And the reason is because Microsoft bundled it in Teams, gave it to everybody and they cut off their growth and they won. [SPEAKER_01] And so in the olden days, last decade, you had five to seven years of fighting other startups and then an incumbent would come in. [SPEAKER_01] And if you could escape that incumbent, you won. [SPEAKER_01] Now, there's less time for the incumbent triad because you can iterate so quickly on code and the markets are open and everything else. But the incumbent can also build share way faster, even if they're getting in their own way. And so there's this interesting dynamic that that timeline should shorten on both sides. And so we'll see what happens there. That's interesting though. [SPEAKER_00] I mean, you could argue that whatever baseline speed you are starting off, a startup should be able to accelerate it at a faster exponent. [SPEAKER_00] So you should have technically an advantage in a world where you can crack out on code. [SPEAKER_00] And I would argue it's probably still harder for Microsoft or Google to generate a bunch of, but you never know. [SPEAKER_00] I get it. [SPEAKER_00] They're fast moving. [SPEAKER_00] They're also good companies. Yeah. They're good companies and they have the distribution and traditionally they, but then, you know, probably the biggest tactic for them is in that five to seven years and maybe they have two or three years. Right. But there is time. [SPEAKER_00] There is time. [SPEAKER_00] Yeah. [SPEAKER_00] What is the most common mistake that you see AI startups making nowadays? [SPEAKER_00] Or maybe even call it startups. [SPEAKER_00] Is there a common failure mode that is emergent based on the environment that you're in right now? [SPEAKER_00] They're also good companies. Yeah. They're good companies and they have the distribution and traditionally they, but then, probably the biggest tactic for them is in that five to seven years and maybe they have two or three years. [SPEAKER_01] Right. [SPEAKER_01] But there is time. There is time. Yeah. What is the most common mistake that you see AI startups making nowadays? Or maybe even call it startups. Is there a common failure mode that is emergent based on the environment that you're in right now? You know, I think there are three types of failure mode. I think one type of failure mode is your thing isn't working and you stick with it too long. And it's back to you should eventually iterate into something really working. And if it's taking two years and nothing's working, you should rethink. In most cases, not all, again, there are counter examples. The second thing is there's really bad advice given to things that are working from people who've never had anything work. So what used to happen is you'd get advice from founders who'd seen product market fit to founders without it. And they say, go hire a sales team and scale really fast and burn a lot of money into it. And that was terrible advice. But it goes the other way. If something's really working and people tell you to stay as lean as possible, hire executives, don't scale that. It's awful advice. And that's where a lot of companies break for a while and other people, competitors who are in a scale, come in. And it's a very common pattern for things that are working. The founders are going to actually build out their company as a product and they don't really go for it in an aggressive enough manner. So I think that's a big failure mode. And another failure mode is spending money on stuff that just doesn't matter. You know, you start training a crazy model on something instead of just testing something with an existing tool and just seeing if people want it. [SPEAKER_00] You know, maybe for that last one, I'm curious. I mean, you're deploying large amounts of capital often to a pretty early stage companies, I guess, you know, for a lot of companies that are raising a hundred, 200 pretty early on in their life cycle. Have you found that they actually know how to use it? Well, it's hard, you know, like if you're training the models, maybe it's all ultimately that. I actually have not funded many things like that. [SPEAKER_00] Okay. [SPEAKER_00] I mean, so I suspect that we have the same priors because I think it's quite hard to spend a hundred. You know, if you raise a hundred million dollars as a seed round, you can't really pay yourself a hundred million dollars. You know, the founder, you have. [SPEAKER_00] Yeah. [SPEAKER_00] I mean, you could. [SPEAKER_00] If anybody wants to hear it all this way. You have seed in the back, seed creation in the back. [SPEAKER_00] Right. [SPEAKER_00] And some salary. [SPEAKER_00] Yeah. [SPEAKER_00] But it's quite hard because your range of what you can pay yourself is pretty bounded, I think, in terms of what is accepted in Silicon Valley. There's only a certain rate in which you can hire and create people. It's not easy to hire well and quite. So often if you raise a hundred at five hundred, I tell people that you can do it, but it requires everything to go right. [SPEAKER_00] Right. [SPEAKER_00] For 12 to 18 months. [SPEAKER_00] And if you had any hiccups along the way, it's really hard to zig and zag the company, which is actually opposite of what you'd expect. Because most people are like, why raise a bunch of money now, I can afford to get it wrong. But I actually think you can. It's much harder to get things wrong. You might have the money, but you don't actually have room to move. [SPEAKER_01] Yeah. I think you're setting different expectations for yourself. I do think there are circumstances where you raise a hundred million dollars and you build, you know, like what Pi is doing, you know, on the robotics side, you're building these foundation models for robotics. That needed a lot of money because they need a lot of computing, they need a lot of training. You know, great. There's a rationale behind it. I just think it's tough depending on what you're doing. And so it really has to be tailored against what are you actually trying to accomplish? And one proportion of it is GPU versus headcount. And often you see people raise big rounds and 80% of it's supposed to go to GPU or something and the other 20% is headcount. So you see these differential ratios versus I'm raising a hundred just to have people, which I think is a tougher thing for a very early company. [SPEAKER_00] I mean, when you came here last, we're going to switch topics a little bit. It's a lot of AI. So we're going to switch topics a little bit. We'll come back to the AI, all right? I think when you were here last, crypto and Web3 was still pre the crash, pre the winter. And I think it's actually pretty interesting. You know, we're starting to see some interesting companies come about again. So I guess trying to figure out where are you in terms of crypto rails, distributed financial transactions. Are you doing any investments there? And there's obviously a lot of talk around how agents might end up being one of the key catalysts to essentially allowing the transaction rails that only exist in Web3. Are you a bull there? Do you agree? You know, I think ultimately there are a lot of APIs for payments. And so anything that's programmatic and an agent can interact with, sure. And that could be a stable coin or it could be a traditional payment system, but it's more just [SPEAKER_00] Rails, distributed financial transactions in terms of are you doing any investments there? And there's obviously a lot of talk around how agents might end up being one of the key catalysts to essentially allowing the transaction rails that only exist in Web3, are you a bull there? Do you agree? I think ultimately there's a lot of APIs for payments. So anything that's programmatic and an agent can interact with, sure. And that could be a stablecoin or it could be a traditional payment system, but it's more just what sort of access do you provide to things. And Stripe I think has done both effectively. They've been thinking about agent APIs for payments and they've also been doing a lot of really interesting work in stablecoins. And I kind of use Stripe as the hidden crypto company that nobody's talking about that's doing pretty crazy stuff between their acquisitions and help. Yeah. They're doing some other stuff. I'm a long-term crypto bull, but it has cycles and we're obviously in a down cycle right now. And my anticipation, this is not financial advice. My anticipation is that it's going to get worse before it gets better. And if you look at historical cycles, Bitcoin should probably drop somewhere between 35 and the forties. And maybe it doesn't do better than that this cycle and some of fifties or sixties or whatever, but it's already in the sixties or it has been. But I do just look at it. You have a very standard crypto cycle and you have halvenings and it runs again. It's the same stuff. And maybe it not, people always say it won't apply this cycle and then it applies again. And at some point it won't apply anymore. But that kind of drives a lot of the ancillary behavior in crypto because so much wealth in the crypto world is tied up in that. But you see people trading sequentially across the different crypto assets with Bitcoin being one of the more stable pools. And then you have a bunch of other tokens on top of that, and then you have other companies. And so I think we have these boom bust cycles that also draw founders in and out. One thing that I think is really exciting and interesting is there's a certain type of technical founder that depending on when they graduated from school, they either went into crypto or they went into AI and it's like a six month difference. Literally. It really is. [SPEAKER_01] They've actually done analysis of this in the context of pre-SVF post-SVF, right? Yeah. [SPEAKER_01] Seriously, it really is, or pre-ChatGPT, post-ChatGPT, right? Or whatever it is. And it really swings people's outcomes and careers. And there's actually really interesting data in general that I remember years ago where if you graduate into a recession versus a boom cycle, you make dramatically less over your entire career. And the reason is you have fewer opportunities. You don't get jobs. You never manage people early. All the things that come with the boom cycle you miss and all the hardness of the bear cycle you had. You're also just less optimistic. I read that same study, which is that you expect the world to be freer. And it's in my life. [SPEAKER_01] Yeah. But you know, I think you just expect that your expectation around the world is that growth is low. And that there are larger systemic forces which are actually preventing you from flourishing, as opposed to YOLO, it's just you go up and up, right? Yeah. And so I think that same thing is, I'm not saying financially, I just mean in general, there was that crypto AI divide and you had founders who started crypto companies and they stuck with them in many cases. And then you had other people who started AI companies and it was a few months of difference. It's the exact same type of person. And so I think that's fascinating. And there's all these weird cohort effects in the medical, all this stuff we don't talk about that's happening in the background. I think another interesting question, which we can skip if you don't want to talk about it, but I think you also see Silicon Valley kind of moving these five to seven year cohorts of people. And so there's this interesting question of how do people maintain your relevance across multiple cycles or what does that mean? Or why do two people who seem equally good have very different outcomes? Is it them? Is it probabilistic? Is it, if you run a Monte Carlo simulation of a person's life, you rerun their life a billion times, what's the expected outcome of that person? And how do these circumstances impact that? So there's all these really interesting things that you can think about in terms of talent. I'll give you another talent question. Jensen Huang, amazing CEO, brilliant, amazing strategist, so good at running the company in a contrarian way, biggest market cap in the world, high EQ, very technical, amazing. For years, he was running a 6 billion dollar company, right? Pinned for 30 years, for 30 years he ran a company and he was this hidden gem of a brilliant CEO who eventually steered things into AI and game over. How many other Jensen Huangs are there out there running public companies where it's just in a market that right now isn't boomy, but they're exceptionally talented? Is it zero? Is this uniquely good? And that may be true. Is it a hundred? How do you identify these people and how do you unlock them? And so I think there's these really interesting talent questions. Like what is the aggregate available talent on the planet and how do you harness it in different ways that I think is never discussed really. I'm curious to get your take. You kind of mentioned one of the central questions is like, in venture, are there not enough good founders or are there not enough good funders? But where do you kind of fall at? Most likely there's enough good founders who are just not pointing at the right things. Is it a hundred? How do you identify these people and how do you unlock them? And so I think there's these really interesting talent questions. What is the aggregate available talent on the planet and how you harness it in different ways that I think is never discussed really. I'm curious to get your take. You mentioned one of the central questions is in venture: are there not enough good founders or are there not enough good funders? But where do you kind of fall at? Jay Shahid Most likely there's enough good founders who are just not pointing at the right things. Most likely there's a faulty search function. And if you think about it, entrepreneurship is a distributed search around the economic landscape of the world. That's really what's happening, right? You're running a giant search function across the economy. And it's not an efficient process and there's unequal information and unequal access and all the rest of it. And so it doesn't work well, but I think there's probably enough good people that if you pointed them in the right thing. Now the founders side of it is, if you ask people how many great product people exist on the planet, I asked a very well-known public market CEO who runs one of the most interesting product companies that question. I said, how many great product people do you think exist in Silicon Valley or in tech? And he said at most a few hundred. And so then each company has at most a couple of those. And those are the people with high agency who are grinding on big things for those very large companies. Right. And so there's, he may be wrong. Maybe there's tens of thousands in the world. Maybe there's like 50, I don't know. But it is striking. If you look at that overlay, if everything's a bell curve, then you need outliers on multiple aspects of multiple bell curves to do certain things exceptionally. And so then you're compounding small probabilities. So one argument is for any given thing, there's not that many people. And one of the parts of that bell curve is almost always agency. I was going to say, the longer that I am in Silicon Valley, the most determining bit that I find is actually the site agency. It's given a particular situation, is your first reaction like, I'll just do it. I'll figure it out. Given any problem, your reaction is not that this seems super hard or seems impossible. It's like, yeah, and I have to do it. We'll figure it out. There's some way. Nothing can in some way stop you. That doesn't mean you won't get stopped, but your first reaction is always like, oh yeah, I got to figure it out. When I'm going my way out of it. Right. I think the open question for me and my variant of it was, is that something that can be truly inculcated or taught or developed versus is there something a little bit more intrinsic about it? [SPEAKER_01] Yeah, I think it's probably both. And I think one of the questions I ask is in this incredible age of technological capabilities, what do you want your kids to have? And the first answer is just agency. So you want all your kids to be, you're going to say Bitcoin. Bitcoin. That's what I think about. Yeah, Bitcoin, I mean, it's steady. So my, Ruchi's brother was visiting us this last week and turns out that I gave all of his children a bunch of Bitcoin six years ago. And then I asked him like, do you know where the keys are? And he's like, lost it. [SPEAKER_01] I didn't see. Sounds like the right answer. Coming back to it, you can buy Bitcoin ETFs now, by the way. So it makes a difference. But they also go down when Bitcoin goes down. Yeah, they have many ETFs. Yeah, it's interesting. So Elad wrote a great book called The Hardcover Handbook. Pruduchi actually has a chapter in that book. I think it's the wolf, named after Pulp Fiction's Harvey Keitel character, the wolf. But that book was written, what, ten years ago? Seven, eight years. That's been a while. What's the biggest thing you would revise or update in it? And make it shorter. Now, this headache, that can be an AI's read it well. [SPEAKER_01] Yeah, so yeah, it's been giving it thought on it. Yeah, I don't think a lot of it is still reasonably relevant because a lot of it was about people-related stuff. It's true. The basics of fundraising or how do you fire somebody for the first time or how do you hire executives or how do you do M&A. So I think there's some aspects that were kind of shifting in time, like some of the major funding of startups and stuff like that. And there's a section on that. But I think in general, it reasonably still works. I'm actually writing another book for starting from scratch, which is more about the zero to one phase of startups. And so I've been working on that and that's been pretty fun as a project. So what is high growth even meaning for in 2026? Like when you take a look at a company, what is your marker for that company will grow and that fits into the large category? Like you will definitely pay attention to the numbers. Yeah, I don't have anything scripted because I think it's a little bit market-segment dependent. And so if you're looking at a defense tech company or something, it's going to be pretty different than if you're looking at an AI vertical company. But in general, everything is growing faster than one would expect. And on the AI side, in particular, obviously we're seeing these massive ramps. I think, you know, what an insane ramp. And you see multiple companies growing really fast. Harvey's growing really fast. Decagon and others are growing really fast. So you just see these things lift off. And I think it's back to these markets are open. The capabilities are massive. The transition is large. The ability to enter a product is high. Everybody wants to try things right now. So again, it's a very magical moment. [SPEAKER_00] General, everything is growing faster than one would expect. And on the AI side, in particular, obviously we're seeing these massive ramps. I think Chris now we're murdered at today. I was just, what an insane ramp and you see multiple companies growing really fast. Harvey's growing really fast. Decagon and others are growing really fast. So you just see these things lift off. And I think it's back to these markets are open. The capabilities are massive. The transition is large. The ability to enter a product is high. Everybody wants to try things right now. So again, it's a very magical moment and a very manic moment. And I was looking back in history at the nineties and in 99, 450 companies went public in the first few months of 2000 and another 450 companies went public. So say that you had 1500 to 2000 companies go public over a five year span. How many of those are still relevant? I don't know the number. It's a dozen, two dozen. It's very, very few. Most of the companies went to zero. Those are the most successful companies. They went public, right? It's not the average company. It's the most successful companies, 90 something percent are just gone. And so then you think about that in the context of AI and you're like, okay, most of these things are not going to exist. A handful of things are going to be Amazon and Google and et cetera. Right. And so then as a founder, how should you think about that? And for every company, there's a handful of companies that'll keep going forever, right? That's probably OpenAI and Anthropic and other stands all the same going forever. There's a lot of companies that are looking really good right now. They should probably sell at some point. And for every company, there's a 12 month period, which is the value maximizing period. That's going to be the most valuable and important it'll ever be. And then in many cases, these things go to zero or close to it. So one thing that some people do to have good hygiene around this, and by the way, I think there's some companies that should absolutely never sell, right? The companies that really work should never, ever, ever sell. But how do you know if you're one of those? Right. And so good hygiene, I think for companies, especially if you have a board is once a year to pre-schedule a board meeting that's talking about exits. Should we sell? And if so, when, is this a value maximizing moment? And because you pre-schedule it and it's annual, it takes the emotion out of it. It doesn't look like you're trying to sell or want to sell. It doesn't look like you're against it. You're just going to have a rational, logic-based conversation on the market, on competition, on your position, on your growth rate. If your growth is like this, that you see the second derivative changing, that may be a good time to sell once you can fix it. Right. And so I think it's worth just having that conversation ongoing because again, in most of these cycles, 97% of things end up not working, even if they're looking like they work. Even the CEO's job in today's world is a little different than quite a while back. So like should every CEO, do they have to be building as an example? Is it possible to be at the forefront of this technological? [SPEAKER_00] Was Steve Jobs building? Was Jeff Bezos building? Did they have access to code? I don't know. Who knows, that's why Amazon is successful. I mean, I think Jeff Bezos is back to building now, right? How do you define building? [SPEAKER_00] Yeah. Right. He wasn't technical at all. And so he hired some guy to build stuff. If you look at Uber, right, Travis has talked about this publicly. Travis and Garrett set it up as a side project. They found the CEO Ryan on Twitter. They literally tweeted, hey, we have this side project, who wants to run it? And somebody put up their hand on Twitter. That was Ryan Graves, right? And then I think they initially outsourced the app to third-party developers and they brought it in-house, right? Were they building? [SPEAKER_00] Yeah. Good point. With Garrett, it always seems to build the first version. Maybe he did. [SPEAKER_00] Yeah, maybe he did. But now you're quite the builder. [SPEAKER_00] I don't know. What is that? Again, I wasn't in the middle of it. I'm just saying it's interesting to look at historical precedents. And I think we have all these myths. And maybe 90% true. [SPEAKER_00] I was just actually walking by the Cove office, which was 148 Towns the other day. And I remember because it was impossible to be in a cab in San Francisco. So my first ever Uber was like, they Uber Black and from 148 Towns to our house in Noe Valley. And it was amazing because I couldn't find a cab in San Francisco. So I was just thinking about that. One aside, by the way, I do think builder CEOs tend to do better on average. So I'm not at all negating that. I'm just saying there are counter examples. And it's because they're in the weeds on the product, attending to the details. They're doing stuff. They don't need another person to enable that. So they can get things done. There are all sorts of reasons why it's a huge plus. So not at all. In general, I tend to work with technical founders. I'm just saying there are examples where that isn't true. [SPEAKER_00] And maybe the larger follow-up question is, as a CEO, should they be a little bit more hands-on than they were five years ago? Because the tools are there, because the advisors are there, the bifurcation between manager mode and maker mode was pretty stark. I think people should have always been hands-on and it's back to where I think micromanagement is underrated. I think being able to be hands-on is underrated. Yeah, I agree. I think that transition's come back. I think it went from delegate, delegate, delegate to actually, you should delegate a ton of stuff. But there's a lot of things that you should be in the weeds on and there's a lot of things that you personally should micromanage and those are the most successful companies. And I think it's that balance. It's not one or the other. And I think people have over-indexed in either direction over time. Yeah, you should actually micromanage itself. That's the highest server and the most important things. And that you're actually good at. Yeah, no, I agree. I think that transition has come back. I think it went from delegate, delegate, delegate to actually, you should delegate a ton of stuff. But there's a lot of things that you should be in the weeds on and there's a lot of things that you personally should micromanage and those are the most successful companies. And I think it's that balance. It's not one or the other. And I think people have over-indexed in either direction over time. Yeah, you should actually micromanage itself. That's the highest server and the most important things. And that you're actually good at. Yeah. Yeah. [SPEAKER_00] Thank you. You lots of comment? [SPEAKER_00] I have a fun. [SPEAKER_00] Thank you so much. [SPEAKER_01] Thank you so much. [SPEAKER_01] Thank you so much. [SPEAKER_01] Thank you so much. 200 pretty early on in their, I would say life cycle. Have you found that they actually know how to use it? Well, it's kind of hard, you know, like if you're like, I mean, maybe you're training the models, maybe it's all kind of like ultimately. That's that I actually have not funded many things like that. Okay. I mean, so I suspect that we have the same priors because I think. It's quite hard to spend a hundred, like, you know, if you raise a hundred million dollars as a seed round, you can't really pay yourself a hundred million dollars. You know, you know, the founder, you have. Yeah. I mean, you could. If anybody wants to hear it in all this way, your face. You have seed in the back, seed creation in the back. Right. And some salary. Yeah. But it's quite hard because your range of what you can pay yourself is pretty bounded, I think, in terms of what is accepted in Silicon Valley. There's only a certain rate in which you can hire and create people. It's not easy to hire well and quite. So often if you raise, you know, a hundred at five hundred, I tell people that you can do it, but it requires everything to go right. Right. For 12 to 18 months. And if you had any hiccups along the way, it's really hard to zig and zag the company, which is actually opposite of what you'd expect. Because most people are like, why raise a bunch of money now I can afford to get it wrong. But I actually think you can. It's much harder to get things wrong. You might have the money, but you don't actually have room to face. It's that I'm adding on it. Yeah. Yeah. I think you're setting different expectations for yourself. I do think there are circumstances where you raise a hundred million dollars and you bill, you know, like what Pi is doing, you know, on the robotics side, you're building these foundation models for robotics. Like that needed a lot of money because they need a lot of computing, they need a lot of training. They need a lot of training. You know, great. There's a rationale behind it. I just think it's tough depending on what you're doing. And so it really has to be tailored against what are you actually trying to accomplish? And one proportion of it is GPU versus headcount. And often you see people raises the grounds and, you know, 80% of it's supposed to go to GPU or something and the other 20% is headcount. So you see these differential ratios versus I'm raising a hundred just to have people, which I think is a tougher thing for a very early company. I mean, when you came here last, we're going to switch topics a little bit. It's a lot of AI. So we're going to switch topics a little bit. We'll come back to the AI at all, right? I think when you were here last, crypto and Web3 was still pre the sadness, pre the winter. And I think it's actually kind of, you know, we're starting to see some interesting companies come about again. So I guess trying to where are you in terms of like, you know, I would say, whether it be crypto rails, distributed financial transactions in terms of are you doing any investments there? And there's obviously a lot of talk around how agents might end up being one of the key catalysts to essentially allowing, you know, kind of like the transaction rails that only are utilizing the transaction rails that only exist and if it triple and Web3, are you a bull there? Like do you agree? You know, I think ultimately there's a lot of APIs for payments. And so, you know, anything that's programmatic and agent can interact with. Sure. And that could be a stable coin or it could be a traditional payment system, but it's more just like what sort of access to you provided things. And Stripe I think has done both right effectively. They've been thinking about agent APIs for payments and they've also been doing a lot of really interesting work in stable coins. And I kind of use Stripe as like the hidden crypto company that nobody's talking about. That's doing pretty crazy stuff between their acquisitions and help. Yeah. You know, so the acquisition and. Yeah. They're doing some other stuff. You know, I'm, I'm a, a long-term crypto bull, but it has cycles and we're obviously in a down cycle right now. And my anticipation, this is not your best advice. My anticipation is that it's going to get worse before it gets better. And if you look at historical cycles, you know, Bitcoin should probably drop just somewhere between say 35 and the forties. And, you know, maybe it doesn't, it does better than that this cycle and some of fifties or sixties or whatever, but it's already in the sixties or it has been. Um, but I do just kind of look at it. You, you have a very standard crypto cycle and you have them having and it runs again, you know, it's the same stuff. And maybe it not, people always say it won't apply this cycle and then it applies again. And at some point it won't apply anymore. But that kind of drives a lot of the ancillary behavior in crypto because so much wealth in the crypto world will tie it up in that. But you see people trading sequentially across, um, the different crypto assets with Bitcoin is sort of one of the more stable pools. And then you have a bunch of other tokens on top of that, and then you have other companies. And so I think we have these boot bus cycles that also draw founders in and out. One thing that I think is really exciting, that interesting is there's a certain type of technical founder that depending on when they graduated from school, they either went into crypto or they went into AI and it's like a six month difference. Literally. It really is. Uh, they've actually done analysis of this in the context of, um, it's like pre SPF post SVF, you know? Yeah. You know, seriously, it really, it really is, or pre chat to BT, post chat to BT, right? Or whatever it is. And, uh, it really swings people's outcomes and careers. And there's actually really interesting data in general that I remember years ago, where if you graduate into a recession, versus a boom cycle, you make dramatically less over your entire career. And the reason is you have fewer opportunities. You have your jobs. You never manage people early. Like all the things that come with the boot cycle you miss and all the things and the hardness of the bear cycle you had. You're just also less optimistic. Like I read that same study, which is that you expect the world to be free. And it's in my life. Yeah. I think you're doing a cancer. You're mine. But you know, I think you just expect that like, hey, your, your expectation around the world is that growth is low. And that there are larger systemic forces, which are actually preventing you from flourishing. Um, as opposed to kind of like YOLO, it's just like, you know, you go up and up. Right. Yeah. And so I think that that same thing is, I'm not saying financially, I just mean in general, there was, there was that crypto AI divide and you had founders who started crypto companies and they stuck with them in many cases. And then you had other people who started AI companies and it was a few months of difference. It's the exact same type of person. And so I think that's kind of fascinating. And you know, there's all these weird cohort effects in the medical, like there's all this stuff we don't talk about. That's kind of happening in the background. I think another interesting question, which we can skip if you don't want to talk about it, but I think, um, you also see Silicon Valley kind of moving these five to seven year cohorts of people. And so there's this interesting question of like, how do people maintain your elements across multiple cycles or what does that mean? Or why do two people who seem equally good have very different outcomes? Is it them? Is it probabilistic? Is it, you know, if you run a Monte Carlo simulation, a person's life, you rerun their life a billion times as the expected outcome of that person. And how do these circumstances impact that? So there's all these really interesting things that you can think about in terms of talent. I'll give you another talent question. Jensen Wei, amazing CEO, brilliant, amazing strategist, like so good at like running the company in contrarian way, biggest market cap in the world, like high EQ, very technical, like amazing. For years, he was running a $6 billion company, right? Payne for 30 years, for 30 years he ran a company and he was this hidden gem of a billion CEO who eventually steered things into AI and, you know, game over. How many other Jensen Wangs are there out there running public companies where it's just in a market that right now isn't boomy, but they're exceptionally talented? Is it zero? Is this uniquely good? And that may be true. Is it a hundred? Is it like, how do you identify these people and how do you unlock them? And so I think there's these really interesting sort of talent questions. Like what is the aggregate available talent on the planet and how you harness it in different ways that I think is kind of never discussed really. I'm curious to get your, you kind of mentioned one of the central questions is like in value is are there not enough good opportunities or are there not enough, you know, good funders? Um, but where do you kind of like, you know, almost fall at? Jay Shahid Most likely there's enough good founders who are just not pointing at the right things. Jay Shahid Most likely there's a faulty search function. And if you think about it, entrepreneurship is a distributed search around the economic landscape of the world. Jay Shahid Most likely that's really what's happening, right? You're really running a giant search function across the, across the economy. And, um, it's not an efficient process and there's, you know, uh, unequal information and equal access and all the rest of it. And so, um, it doesn't work well, but I think there's probably enough good people that if you pointed them in the right thing. Now the folks side of it is, um, if you ask people how many great product people exist on the planet, I asked a very well known public market CEO who runs one of the most interesting product companies. That question I said, how many great product people do you think exist in Silicon Valley or in tech? And he said at most a few hundred. And so then each company has at most a couple of those. And those are the people with high agency who are grinding big things for those very large companies. Right. And so there's, there's now he may be wrong. Maybe there's tens of thousands in the wrong server. That's where maybe there's like 50, I don't know. But, um, it is striking. If you look at that overlay of, you know, if everything's a bell curve, then you need outliers on multiple aspects of multiple bell curves to do certain things exceptionally. And so then you're compounding small probabilities. So one argument is for any given thing, there's not that many people. And one of the parts of that bell curve is almost always agency. I was going to say, I actually think that the most, um, longer that I am in Silicon Valley, the most determining bit that I find is actually the site agency. It's like kind of given a particular situation is your, is your first reaction is like, I'll just, I'll just do it. I'll just like do things. I'll figure it out. I get, isn't it like given any problem, your reaction is not that, you know, it's like, Oh, this seems super hard. It seems to possible. It's like, yeah, and I have to do it. We'll figure it out. There's some way there's nothing that can in some way stop you. That doesn't mean you won't get stopped, but your first reaction is always like, Oh yeah, I got to figure it out. When I'm a guy doing my way out of it. Right. I think the open question for me and my variant of it was, is that something that can be truly inculcated, slash taught, slash developed, uh, versus is there something a little bit more intrinsic about it? Yeah, it was about, I think it's probably both. And I think, you know, one of the questions I ask at the end, you tell from you, you just ask now, which is in this incredible age of technological capabilities, what do you want your kids to have? And the first answer is just like agency. So you want all my kids to be like, you're going to say Bitcoin. Bitcoin. That's what I think about. Yeah. Bitcoin, I mean, it's steady. So my, Ruchi's brother was visiting us, uh, this last week and turns out that I gave all of his children a bunch of Bitcoin six years ago. And then I asked him like, do you know where the keys are? And he's like, lost it. I didn't see sounds like the right answer. Um, okay. Coming back to it. Um, you can buy Bitcoin ETFs now, by the way. So it makes a, but they also go down when Bitcoin goes down. Yeah. They have many ETFs. Yeah. It's interesting. So Elad wrote a great book called The Hydrows Handbook. Uh, Pruduchi actually has a chapter in that book. I think it's the, uh, the wolf, uh, it's kind of named after Pulp Fiction's Harvey Tidell, you know, kind of like character, the wolf. Uh, but that was, I'm going to think book was written, what, uh, 10 years ago. Something like that, seven, eight years. That's been a while. What's the biggest thing you would revise or update it? And make it shorter. Now, this headache, um, that can be AI's read it well. Yeah. So yeah, it's been a GI on it. Um, yeah, I don't think, I think a lot of it is still reasonably relevant because a lot of it is, was about people related stuff. It's true. Are the basics of fundraising or how do you fire somebody for the first time or how do you hire executives or how do you do M and a or, you know, so I think there's some aspects that were kind of moving in time, like for some of the major funder of startups and stuff like that. And there's like a section on, uh, but I think in general, uh, it's, it reasonably still kind of works. I think, uh, I'm actually writing another book for starting to start best me to write this, um, which is more about, um, the zero to one phase of startups. Okay. And so I've been working on that and that's been pretty fun as like a project. Uh, so what is high growth even meaning for in 26? Like when you take a look at head of like a company, what is your marker for like that is guy busters will grow, like that kind of fits into the E-large. Like you will definitely pay attention to the numbers. Yeah. I don't, I don't have anything for scripted because I think it's a little bit market segment dependent. And so if you're looking at like a defense tech company or something, it's going to be pretty different than if you're looking at an AI, uh, vertical company. But in general, um, you know, everything is growing faster than one would expect. And, um, on the AI side, in particular, obviously we're seeing these massive ramps, you know, I think Chris now we're murdered at today. Uh, uh, I was just, you know, what an insane ramp and you see multiple companies growing really fast. You know, Harvey's growing really fast. Like, uh, Decagon and others are growing really fast. So you just see these things lift off. And I think it's back to like, these markets are open. The capabilities are massive. The transition is large. The ability to enter a product is high. Everybody wants to try things right now. So again, it's a very magical moment and a very manic moment. And, you know, I've, I was looking back in history at the nineties and in 99, 450 companies went public in the first few months of 2000 and other 450 companies went public. And so say that you had 1500 to 2000 companies go public over a five year span. How many of those are still relevant? I don't know the number. It's a dozen, two dozen. It's very, very few. Most of the companies went to zero. Those are the most successful companies. They went public, right? It's not the average company. It's the most successful companies, 90 something percent are just gone. And so then you think about that in the complex of AI and you're like, okay, most of these things are not going to exist. A handful of things are going to be Amazon and Google and et cetera. Right. And so then as a founder, how should you think about that? And for every company, there's a handful of companies that'll keep going forever, right? That's probably open AI and and drawback and PRD stands all the same going forever. There's a lot of companies that are looking really good right now. They should probably sell at some point. And for every company, there's like a 12 month period, which is sort of the value maximizing period. That's going to be the most valuable and important it'll ever be. And then in many cases, these things go to zero or close to it. So one thing that some people do to have good hygiene around this, and by the way, I think there's some companies that should absolutely never sell, right? The companies that really work should never, ever, ever sell. But how do you know if you're one of those? Right. And so good hygiene, I think for companies, especially if you have a board is once a year to pre-schedule a board meeting that's talking about exits. Should we sell? And if so, to be when, is this a value maximizing moment? And because you pre-schedule it and it's annual, it takes the emotion out of it. It doesn't look like you're trying to sell or want to sell. It doesn't look like you're against it. You're just going to have a rational, logic-based conversation on the market, on competition, on your position, on your growth rate. Like if your growth is like this, that you see the second derivative changing, that may be a good time to sell once you can fix it. Right. And so I think it's worth just having that conversation ongoing because again, in most of these cycles, 97% of things end up not working, even if they're looking at work. Even the CEO's job in today's world is a little different than the quite a primitive boss. Oh, so. Like should every CEO, like do they have to be building as an example? Like, is it possible to kind of be at the forefront of this technological? Was Steve Jobs building? Was Jeff Bezos building? Did they have access to cloud codes? I don't know. Who knows, that's why Amazon is successful. I mean, I think Jeff Bezos is back to building now, right here. Oh, how do you define building? Yeah. Right. He wasn't technical at all. And so he hired some guy to build stuff. If you look at Uber, right, Travis has talked about this publicly. Travis and Garrett set it up as a side project. They found the CEO Ryan on Twitter. They literally tweeted, hey, we have this side project, who wants to run it? And somebody put up their hand on Twitter. That was Ryan Graves, right? And then I think they initially outsourced the app to third-party developers and they brought it in-house, right? That was early Twitter. I mean, early on. Heber, I excuse me. We're CEO. Yeah. And they found them on Twitter. Right. And so, were they building? Yeah. Good point. With Garrett, it always seems to build the first worker. Maybe he did. Yeah, maybe he did. But now you're quite the builder. I don't know. What is that? Again, I wasn't in the middle of it. I'm just saying like, it's interesting to look at historical precedents. And I think we have all these myths. Yeah. And maybe 90% true. I was just actually walking by the Cove office, which was 148 towns in the other day. And I remember because it was impossible to be in a cab in San Francisco. So, my first ever Uber was like, they Uber black and from 148 towns to our house in Noye Valley. And it was amazing because I'm like, I don't find a cat in San Francisco. So, I was just thinking about that. One aside, by the way, I do think builder CEOs tend to do better. Yes. On average. So, I'm not at all negating that. And just saying there's counter examples. And it's because they're in the weeds on the product or detailing to the, they're doing stuff. They don't need another person to enable that. So, they can't get things and do that. You know, there's all sorts of reasons why it's a huge pilot. So, not at all. In general, I tend to work with technical founders. I'm just saying it. There are examples where that isn't true. And maybe the larger follow-up question is more as a CEO need. Should they be a little bit more hands-on than they were five years ago? Because the tools to, because the advisors, you know, essentially, you know, the bifurcation between manager mode and maker modes was pretty stark. I think people should have always been hands-on and it's back to, where I think micromanagement is underrated. I think being able to be, it is underrated. Yeah, no, I agree. I think, and I think that transition's come back. I think basically it went from delegate, delegate, delegate to actually, you should delegate a ton of shit. But there's a lot of things that you should be on the weeds on and there's a lot of things that you personally should micromanage and those are the most successful companies. And I think it's that balance. It's not one or the other. And I think people have over-indexed in either direction over time. Yeah, you should actually micromanage itself. That's the highest server and the most important things. And that you're actually good at. Yeah. Yeah. Thank you. You lots of comment? I have a fun. Thank you so much. Thank you so much. Thank you so much. Thank you so much. so muchm