The Exchanges

Q I'm going to do a quick fire round with you. Uh, you have to go and be a public company CEO, I know. Um, so what have you changed your mind on in the last 12 months most significantly?

A I do think that, that I've, I've become more convinced that software is headless in the past year than I was maybe three years ago. And it's because of the, the level of agentic capabilities on tool calling and searching across systems and the accuracy of that. Uh, and that, that has happened faster than I, I would have, uh, perceived. So two to three years ago, if you were to kind of You know, wire up an agent and tell it, hey, go work inside a box and find a document to work with and do some process. It would, it would basically almost always find the wrong document and it wouldn't be able to handle actually like cracking open the file and reading through it. And so thus, you know, going headless wasn't sort of the, the most urgent priority, uh, from an agentic standpoint. And in the past year, those capabilities have just absolutely accelerated. To the point where I'm fully convinced that you just, you have to be, you know, headless first as a software platform.

AI assessment note: “I've become more convinced that software is headless in the past year”

Q bunch of checks as we have done for the last 10 years, and a next generation, or to your point, a back to the future era of venture capital where you co-found the business side by side? Can they run side by side? Or are we actually entering an era where we're back to the future era, as you say, where value accrual is in the co-founding and incubation side?

A Um, I, I think it's very hard for them to coexist inside of one person, and it's very hard to coexist sometimes inside of even one firm because, you know, the reason I'm sitting here at Periodic Labs, I work here three days a week. Every day from eight a.m. to 8:30 a.m. for the last year, Liam Doge and I have had a stand-up every morning where we go through the priorities of the company, and then we, we make them, we prioritize, we go and execute. I mean, the compute team of AMP is sitting upstairs, procuring compute for, for the periodic guys. My role models have always been the Arthur Rocks and the Bob Swansons and the Mike Marcola, personal computing. Effectively, the first CEO for the first year of Apple was Mike Marcola. He was an angel investor, and he was the one doing all the CapEx, you know, supply chain and capital and all of that stuff that allowed Steve and Jobs and Woz to focus on the product and the engineering. And, and that kind of deep partnership is what I get really excited about.

AI assessment note: “it's very hard for them to coexist inside of one person”

Q When you think about that, is continuous learning the next breakthrough that you're most excited by?

A I think there's quite a few things that are missing. There's, there's continual learning. I think there's a lot of, uh, I think a lot of mileage in looking at different memory systems. Um, at the moment we have these long context windows, which are kind of a bit brute force. You just put everything in them. Um, I think there's, there's, there's a lot of, uh, interesting, probably architectures to be invented there. Um, and then there's stuff like, uh, longterm planning, you know, hierarchical planning. These systems are not very good at planning at long time horizons, you know, many years into the future, uh, which we is, you know, with our minds we can do. So, um, there's quite a lot of, uh, problems I think that are still left to overcome. Maybe one of the biggest is consistency. So, you know, I sometimes call these systems jagged intelligences because they're really amazing at certain things, uh, when you pose the question in a certain way. But if you pose a question in a slightly different way, they can actually still fail at quite elementary things. So a general intelligence shouldn't be that sort of jagged.

AI assessment note: “I think there's quite a few things that are missing. There's, there's continual learning.”

Q You're gonna leave me with a cliffhanger gockle. You're like, there's some very durable characteristics. We can talk about them if we want. I would love it if we could talk about them. Can you please help me understand?

A It's basically a play on Hamilton Helmer's seven parts, but it's slightly different. I call it the eight motes. The first mote is data mote, which we all talk about, but it's truly has to be proprietary. It has to be data that nobody else has access to. I think Spotify is a good example. If you look at their Uh, the discover product, it uses a decade of listening behavior across hundreds of billions of people. You can't create that listening product, that, that discover product easily. Second is the workflow mode, which a lot of people argue it's a weak mode. I agree by itself. It's a weak mode, but the deeper you're embedded in the company, running their operations, moving their money, the, the deeper the workflow mode is just by itself. I don't think it's enough in perpetuity, but the deeper you embedding is, for example, NetSuite is an ERP that runs your business. They have a much, much deeper mode than say Zendesk. Which is a lighter workflow mode. So that is a mode. You can say it's one, maybe Zendex is a zero point, financially it is a one. Third one is regulatory mode. So licenses, uh, capital require multi-year procurement contracts. Coinbase, when I'm on the board, is a great example. They have MTLs, money transmission licenses, state by state. There is with the Fini, CN, all of those things. It makes it impossible for a company to use anybody else than Coinbase to cus…

AI assessment note: “I call it the eight motes. The first mote is data mote”

Q product engagement? And what I mean by that is, if you want simplicity, it wouldn't necessarily be depth of product engagement. Hey, make me a website that's like BBC News, but make it just for VCs and for funding rounds. Boof. That's not very deep engagement, but it could be a happy user. How do you think about that product engagement need versus actually just what's best for the user?

A So I think about it on multiple vectors. There's intensity of engagement. So how deep do I go? How much time do I spend? How complicated of the task that I'm trying to accomplish? That's one vector intensity. And you're right. Intensity is very powerful for social platforms because the more intense you are on them, the more The better user you are for them. For simple productivity tools, intensity is almost an anti-metric because intensity means like I'm getting stuck, like I'm doing too much where this actually is supposed to be easy. So intensity is often the anti-metric. However, there's also frequency. So how often do I come back and do it? Frequency is a big one. You always want to be in a habitual zone for our minds. Habitual zone for our minds is somewhere on daily or weekly basis. Anytime you move into being monthly, You're in the forgettable zone. I don't remember what I did last month. Like the world is moving too quickly. I don't even remember if I looked at the web, like some software last month, or even if I tried it last month. So trying to be in that daily or weekly habitual zone is super important. And then on top of it, you look at just what actions are meaningful. The worst thing is when you create frequency of engagement based on logins. That's a vanity metric. Everybody can log in. That doesn't mean they get value. So setting up that frequency engagement on …

AI assessment note: “For simple productivity tools, intensity is almost an anti-metric”

Q product engagement? And what I mean by that is, if you want simplicity, it wouldn't necessarily be depth of product engagement. Hey, make me a website that's like BBC News, but make it just for VCs and for funding rounds. Boof. That's not very deep engagement, but it could be a happy user. How do you think about that product engagement need versus actually just what's best for the user?

A So I think about it on multiple vectors. There's intensity of engagement. So how deep do I go? How much time do I spend? How complicated of the task that I'm trying to accomplish? That's one vector intensity. And you're right. Intensity is very powerful for social platforms because the more intense you are on them, the more The better user you are for them. For simple productivity tools, intensity is almost an anti-metric because intensity means like I'm getting stuck, like I'm doing too much where this actually is supposed to be easy. So intensity is often the anti-metric. However, there's also frequency. So how often do I come back and do it? Frequency is a big one. You always want to be in a habitual zone for our minds. Habitual zone for our minds is somewhere on daily or weekly basis. Anytime you move into being monthly, You're in the forgettable zone. I don't remember what I did last month. Like the world is moving too quickly. I don't even remember if I looked at the web, like some software last month, or even if I tried it last month. So trying to be in that daily or weekly habitual zone is super important. And then on top of it, you look at just what actions are meaningful. The worst thing is when you create frequency of engagement based on logins. That's a vanity metric. Everybody can log in. That doesn't mean they get value. So setting up that frequency engagement on …

AI assessment note: “For simple productivity tools, intensity is almost an anti-metric”

Q Is there a parallel to a prior time where you can remember this? Whoa, immense uncertainty. I do not know what happens. I'm better to sit out and watch because I don't want to see Monday drop another.

A Yeah. But so March of 2000, uh, almost 26 years ago, tech stocks dropped between 30 and 40% across the board. And then if you missed a quarter, you were down over 50 or 60%. And we lived in that malaise Uh, March through the summer. And by the way, we were still able to get a couple of IPOs out at, but at lower, um, multiples during that time. And we just sat in misery, um, until nine, 11, which what I call was the coup de gras to finish us off. And, uh, the market got destroyed and we were thinking, Hey, you know, we don't do.com investments and insight. We're going to be fine. Cause we knew Then in 99, that there was a bubble and we knew that dot coms were going to blow up. We said, look, it just can't be sustainable. There's not enough commerce. There's not enough people on dial up. You can't do commerce on dial up and there's not enough fiber in the ground. So sure enough, there was a burst. However, the tsunami came in, took out the dot coms and then took out all software. We all went down and we were in misery for years.

AI assessment note: “March of 2000, uh, almost 26 years ago, tech stocks dropped between 30 and 40%”

Q How does the UI paradigm change in the world of AI? Like, everyone's like, now we're just all going to be voice. Do you agree with like, it's all voice.

A I think, so voice is amazing for enterprise. I think that, uh, one dynamic UIs and two chat UIs are overstated in consumer. And the best thinker on this is actually Eugenia who founded Replica and now Wabi. She's, she's great on this. And what she would tell you if she was here is that, look, most people don't want to save time. They want to spend time. Okay. And the products are designed by the most high agency people in the world. Like Sam and Elon are the most high agency people in the world. For them, the optimal UI is a chat box where you say exactly what you want and like, voila, there it is. But for many people, they're again, looking to waste time, spend time. They want a browse based interface. They're not quite sure what they want. Can't always articulate it. So I think that in a world where we have intent-based and browse-based, browse-based largely stays the same. And perhaps the future of intent-based is chat. I'm still a little skeptical.

AI assessment note: “browse-based largely stays the same. And perhaps the future of intent-based is chat.”

Q How does the UI paradigm change in the world of AI? Like, everyone's like, now we're just all going to be voice. Do you agree with like, it's all voice.

A I think, so voice is amazing for enterprise. I think that, uh, one dynamic UIs and two chat UIs are overstated in consumer. And the best thinker on this is actually Eugenia who founded Replica and now Wabi. She's, she's great on this. And what she would tell you if she was here is that, look, most people don't want to save time. They want to spend time. Okay. And the products are designed by the most high agency people in the world. Like Sam and Elon are the most high agency people in the world. For them, the optimal UI is a chat box where you say exactly what you want and like, voila, there it is. But for many people, they're again, looking to waste time, spend time. They want a browse based interface. They're not quite sure what they want. Can't always articulate it. So I think that in a world where we have intent-based and browse-based, browse-based largely stays the same. And perhaps the future of intent-based is chat. I'm still a little skeptical.

AI assessment note: “voice is amazing for enterprise. I think that... chat UIs are overstated in consumer”

Q terms of like having a command of technology, and you said earlier about the challenge of shipping products and getting from zero to one, showing that you've had a success building in the past is a great way to prove that you can do it moving forwards. Do you have an unreasonable or an unwavering leaning towards serial founders who've proven that they can do it because of their track?

A Yeah, I'll give you a nuance take on this. So I think that repeat founders working in their domain of expertise Are formidable. Like the Clutch guys sold a company to Carvana. Uh, you know, they weren't super happy with the way that the whole thing, you know, ended up in terms of their startup achieving their ambitions. They went and then started another company out of that also in the auto space called Clutch. It's going extraordinarily well. And they know, you know, they're taking all the shortcuts because they know the market. So I do think particularly in enterprise, um, working in the same domain and, you know, being a repeat entrepreneur is a huge, uh, source of alpha. I actually think conversely in consumer, having a beginner's mind and a high willingness to be embarrassed is a competitive advantage because so many consumer products feel embarrassing and, you know, they're immediately dismissed as embarrassing or impossible or a silly, non-serious thing to be working on. When you're 25 and like the stakes are low, you just want to make something happen in the world. That is a perfect setup. Once you've sold a company, all of a sudden it's like, your venture friends are like, what are you working on? You know, like your girlfriend or your boyfriend's like, what are you working on? You want to sound cool at dinner parties or at the bar, and that slight hesitation to be emb…

AI assessment note: “repeat founders working in their domain of expertise Are formidable”

Q So when you think about market composition for that market and the kind of developer tooling, Uh, space. Does that look more like cloud, or does that look more like Uber and Lyft?

A I don't think it looks like Uber and Lyft, right? I think Uber and Lyft are the, to my mind, the most extreme examples of pure substitutes, and a lot of the sort of price has been computed away. You look at cloud, you sort of have this oligopoly where they all actually have pretty reasonable margins, right? And, you know, you can squint and say, of course they have their specializations, but they're roughly substitutes, and yet they've all done well. I think the, uh, the foundation model companies look a little bit like that. And I think in the apps layer, you're just going to have people that want to Consume the code they generate through a rich IDE and those that want to be closer to the metal, and that's probably closer to AWS Google Cloud than it is Uber Lyft.

AI assessment note: “I don't think it looks like Uber and Lyft, right?”

Q You mentioned, um, OpenAI and Sam as the deal maker as well with this. Very good. I liked it. Um, I heard that you cold called Sam in the summer of twenty-twenty-two. Can you just tell me about that before we do it? Yeah.

A So it wasn't a cold call. We, we cold emailed them, um, and we cold emailed Sam and Jason Kwan. Um, what we had basically done was we went on r slash legal advice, which is basically like a subreddit for asking legal questions. And we grabbed a bunch of those questions, ran a chain of thought product that we had basically built on top of it and gave it to a bunch of landlord tenant. Attorneys. And then we basically said, just like, look at these questions and tell me if they're, you would send the answer. We didn't say anything about AI to the consumer who asked the question. And of 86 out of a hundred questions, three out of three said, this is a perfect answer. I'm done. And we cobbled all that together and we just sent a cold email to Sam Altman and Jason Kwan. Um, the idea was basically, hey, did you guys know that, you know, at this point it was GPT-III and just the API was public. I think The end of 2021 or beginning of 2022, they had an API. Um, did you know it was this good at legal? That was it. That was basically the subject line of the email was like, did you know it was this good at legal? Um, and we met them like pretty recently after that.

AI assessment note: “So it wasn't a cold call. We, we cold emailed them”

Q When you look back at the distribution, the channel strategy, what mistake do you think you made?

A The mistake I think we made is not hiring early enough. We had super strong PMF right from the start. We had channels working right from the start, but we didn't actually have anyone focused on certain channels. So for example, affiliates, we set up over a year and a half ago, and that now is bringing in over, you know, tens of thousands of dollars of MRR Per month. But this was set up a year and a half ago by one engineer in one week, and then no one's touched it since. And what we should have done is, once we've seen signs of life, given we already have PMF overall, staff one person who can focus on growing that one product, and those one set of KPIs, and nothing else. Keep them laser focused. So yeah, overall, I wish we had staffed up more channels with more dedicated people sooner.

AI assessment note: “The mistake I think we made is not hiring early enough.”

Q I have not heard of this. Are we reading? Carcerating Titanic? Like, why do we need motion designs?

A So video is hugely underrated as a medium for growth, and so every launch video we do, the keystone marketing piece we ship with it is the video. There's different types of videos you can do. You can do motion design, which is basically animated graphics. You could do a founder-led video where they're speaking to the camera. You could do a more like screen share style video where they're kind of recording and interacting, and it's very product focused. But why I love motion design is it's abstract, so you can really focus on what are the core value props. You can make it catchy and attention focused and mix in the abstract UI elements, and the big mistake that I see with most people's videos is that they put the founder themselves, they do a full five minute monologue themselves about the company mission, and unless you have the editing skills of MrBeast, you're just not going to keep people for the launch video. So make it sure Get the key message across in the 1:30 seconds, and motion design is fantastic for this.

AI assessment note: “why I love motion design is it's abstract, so you can really focus”

Q Work ethic gets you a long way. In the early days, you built the MVP with just a million bucks, right? Before any outside funding. Can you just talk to me about that and what you'd advise founders on MVPs and going to V-one with minimal resources having done that?

A Yeah, so yeah, so at the time we wanted to build a full stack energy company, and we had a million bucks. So what we need for full stack energy company, we need power generation, we need a license, we need someone to help us guide through regulation, we need a qualified trader, so you need to be qualified to trade energy, and we also need a qualified electrician. So we found a single wind turbine in north of Scotland, um, that the seller is willing to sell for 750 K. Uh, we found a license for 75 K. Um, we convinced, uh, the former CEO of Ofgem to become an advisor with just equity and no cash. Um, and my co-founder Charles became a qualified trader and electrician. So we acquired all the skills and assets we need to build a MVP for Sac Energy Company. So my advice is I don't think you need a lot of capital. You don't need a huge seed round to get started. You can just

AI assessment note: “we found a single wind turbine in north of Scotland... willing to sell for 750 K”

Q We've covered quite a few different products from, you know, your WeTransfers to Evernotes to HQ Trivia Clones to your Play Ons. How is the team structured? I think people have this conceptual idea That basically Bending Spoons acquires companies, fires the team, and centralizes engineering and product.

A You can think of Bending Spoons as a horizontal platform to which we plug individual products. The platform is basically everything that you don't need to have at the product level. For instance, let's make an example, talent and recruiting, or accounting, finance, legal. But not only that, also Deep technological teams that power like library data pipelines, monetization libraries, stuff that can be adopted by each product. This way you don't need an Evernote recruiting team. You don't need a common monetization technology team. You just need to have these pieces at the platform level, and then to this platform that where we invest a lot, you can plug new products all the time. And so this means that the individual product team Can be much, much leaner than it could be as a standalone company. And often we, we see the difference that it makes. Like we can acquire companies that are very large and run them with very few people in comparison.

AI assessment note: “You can think of Bending Spoons as a horizontal platform to which we plug individual products.”

Q Why would you need a custom model? When you look at a lot of the customers that you mentioned there, for the ones where you have FDs who go in and build custom models, what is the reasoning around that? And is that a temporary moment in time? Or is that a permanent requirement from them for a certain reason?

A I think it's a permanent requirement. I'll give you an example. Um, let's pick an insurance company, right? Uh, and for insurance companies, two really important problems they have to solve are underwriting and claims processing. Let's pick underwriting, for example. So with underwriting, the problem statement is you might get multiple types of unstructured medical data. Uh, it could be Somebody taking a picture of their medical history on a, on their smartphone, or it could be some OCR data from somebody's medical history or data in PDFs, et cetera. And a human has to look at that person's medical information and then decide, is this person high risk, medium risk, or low risk? What medical conditions do they have? Do they have cardiovascular? Do they have renal? And how do you price insurance for somebody like this? Do you even take them on as a, As a client, if you're an insurance company, right? Now, this is a problem that, um, an LLM can solve really well with a human in the loop system. Now, you may not need a trillion parameter world model to do a task like this. Um, in fact, uh, there's lots of research that shows a smaller language model will actually be faster and more accurate at a task like this, um, than a giant world model. And the insurance company also may not want their data to go back to like a frontier, uh, model. So oftentimes in these cases, uh, what we woul…

AI assessment note: “I think it's a permanent requirement. I'll give you an example.”

Q Were you nervous about that? And I don't mean that badly, but it's a little bit like a GP at Excel or Sequoia or you name it, spinning and doing their own. It's like, gosh, I've crushed it with this, but I, I did have some help.

A Oh, I absolutely did. I think a few things are like, one, it's fun to make it a little harder. Absolutely, it's gonna be a little harder if Sam's not involved, but like, that's good. It's good to be pushed and have to, like, work for what you want. I also think I was really fortunate that I kind of grew up with a Sequoia Founders Fund level, like, training and background. Like, we all lived in a house in the mission, and all the best people, whether it's like a Peter Thiel or a Keith, are coming by the house talking about investing, You know, I think we had about 17 unicorns across the fund, because it's just a small world, and everyone's coming by, so I'm like, oh, these are the founders of Instacart, or Reddit, or Rippling, or Boom, and it's like, this is the bar. So I learned the bar, and I got my, like, really amazing education, so I felt pretty comfortable going in, and I think my sort of ace in the hole is, dude, Harry, I, I can't fail. You know how embarrassing it would be if this fund, like, shat the bed? I can't fail. And I think it's a great motivator.

AI assessment note: “Oh, I absolutely did. I think a few things are like, one, it's fun”

Q Did you notice that different LPs wanted different things? Endowments, foundations, pensions, corporates, family offices. Did you notice a A disparate communication requirement.

A Of course, they're all incentivized completely different. A fund of funds or someone that works at a family office is incentivized to get the best returns possible because they're paid out on the upside. An endowment or a foundation, for the most part, just wants to keep their brand intact, have a nice relationship with you, make sure you don't do anything dumb, but like, why do they care if you're a seven X fund or a two X fund, at least financially? So we absolutely saw that as like for the more traditional Institutional capital. Like, give them a story and a product that feels good for what they're invested in, for a family office or a fund of funds. Tell them how much money they're gonna make if they invest with you.

AI assessment note: “Of course, they're all incentivized completely different.”

Q Did you notice that different LPs wanted different things? Endowments, foundations, pensions, corporates, family offices. Did you notice a A disparate communication requirement.

A Of course, they're all incentivized completely different. A fund of funds or someone that works at a family office is incentivized to get the best returns possible because they're paid out on the upside. An endowment or a foundation, for the most part, just wants to keep their brand intact, have a nice relationship with you, make sure you don't do anything dumb, but like, why do they care if you're a seven X fund or a two X fund, at least financially? So we absolutely saw that as like for the more traditional Institutional capital. Like, give them a story and a product that feels good for what they're invested in, for a family office or a fund of funds. Tell them how much money they're gonna make if they invest with you.

AI assessment note: “Of course, they're all incentivized completely different.”

Q So can we just unpack this? You said about the Oracle credit default swaps. Can you help me understand what's going on there and why that's important?

A Okay. Oracle has a big deal with open AI. Oracle needs to build lots of data centers to build those data centers. They borrow money like a mortgage and they've borrowed money. And there's a thing called a credit default swap, which you may remember from the great financial crisis, which is the odds that Oracle defaults on its debt that they cannot pay their mortgage. Google and Microsoft and other major technology companies are, um, at a certain level, which is basically the same rate as the federal government, right? So government grade and Oracle is three times that in the last three or four days. So it's a big move. It's the, the risk is still quite small. So the overall probability of an Oracle default is small. The magnitude of the move suggests a meaningful repricing of risk.

AI assessment note: “The magnitude of the move suggests a meaningful repricing of risk.”

Q Do you think there are any screaming flags from the last week?

A I don't think so. I mean, most of the hyperscalers GPU capacity is sold out for the next two years. Um, they generate cash. The debt as a percentage of free cash flow is really small. I think the major red flags for me are customer concentration risk is higher than it's ever been. So Nvidia, two customers for Nvidia represent more than 40% of revenues. Four percent represent more than 50% of revenues. I went back and looked at the dot-com era, the networking companies. NVIDIA is 10 times more concentrated in terms of revenue than Lucent was. So I think that's an issue. But most of NVIDIA's customers are super cashflow positive, right? Google and Meta and others are spitting out cash, and they can decide to stop at basically whatever point. Um, so I think it's all okay. How does this merry-go-round stop? Like, if the game of musical chairs were to collapse and everyone falls on their But, uh, in the seat, what happens it's inference demands slows. And if there's a hiccup, if Google says we built this amount of capacity and we can only fill 80%, if that happens, then you see a.

AI assessment note: “I think the major red flags for me are customer concentration risk”

Q So can we just unpack this? You said about the Oracle credit default swaps. Can you help me understand what's going on there and why that's important?

A Okay. Oracle has a big deal with open AI. Oracle needs to build lots of data centers to build those data centers. They borrow money like a mortgage and they've borrowed money. And there's a thing called a credit default swap, which you may remember from the great financial crisis, which is the odds that Oracle defaults on its debt that they cannot pay their mortgage. Google and Microsoft and other major technology companies are, um, at a certain level, which is basically the same rate as the federal government, right? So government grade and Oracle is three times that in the last three or four days. So it's a big move. It's the, the risk is still quite small. So the overall probability of an Oracle default is small. The magnitude of the move suggests a meaningful repricing of risk.

AI assessment note: “The magnitude of the move suggests a meaningful repricing of risk.”

Q You said about synthetic data, and that also being a very important segment to consider. Do you get model degradation when you get this kind of reinforcing loop of models learning on synthetic data, which creates more data for synthetic, and it actually degrades, or does it improve?

A It really depends how you're generating your synthetic data. So in some domains, if you think like images, languages, like LLMs talking to each other at some point, you definitely get the degradation and that degradation is due to essentially like a loss of diversity of your data. So, you know, you can make an analogy, you know, you, you take a bunch of people, put them on an island and let them reproduce, you know, at some point the genetic diversity is gonna keep shrinking. Um, and so you get a reasonably similar phenomenon with, uh, with models. Because you're not injecting diversity into the data. So for their domains where lack of diversity means you get a collapse of distribution. There's other domains where, um, you don't need diversity. If you think of like, you know, playing chess, playing go, these kinds of games, we know exactly how to generate board configurations. And so we can generate tons of synthetic data. Not endless because it's a closed world, but still tons of synthetic data and through that learn for a long time. Then there's domains that are sort of in between. If I think of coding, we can generate synthetic code. You take normal code and we know how to inject diversity into the code. Like I can take a couple of repositories, mix and match, apply an LLM to transform it. And so there's a way to generate synthetic data. The language is predictable enough, a…

AI assessment note: “It really depends how you're generating your synthetic data.”

Q said that I could just take this in any way that I wanted. I find it difficult when you have a public persona to actually separate who you are from the showman. You know, you've got to be on show when you're doing podcasts, when you're speaking, and I really struggle with that. How do you think about retaining your true self when you also have to be a showman?

A Well, I have thought about this very consistently since the early days because You have to be careful which mask or masks you wear because you can become the mask very easily, and I think that's never been more true than now. That's the first warning. I would say another is that if you simply respond to your audience by doing more of what they respond most strongly to, you will end up exaggerating tenfold all of your most extreme behaviors and views. And over time, what does that mean? You will become the embodiment of those things, even when you're off camera, off mic. And to try to counteract some of those, I have done certain things. For instance, you probably have noticed, I'm sure a lot of people have noticed, I have not fully embraced video. I have not gone into full production, television studio land. I have not tried to spend a A lot of time on algorithms on YouTube as a driver of growth. And part of that is because I think video puts all of those inclinations and all of those risks on steroids. It also happens to obliterate your privacy just through giving you broader facial recognition in the world. And as I start looking towards future chapters with parenting and kids and so on, I want less and less of that.

AI assessment note: “to try to counteract some of those, I have done certain things.”

Q last few months, actually, is like sponsored UGC, which is like these UGC platforms, often at universities or around young communities, where they say, hey, give us a 25,000 dollar spend, and they literally get like a thousand people at university to post a review of your Chipotle Dinner. I'm just making it up. And they give them 25 bucks. How do you feel about sponsored influencer elements like that?

A I think they're great, and I, the UGC example is a really good one we should dive into, because UGC itself, one of the goals is distribution. But we also have to think about the other byproduct that you get out of UGC, which is that you now have a video asset that you can use for other types of surfaces. Think about all the places that you could Add that to on your homepage, on your website overall. One of the most powerful UGC strategies actually that I've seen and has worked well for a lot of the companies that I work with recently is turning UGC into a whitelisting ad strategy. So for those who maybe aren't familiar with the concept of whitelisting, it's essentially where you work with a third party who's not necessarily directly employed by your company, and you can actually promote your content directly through their account. There's a lot of benefits to doing that. Firstly, that It feels less salesy when it's not you that's talking about the product that's at hand, but instead basically promoted through the name or the persona of somebody else who seems to be a customer of the product itself. So whitelisting actually has great effects for essentially paid social ads where now instead of let's say that you're Descript and you have to promote all of your content through the Descript account, now the Creator that you're working with, um, let's say, you know, podcaster Pat Fl…

AI assessment note: “I think they're great, and I, the UGC example is a really good one”

Q terms of the different models there. Again, sorry to cite it, but it's, it's kind of handy having just done it. Jonathan said that you would definitely have OpenAI and Anthropic build out their own ships, because then they would have control of their own destiny. Do you think OpenAI and Anthropic build their own ships so they don't have self-reliance on NVIDIA in the way that they do today?

A I think that there is a long history of software companies failing to build chips. The list is, is very large. I think, uh, whether, uh, OpenAI can do it, uh, whether they can do it through partnership with other vendors, with Broadcom, with smaller, more innovative companies is an open question. Um, but I, I think that, uh, you know, companies at the size of Microsoft have been unable to deliver, uh, chips, right? I think, uh, uh, there are plenty of examples as you look across the Fang, uh, group where chips were tried. I mean, probably the most successful is Google and they're 10 years in, right? Maybe longer. You know, soft, modern software does not fit well In a chip making framework. I mean, weekly sprints don't work well on two year long projects. Um, you know, move fast, break things often is not the way you think in the chip world. The way you think in the chip world is measure twice before you cut once because your bugs cost you six months and tens of millions of dollars. And so it's a very different mentality. And where there's been success, it has frequently been acquired. Apple got into the chip business through buying PA Semi. Amazon got into the chip business through acquiring Annapurna. Um, uh, Google acquired the talent from a collection of companies, uh, and then set it in a BU that was a side and under somebody who, who had sort of enormous respect in the org…

AI assessment note: “there is a long history of software companies failing to build chips.”

Q Palta is this kind of weird beast in, in a respectful way. It's amazing. But like, what is Palta for those that don't know? And then we can dive into the machinery of how you build incredibly successful.

A So Palta is a builder. We are a venture builder. It sits in the intersection of current company and venture capital. We create the ideas, and then we partner with founders to make those ideas happen. And we give them everything from the very beginning To make it scale really fast. So we give them growth, analytics, data infrastructure, finance, legal support. We build such category leaders as Flow, Simple, Lave, and Zing. Palt is going to hit five hundred fifty million this year in revenues. We grow each year 50% and even more. We raised around two hundred million from top tier investors. One of them is VNV Global. And palta is avocado in Spanish, so we are building health tech B to C mobile subscription.

AI assessment note: “So Palta is a builder. We are a venture builder.”

Q Why the hundred million? That feels like a bit of an arbitrary number. If you can get it to ten million, and it's super efficient and profitable, why is that bad?

A Because each business gonna be compared to flow, and when you have flow in the portfolio, which is a more than seventy-five million more, six million active subscribers, Second year retention of 80% and more than 65% organic traffic, the numbers that companies can just dream about. You need to match those numbers with the new kids that you are raising, right? We have Simple. Simple is a weight loss management tool. They're going to hit two hundred million in ARR. We have Zing, which is a powered fitness coach with hyper personalized and super engaging workouts. They're gonna hit around fifty million in RR. They grew four times since last year. And we have Lavi. Lavi is an AI cosmetologist making face care accessible. They're gonna hit around twenty million in RR this year. They're growing 30% month over month.

AI assessment note: “Because each business gonna be compared to flow”

Q Palta is this kind of weird beast in, in a respectful way. It's amazing. But like, what is Palta for those that don't know? And then we can dive into the machinery of how you build incredibly successful.

A So Palta is a builder. We are a venture builder. It sits in the intersection of current company and venture capital. We create the ideas, and then we partner with founders to make those ideas happen. And we give them everything from the very beginning To make it scale really fast. So we give them growth, analytics, data infrastructure, finance, legal support. We build such category leaders as Flow, Simple, Lave, and Zing. Palt is going to hit five hundred fifty million this year in revenues. We grow each year 50% and even more. We raised around two hundred million from top tier investors. One of them is VNV Global. And palta is avocado in Spanish, so we are building health tech B to C mobile subscription.

AI assessment note: “So Palta is a builder. We are a venture builder.”