TBPN Live - YC Demo Day Lightning Round, New Snap AR Glasses, SpaceX Rips | Garry Tan, Andrew Lee, Stamatios Floratos, Hugo Frisk, Efraín Torres, Russell Smith, Payton Case, Akshay Trikha, Diana Hu, Harj Taggar, Connor Hayes, Luke Burgis, Anda Gansca

Starting point is 00:00:00 You're watching TVPN. Today's Tuesday, June 16th, 2026. We are live from the TVVIN Ultramal, the Temple of Technology, fortunes, finance, capital, capital. Let me tell you about ramp. Time is money, save both, easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place. Sorry about the delay.

Starting point is 00:00:16 We had a little bit of a technical issue. Glad everyone's here. Thank you for tuning in. It's YC Demo Day. Thank you for never doubting. Never doubt us ever. No, it's okay. Every once in a while.

Starting point is 00:00:28 You can doubt us a little bit. in the comments, pop off in the chat. But we have YC Demo Day. We have one, two, three, four, five, six, seven YC founders joining, then three YC partners. And then we're going around the horn, talking to some other folks. Threads just hit 500 million monthly active users. We're having Connor Hayes back on the show for that. Luke Burgess is here. Burgess is here with his new book, the one and the 99. And so we'll go through that. Not a lot of time for the news, but we'll take you through it anyway. And in the middle of guests, we'll jump in for little news.

Starting point is 00:01:02 Exactly. We love. Exactly. We love the news. The big news is that SpaceX ran like absolute crazy after hours last night. I kept getting push notifications and thinking, that can't be right. And then I would double click in, check a little bit. And I would say, yeah, the headlines are crazy.

Starting point is 00:01:26 Yeah, so the big one is... There's so many of it's just like, it's passing this company, passing this company, passing that company, and you're like, whoa, this is a crazy, crazy time. So, SpaceX sitting comfortably in the top five company. You got Nvidia at $5 trillion. Alphabet and Apple right around four, four and a half. Then Microsoft and SpaceX, neck and just shy of $3 trillion.

Starting point is 00:01:51 And it basically makes their cursor acquisition free. They paid $60 billion in Newspan. stock for the company and their market cap, SpaceX's market cap, more than quadrupled and added more than 4xed the price of a cursor acquisition. I believe they had until Q4 to actually pull trigger on the cursor deal, like they built in a window. I bet you, I bet you there was some terms that weren't public that that cursor would have wanted to make sure that they got in as quickly as possible. Sure, sure. If the stock's running, you're not. It was done really well.

Starting point is 00:02:27 Yep. Who knows where SpaceX will net out? I mean, I thought we'd be at $2.4 trillion by the middle, by basically the last hour of trading on Friday. I was incorrect at the time, but it's obviously continued to run. And there was another scenario where SpaceX could have effectively acquired Cursor at an even higher valuation,

Starting point is 00:02:53 but I think it's a great, Fantastic outcome for the cursor team, obviously, you know, Open AI, it's a big cursor investor, Thrive, A16Z, CO2, and others that I'm fortunately forgetting. But who is Rememort saying about this? He said it's the biggest, the biggest VC-backed, in the past five days, we've seen the biggest VC-backed IPO ever and the biggest VC-backed strategic sale ever. We've never had an MNA of a VC-backed company, young startup, north of 50 billion. Like 60 billion is so, so big. We sort of lose sight of it because we're talking about a trillion dollars for this company, three trillion for that company.

Starting point is 00:03:40 But 60 billion is beyond a home run. And it happened in an M&A, which is crazy. And there's, I would assume, a large number of the retail investors investing in SpaceX, we're not necessarily even that familiar with Cursor up until today. I did see the headline today. If you're waking up today, you're like, wait, SpaceX just acquired one of the hottest AI companies in the world with billions of dollars in revenue. Like, this is so bullish. Obviously, a lot of the more institutional investors or people that had been following the company privately were well aware that they had the option to do this, that they would almost

Starting point is 00:04:23 certainly take it. Yeah. But it takes are all over the timeline. Quinn Thompson says, this is brilliant corporate finance, user newly printed, low float, retail inflated currency to acquire real businesses ahead of the lockup expiring,

Starting point is 00:04:36 probably the most creative, accretive way to sell as much equity as possible into IPO pump. I wonder what acquisition is next. That's very interesting. Nick Carter fires back and says, you'd think the people buying the equity would also be aware of this,

Starting point is 00:04:50 and that's a good point. But the question is, like, are there more acquisitions in the pipeline? Because Tesla has not done a lot of acquisitions. Elon's always been a build, not by operator or founder. Tesla itself. Yeah, that's a very different scenario. What I'm talking about is like there are companies, meta is very acquisitive, Google, Apple. Apple's like on the smaller side, but they still do deals more frequently. It would be very interesting if we're seeing, you know, one deal a month and he's sort of putting together the pieces of buying up Neo Labs and compute

Starting point is 00:05:28 capacity and Neo clouds and rolling everything up. That would just be a completely different operating philosophy. But given where the stock is, given how smooth this deal went and where the market cap of SpaceX is, it's not the craziest thing. It would be a big pivot in his strategy, though. Business Insider has a deep dive on Cursors, Wild Ride, Michael Truel. Didn't pay himself for years. Interestingly, Cursor once made up 40 to 50% of Anthropics revenue. That is a crazy, you know, gyration in the market. It just shows you how quickly things are changing.

Starting point is 00:06:05 Went from, you know, it's all cursor. That business is just funneling to Cursor. Cursor is the front door to this business. Then it's like, oh, it's all these hyperscalers that are spending a billion dollars, half a billion dollars a month. And so the whole landscape is changing. nothing tells that more than anthropic telling cursor that Claude was just a research effort. This goes back to the Dillon Field, like, were they consistently candid?

Starting point is 00:06:28 Or maybe they really just did think, hey, yeah, this is just a research effort. Like, you do you and then realized later like, whoa, whoa, we need to be a player in this. We cannot be in the middle of this situation. So we got to do our own thing. Owning the end customer relationship. Yeah. And so wild, wild ride. Yeah, similar to, I think, the messaging to Figma and Canva around.

Starting point is 00:06:50 Anthropics design tool. Yeah, it's tricky. Every AI company is a general use technology. They're using it for everything. There's questions about like, will you maintain value if you're in the token path? Do you need to be like, do not build a company that

Starting point is 00:07:06 depends on the model's getting better, but if the models get really, really good, like what can't they do? And so constantly we see every company competing with every other company as every hyper-scaler has an LLM at this point, and every SaaS product has other products because they can launch features faster. And so the competition is heating

Starting point is 00:07:25 up across all these different companies. Next time we have Michael on the show, we got to ask how he paid the bills for all those years. What's doing? Crash on couches or something? Got to ask. I don't know. Taking out massive loans against the company, maybe. Anyway, let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI. I own the data platform that powers it. Snap also has some big news today. We're going to get to it. We'll get to that after our first interview with Andrew Lee from Task Lit.

Starting point is 00:07:59 He's the founder and CEO. Andrew, welcome to the show. Thank you for the patience and joining us on such a busy day. How is Demo Day going? How are you? I'm doing great. Thanks for having me. I want to be up front.

Starting point is 00:08:12 I can neither confirm nor deny the SpaceX acquisition rumors. Oh, okay. I see. They're flying around. flying around Demo Day. Elon's probably there. Demo Day, shaking hands, trying to scoop up, roll up the entire batch, maybe.

Starting point is 00:08:24 Who knows? He's got to stop to do it. No comment. Demo Day is great. We're having a good time. There's 1,000 people here. There's 200 companies. Wow.

Starting point is 00:08:33 It's a good time. Amazing. Did you pitch already? Or is that ahead of you still? Not yet. We're in group four, which is going to be this afternoon. And it's actually my co-founder who's going to be up on stage, giving the talk.

Starting point is 00:08:45 Cool. I'll just be the pretty faceover on the same. Okay, well, give us some backstory on yourself and tell us about the business. Yeah, so as background, I was, the thing I'm probably best known for is I was one of the founders of Firebase. So this is my second time through YC I did the summer 2011 batch. We sold that company to Google in 2014. Yeah. I was there for about three years.

Starting point is 00:09:09 This company has been a very interesting story. We actually started in 2020 as a better Gmail. Like we tried to build a new email experience, raised a bunch of money, hired a team. So you quit Google to try to eat Gmail alive. Yeah. I had already left Google a couple years before, but I saw them shut down inbox. And I was like, man, this is a huge, huge product. Like, what are these guys doing?

Starting point is 00:09:33 We should do a better job. This is a terrible idea. Don't do it. Don't invest in it. But just as we were getting ready to shut that thing down, LLMs got good. And we're like, holy crap, we can. take all of this data that you have, and we can fit into a language model,

Starting point is 00:09:49 and we can like draft emails for you, and we can do interesting types of search, and we can do categorization. So we started kind of pulling that string. And around about this time last year, we had this really good agent inside our email client, and we had a bunch of users being like, hey, I love your product,

Starting point is 00:10:06 but could you just get rid of all this inbox crap, right? Like the UI is getting in the way. And so we spun it out, and we had a, it's a totally new product, totally new code base, and this is Tasklet, And we launched that in October, and that's gone super, super well. And that's, you know, that's why we're in YC. And that's gone, you know, at the beginning of the year, we were at about a third of a million of run rate.

Starting point is 00:10:25 We're at a $7 million run right now. So it's like a nice, nice fast ramp. And is that the bar now? Like, is there, you know, are you meeting other companies in the batch that have been able to ramp that, that quickly? I think we have the highest revenue of the batch. There we go. Let's go. There we go.

Starting point is 00:10:44 Well, why don't we go hit the gong? Oh, yes. And then, uh, then we can, boom. That's for you. Um, so, so this feels like, so we were, we're looking at our little summary here. AI agents that connect across work tools, run 24-7 and take ownership of recurring workflows. Uh, does big tech hate you? Like, because I feel like the whole point is wild gardens that what, that's what creates value

Starting point is 00:11:14 accrual. That's what keeps the value accruing to the work tools that, you know, big company CEOs love and extract a lot of value from. And you're coming in here and digging a hole under the wall of the garden and throwing a ladder over the top and digging a hole right through the walled garden. Absolutely. How do you like maintain those integrations if, uh, if, if the actual platform is maybe like hostile. We've seen reporting about this where different companies have tried to sort of integrate across

Starting point is 00:11:48 and then they've wound up with sharp elbows. I think a lot of them have figured it out. But like how do you think about working copacetically with the work tools that you integrate with? Totally. So AI has totally changed the game. You know, two years ago, if you wanted to integrate with a bunch of stuff,

Starting point is 00:12:03 you had to like hand code a bunch of integrations and there were, you know, companies like NADAN and ZAPier that just did that. That was their moat. And our secret sauce is that we can dynamically generate the integrations with AI, like, when you need them. So we have a bunch of canned integration. Some of them we've hand-coded and made them really, really nice. But even if you ask for an integration we don't have, the AI can generate one for you.

Starting point is 00:12:27 And this is super powerful, not just because it lets you go outside the wall garden, but because it lets you connect to things that might not be public at all. So if you're a company, a lot of times you have internal APIs and you're like, hey, I want to have some integration that hooks up to, like, Notion and HubSpot. and Gmail, but then also, like, my random bespoke, like, internal API. And you can't do that with, you know, any of the products from, you know, big tech. You can't do that with, you know, Claude or, or OpenAI. But our product, yeah, sure, you can integrate, you know, via API with this thing, with MCP with this thing, with a pre-can connector with this other thing, and they can all work together.

Starting point is 00:13:01 What's the state of search these days? I feel like we've seen incredible progress in AI and agents. And then when I have the same frustration that you had when you were thinking about working on email, I'll type in some keyword, you know, like YC. And I will get some cookie, some string that's stuffed in a receipt from some coffee shop and it has nothing to do with Y Combinator. The context isn't there. Apple intelligence is rolling out. It takes like five days just to index everything into some sort of rag database. Like, why are we falling behind there?

Starting point is 00:13:38 a key, is that key to your strategy or is it something that it's a problem for someone else to work on? So I think agents, again, have totally changed the game here. So in a short way of our email client, the way we did search was the, you know, the vector database approach, right? We embedded all your emails, we stuck in a vector database. We had this interesting search stack. And the goal there was to have like the search results be good. And it turns out with, if you have a really smart model and you have an agent, you don't actually care that much of the search results are good, as long as you can run lots of searches. So, What we found in Tasklit is we don't need to, like, ingest all your email and index it.

Starting point is 00:14:13 We can hook up to the regular APIs and just run a whole bunch of queries in parallel. Look at the results, adapt, iterate. We just keep refining. And so the results are almost as good without any infrastructure. So it's a lot cheaper. You can connect to more stuff. So I think if you find used Tasklet, you can kind of hook it up to anything. And you'll get surprisingly good search results without any special back-in support.

Starting point is 00:14:34 But I imagine that that's sort of slow right now. And for a company like Google to roll that out to a billion users, you know, for free immediately, that's going to be really expensive. So that's still a ways out. They need to actually chop through that. But we could expect that search will get faster and better over the next couple of years. Yeah. I think that is the big drawback.

Starting point is 00:14:56 And like as an example for, for Demo Day, I wanted to make sure that I chatted with the folks that were here that reached out to me. And I've had people reaching out a period of months. And so I took Fabal 5 back when that was still a thing. And I pointed that at my Gmail and I said, go find everyone who was cold e-belled me in the last and months who might be at Demo Day and give me a spreadsheet with those folks. And it spent like, and stack rank them, right? Like I want to know who should I talk to first and so on.

Starting point is 00:15:25 And it spent like half an hour. It spent like $60 running this thing, right? But the end of that I had this spreadsheet of like 100 plus folks that was like carefully researched and like ordered. And like it found every, like there was no one I could think. of that it didn't find in that process. So if you're willing to spend 60 bucks, if it's worth it to you, like, great. I love it. Are you, last question, are you capital constrained? Like, I'm assuming you have a bunch of offers, maybe you already finished your fundraise, but why even raise? It feels like you guys seem to be making, I would imagine, with that kind of revenue rampant, it'd be hard to not be

Starting point is 00:16:01 making money right now. We are not capital constrained. We actually have already closed a $20 million all around in April with USB and Lightspeed and some other folks. But they're, you know, we're competing directly with Anthropic and Open AI. They've obviously raised a lot more money and I think we can put it to use. So we are raising more money now.

Starting point is 00:16:19 I like just, I like just saying the actual reality because there's a lot of founders that would come on and be like, well, actually like we do something that like, you know, and it's like now everyone is competing all the time

Starting point is 00:16:30 and it's better to just accept it and play the game. 100%. And I think, Corgi set a new bar with, you know, three weeks from, what was it, Series B to C or whatever they did there. So, you know, why wait? Three days. Do it in three days. Three days.

Starting point is 00:16:45 I love it. Thank you for coming on. Yeah, great to meet, Andrew. Congratulations. We'll talk to you soon. Talk to you. Thanks for having me. Have a good one.

Starting point is 00:16:53 Cheers. Let me tell you all about the New York Stock Exchange. Want to change the world, raise capital at the New York Stock Exchange. Our next guest will be sitting in that same seat, slotting in. In just a second. We have Statoos. And we'll be working on Eden robotics.

Starting point is 00:17:11 The microphone. Oh, yes, yes, yes. Good Alexander says, I pay for Starlink. It's a great product. Cursor is also a great product. Codex is a great product. And Claude with Fable

Starting point is 00:17:21 is ridiculous product. That's the main difference between now and 1999. All this stuff is awesome. So total white pell, then read the follow-up. I don't think you can really compare this to 1999.

Starting point is 00:17:33 I think it's actually unprecedented. And if AI doesn't work, the global debt bomb explodes and there's a great depression. The 1999 tech bubble pop, and things were generally fine. Yeah. So it is higher stakes, but it's more impactful. So he's, I mean, it's a good point to not try and draw too much comparison to 1999. Just because it's, it's technology. It's a technology driven boom.

Starting point is 00:17:57 It is very, very different in terms of the financial structure and also the impact, the revenues, the rollout, the opportunity. All of these things, we're just in higher stakes territory, I suppose. Anyway, we have our next guest from Eden Robotics in the YSeed Demo Day stage. How are you doing? Welcome to the show.

Starting point is 00:18:19 Hey, guys. All good. How much yourselves? We're doing great. Thank you so much. Introduce yourself at the company. Yeah, so I'm Stam, CEO and co-founder of Iner Robotics.

Starting point is 00:18:30 We're making general purpose robots for manufacturing warehousing initially and everything else right after that. How general? Are we going humanoid? Are we going wheels? Are we going? Semihumanoid. Semi-humanoid?

Starting point is 00:18:44 What does that mean? Centaur? Yeah, centaur. So the body, the legs are, there's four legs like a horse. But yeah, so semi-humanoid. What does that actually mean? So that means it's a wheeled-based robots. It has wheels, but it has two arms, which are placed like a human.

Starting point is 00:19:02 our shoulders, same way we have, and it also doesn't have Dexter's hand has grippers, which fun fact can do more than 80% of work in industry. Interesting. What about the wheels are important to decide? Like a lot of manufacturing facilities have very flat floor, so I imagine you don't need really any stair climbing ability, which some wheeled robots can do. But do you need to be omnidirectional? Do you need to change path? Is it just about optimizing battery life? Like, what are the tradeoffs that you decide when you're building the base?

Starting point is 00:19:38 Yeah. So when you're building a base, really, when you're going with wheel, it's not going to be as general purpose as legs. Like legs can go anywhere. Yeah. But it doesn't have to be. And you have a massive gain in both costs and energy. Yeah.

Starting point is 00:19:51 Legs are extremely expensive in terms of energy. And also in terms of, like, simply cost. And also, I imagine you can put way more battery pack on. top of like the wheel base, right? Yeah, that's cool. Half our base is just the battery. Yeah. So half the base is just the battery,

Starting point is 00:20:06 and then the other half is the compute. Yeah. And that's it. So how long, is there some sort of like, sweet spot for once you have a robot that can roll around and do work in a manufacturing plant for six hours without needing to recharge?

Starting point is 00:20:21 That's the sweet spot. You know how like 300 miles of range was like the sweet spot for electric vehicles? And under that, you start getting range anxiety. Yeah. So I think six hours is too low. Right now our robot is using a car battery. It can run for 12 hours.

Starting point is 00:20:37 But with our own... Not an electric car battery, just like one of those bricks that's a car battery. Exactly. It's a really big car battery. And with our own battery design, it can go for 20 hours and optimize charging for like these four hours and not have to like stop for four and work for 20. But like optimized throughout the time to charge enough. You got the car battery in there? You're thinking about putting a V8 in there maybe?

Starting point is 00:20:59 Why not? Probably not. So one of the, you know, we've had a bunch of robotics founders on. I've been trying to understand what some of the kind of like economic challenges there's going to be to rolling out, you know, humanoid-esque robots. When I see some of these humanoid form factors, I'm thinking like, look how many different motors it has, how much where is each of, each of the motor is going to be taking at any given point. And then like trying to run the numbers on like the, what is the cost of the unit?

Starting point is 00:21:37 What is the depreciation going to look like over the year? How often are you going to have to be repairing different components to actually make it competitive with a human, which you can hire for, let's say, $30,000 to $40,000 a year, depending on minimum wage in a, you know, a certain area or type of role? because like the you know aside from humans being like cool and funny they also go home after work and they feed themselves and they repair you know they they sort of are repairing them you know repairing themselves while they're sleeping right so that's like sort of off balance sheet

Starting point is 00:22:13 sort of like activity that's happening that ultimately benefits the business because somebody goes home eats sleeps and then they come back to work refreshed ready to go that's a crazy way to describe human workers. But I get it. But yeah, I'm just like comparing, comparing these two things. A robot, it doesn't go home at the end of the day. It might need to be repaired. It might need to be updated. And it doesn't happen unless the company actually pays for it. So the biggest kind of like hardware risk factor is just the actuators. Really nothing else would require maintenance. Everything else would be able to last pretty long time. Right now we're at this point where you have actuators that are actually good enough to last like three years and last

Starting point is 00:23:00 this overusage. It used to be that these actuaries would overheat very quickly and you would be able to run for a very short amount of time. And there are some human noise today. Like, for example, unitary robots, they're very famous for overheating very, very quickly. But there are actors in the market and we are using such actuators that actually can sustain long hours of operation. We ran it, for example, for 10 hours during alumni demo day.

Starting point is 00:23:29 There was no hardware issue when it came to the actuators. And it went pretty reliably. We expect that these kind of actuators would have a useful life of about three years. And in the case of the maintenance, our whole thing, by the way, is we're not selling these robots. So we're selling the labor. And in fact, we're not even leasing them on a monthly basis. We're charging per hour.

Starting point is 00:23:49 So we're charging a $10 per hour, which is actually found that customers love that. idea much more than they love the lease idea because they have no mental framework, right? They have the mental framework of a human worker. And so the maintenance would go is if a customer has a robot somehow breaks, there's a problem with it. And sorry, is it per hour that the robot is actually running? Yes. So you're not just saying, I'm going to charge you 24 hours a day for this period. It's like they're they're only paying if they're getting actual value. It's like a waymo. You're used to paying 20 bucks to get across down. You pay 20 bucks. You pay 20 bucks.

Starting point is 00:24:24 to be across town. Exactly. We do set minimums, obviously, because we need to keep the hardware at the facility, but it's just whatever is enough for us to keep the hardware there. Everything else is adapting based on usage. Well, congratulations. Thank you so much for coming. Did you already get the round done? Not yet, but we're very close. Good luck. Love it. Good luck. Great to meet you. Come back on soon. Have a great one. Thank you. Have a good one. We'll talk to you soon. Well, we bring in our next guest. Let me tell you about Figma. Agents. Meet the canvas. Your AI agents can now create and modify your Figma files with design system context.

Starting point is 00:25:01 Breaking from Semaphore. Yes. Netflix lost out to Fox in pursuit of Roku. How is that possible? I mean, yeah, it is interesting. The lore there, Netflix sort of created Roku. But the real question is, like, why would Netflix want to buy Roku? Like the boxes?

Starting point is 00:25:21 That just seems like it makes so much less sense. So in terms of lost out, I can imagine they lost out. It's a platform. They sit as an application on a platform that represents like 40% of connected streaming. I don't know. I'm not surprised they lost out because I would be surprised if Netflix. Like when Netflix was bidding for Warner Brothers, that made a lot more sense to me. More library, more studio production for Netflix to distribute. But finding a second distributor when they're already competing in the ads-based model, they're moving towards a cheaper distribution thing.

Starting point is 00:25:55 They're already installed by default on most connected TV. I just think it's a way to expand like overall engagement, watch hours. They're already getting into ads. Why would they not want to have like more, basically like more watch hours to monitor? Maybe, maybe, maybe. It seems like in, Roku is in way better hands with Fox than Netflix. So it makes sense why the deal broke that way. Anyway, let's bring in our next guest, Hugo Frisk from Tenet,

Starting point is 00:26:21 industries is the co-founder and CEO. Hugo, how are you doing? What's going on? I'm doing great over here. Yeah. Thank you so much for you doing the time. Introduce yourself in the company. How you doing? Yeah, so I'm Hugo. I'm co-founder of tenant industries and we are focusing on mass producibility and low-cost for the defense system, basically commoditizing it. So I'm from the electronics manufacturing industry and my co-founder, we're three co-founders, but the one of them is from the automotive industry. So basically taking this like mass-producibility, those concepts, and bringing them to the defense industry. How low cost do you want to go?

Starting point is 00:26:59 I mean, we're all familiar with like the couple hundred dollar price point for a DJI product. Where do you want to sit? Yeah, that's about it. It depends on the customer, really. But yeah, we want to match shipper than China is our slogan. Okay. And then how are you thinking about the actual product? We've been hearing a lot about the fiber optic controlled drones

Starting point is 00:27:25 that don't need to deal with wireless jamming or anything like that. Like, where is the demand for the next generation product or is it just what we know from a DJI quad rotor, what we saw there? Are there any specific demands for the defense context that make it a little bit different than what's already out there? Yeah, the main factor is the range, basically. It's range and lift power. How much can you lift?

Starting point is 00:27:49 What's the payload weight? So it's all about increasing the range. And the main limiting factor for the range is oftentimes the radio, right? So what you want, that's why they're using like this fiber radios or this fiber optic cables because then you get longer range without getting jammed. But unfortunately, I think the fiber market, it's kind of only relevant for this bubble in time because to increase the range, which is the main, like what customers want, they want

Starting point is 00:28:20 you can do either a better radio basically you can develop better radios that's maybe the first step and the next step is doing autonomy right if you have autonomy then why do you need any contact at all? Sure yeah everything. Are you capable or do you think it's going to be possible

Starting point is 00:28:36 to do on device autonomy on a drone that size two kilograms that's not enough to run some serious GPU but what is possible these days? No exactly you can put any of your jets on it but the immediate Jetson, it's like five times expensive as the drone itself. Sure.

Starting point is 00:28:53 So then it's not really a one-time use kind of deal. Yeah. So, but there are these embedded ships for, for AI. It's just you can't run an LLM model on it. You can do some visual recognition and maybe you can do some like prioritization of different targets. But the main, the main factor here about the autonomy is, it's not really the vision model. That's like a sole problem. This is possible. The main problem there is, is navigation. How do you get to the near vicinity to your target? How do you arrive in,

Starting point is 00:29:24 like, navigating for 20 kilometers without the GPS, because you can't use a GPS on the front line. It's not possible. It's jammed out. It's super easy to jam out GPS and to spoof it. They will say that you're in China or something while you're in Ukraine or you're in Sweden or wherever. So that's the main problem. How do you navigate without the GPS? Because the last mile, which is basically targeting, that's a solved problem. That was solved in the 60s, missiles. I don't get the companies that are pitching that. What is your, what is your business model going to look like? Or what does it look like today? You're trying to commoditize the category, push pricing down as much as possible. But, you know, what are your margins going to look like?

Starting point is 00:30:06 And how do you expect to structure deals with different buyers? So right now, it's kind of this pipeline. How do you scale a defense company? Because it's not really possible to get this deals as a new startup as deals straight to the state at least not where we're looking in Europe but so what you want to do is you start with with the private companies you sell to them so we're selling to this counter US companies this this month we went from fifth five orders to 350 orders and then you start selling to this subprime it's called like basically you're selling to a company that's selling directly to the government and Those are really interesting because they have already the channels to sell to these governments.

Starting point is 00:30:52 And then once you have established like your manufacturing line and your capability and people trust you, then you can sell directly to the governments. Chat asks, how does their drone survive high power microwave machines like Leonidas? Is the answer that it doesn't, but you can produce at such scale that you can kind of ground these types of... high-powered microwave machines rolled out everywhere, but what is your take on microwave as a counter-UAS force right now? I think microwave is, like we've tested with these people having shotguns and shooting down our drones with shotguns, like radar-controlled, basically. And it's a short-range system. That system only reaches 100 meters. I don't know about this microwave. How long

Starting point is 00:31:42 is to reach that there, but I would guess it's pretty small. So then you need a lot of these systems, and people don't have that. They don't have these systems every 100 meters. Yeah, yeah, yeah. Not every military installation or unit is going to have every possible counter UAS. So it's sort of a game of rock paper scissors on the battle these days. And then the other thing is, like you said, like just send more. Yeah.

Starting point is 00:32:06 Yeah, that's what I was saying. It's like you have, you know, your five exquisite systems, well, we're going to, you produce. Yeah. Is everything hand-built right now? Are you already working on automation? I imagine that low-cost, manufactureability means more on supply chain and design, but what does the actual rollout look like over time? Yeah. So when we started this, we wanted to produce with robots. Still do, but basically what we, our background is this manufacturing background, so we're really good that designing for manufacturing, designing it so it's really easy to assemble. So we designed it.

Starting point is 00:32:45 So how would it be easy for a robot to build this? Well, then it needs a few parts and it needs a few screws and no soldering and it needs to be assembled all from one side. So that was the starting. And then it turns out once you do that, it's very, very easy to assemble. So it takes two minutes for a human to assemble it. So then why do you need a robot? Good point.

Starting point is 00:33:09 batch mate we'll have to make the pitch for you how's uh how's investor interest so far you've been busy the last couple weeks or are you starting a process now yeah yeah super busy it's actually my co-founder emil who's uh who's doing that uh so he's always on call and uh i'm doing customer demos and stuff yeah very cool well congratulations great to meet hugo thank you so much for taking the time to come on the show thanks on all the progress we'll talk to you soon yeah no problem Cheers. Ryan Peterson says with yesterday's 20% SpaceX pop, Elon made more money than Warren Buffett made in his entire career. These headlines are going to get crazy because I mean, I see some of them getting community noted, but it's like a single day will move more than like all of Bill Gates's current net worth, that type of thing.

Starting point is 00:33:56 Those headlines are going to pop off consistently. The crazy one is like it's there's going to be down days. Like, there's going to be just randomly. Elon lost more money today than any person in history combined. He lost more money than Brazil makes in a trillion years, you know. And because, like, whenever the big numbers get contextualized, it's always very entertaining. It was up another 14% after hours, though. But the numbers are staggering when you're in the one trillion dollar club.

Starting point is 00:34:26 J.B. says, is this everyone's first IPO? It's silly now as it approaches Amazon valuation. Well, we pass that, but the float is low until the lockup of nearly every retail investor on Earth wants to be involved. You're going to get stupid moves. Then the float opens up and all the retail people are stuck for years. What does that mean? Just say you haven't seen the mass driver demo, buddy. Or the next data center.

Starting point is 00:34:49 I mean, that's the really interesting thing with Cursor. There's all this debate over like they have a deal with Anthropic. They have a deal with Google. Cursor obviously wants to compute. But talking to Gavin Baker, it sounds like there might be a lot more. terrestrial compute coming online in the very short term, and that is valuable. They're monetizing this very effectively, and so you could see a short-term revenue ramp just driven from sort of the boring neocloud stuff that monetizes really well. And then that provides, you know, it's like

Starting point is 00:35:18 the Model 3. It's going to be the economic engine that provides the capital for data centers in space, mass driver on the moon, all that stuff needs fuel for the fire. Anyway, Ad Ludlow has some data. SpaceX, the current approximate price to sales is 150X. Amazon is 3.6. Microsoft 9.2. See, people read this as a bear SpaceX take, but imagine if Amazon started trading at SpaceX's price to sales.

Starting point is 00:35:49 Yeah. It'd probably be like a $100 trillion company. Anyway, our next guest might be building the next $100 trillion company. We have Peyton Case from Dispatch here, building re-entry vehicles. We're returning space manufactured payloads in semiconductors, biotech and pharma. Welcome to the show, Peyton. How you doing?

Starting point is 00:36:06 Oh, wait, sorry. Do I have the wrong person? I think they were the mix up on this. Oh, we switched around, Ephrae, Torres. Yes. Thank you. Please introduce yourself for us so we get it correct. Yeah, so I remember.

Starting point is 00:36:17 My name is Dr. Efrienne Torres. We're from Adelante, building whole body mobile MRIs for cancer screening. Amazing. There we go. Are there any downsides to constantly getting MRIs? No. No, so MRI is non-cancer causing radiation. Okay.

Starting point is 00:36:30 CT and X-ray, there's limitations. Got it. But MRI, I mean, I've scanned myself thousands of times. Okay. So are you the most scanned person in history? No, my professor takes that one. He likes the town. He's the most scanned and published brain in the history of mankind.

Starting point is 00:36:47 Five times a day or something. Yeah, that's how you get it. How mobile are we talking? Are we talking mobile like it can be in a truck that comes to the town and you go, like getting your body fat measured in a water tank? Or are we talking like Fitbit? I clip it to my waistband and I'm constantly getting MRIed. Yeah, no, we're talking trailer.

Starting point is 00:37:07 So pulled by like a 4.350 shows up outside of a clinic. Let's give it up. I like the 435. There you go, right? We actually calculated it just for that. Yeah. And this whole body's screen, $250 and there you go. 250.

Starting point is 00:37:20 Okay. So what, I'm getting a little feedback right now. Can we do with that? Thank you. So, yeah, sorry. What does a normal MRI cost? Yeah, I mean, Per NUvo, if you look at their pricing, they're $2,500, and it can go up to $3,500. So we're on 10 times cheaper than everything on the market.

Starting point is 00:37:39 And what is driving that? Yeah, so there's two things. The first thing is our technology. Like, our system is 80% lighter, consume 60% less power. It's actually eight times quieter, too. And all of this lets us have the operational gains. So we don't sell the systems. We sell scan. So it makes our internal operation costs really low. Interesting. And the shelf life of these systems are 10 to 12 years. So we can charge 250 a scan.

Starting point is 00:38:09 And actually, a gross margin ends up being 55% because of how long these systems last. Do you have to go through medical device approval with the FDA? Yeah. No, we're going through medical device approval next year as a class to 510K. It's a fairly straightforward path because of the fact that there's a lot of regulatory precedence. where there's regular risk arises when you're like the first MRI, first therapeutic and things like that. Well, then what about patents? What's the backstory on the company and the IP? Did you cook this up with your professor? Is he involved with the business? What's the history? Yeah, I know, not the professor. He's not involved. So me and my co-founder, we've known each other 10 years, went to grad school

Starting point is 00:38:47 together, specifically to spin out a startup to disrupt the MRI industry, if you think about it. And so we developed the IP in grad school, got that pattern issued. Adelante has exclusive rights to it. And, yeah, so it's been cooking for a good bit here. But MRI is a hard problem to solve, and that's why it's been picking us. But yeah, doing well. So you're going through FDA approval. Can you actually sell the product in some advanced form yet?

Starting point is 00:39:13 Or is it just, like, demonstration for medical, for, like, research purposes at this point? Yeah. So I love this question because it's got to one of the creativity of our company. We are deploying this year, but it's a traditional MRIs. Oh. The traditional MRIs, we're going to put them in semi-trucks and it's all FDA cleared. Okay. We can go to our customers.

Starting point is 00:39:33 We have 11 LOIs already at 19.75 million and start building that distribution channel now. Yeah. And over the next year, right, we're going to get five to ten of these trailers out there. All breaking even, but then come Q1, 2028 when our product is cleared, you just switch it and it's like a lever to profitability. That's amazing. Wow. Congratulations. Yeah, okay.

Starting point is 00:39:53 How's the fundraising been going? It's been going well. We only have a small amount of allocation left of our non-leads and having a lot of exciting conversations of potential leads. Are you seeing more interest from investors who have invested in health tech, medical tech, FDA-approved devices, that crowd? Or are there tech investors who are sort of exhausted with the AI stuff and want to broaden their horizons?

Starting point is 00:40:22 Yeah, honestly, we're getting a lot of interest from deep tech, generalists, and just tech. Because MRI is very well known in this space. Sure. Yeah. Everyone has a story of someone that didn't get an MRI in time, whether it be from like a sports injury to something more severe, like the cancer wasn't caught in time. And so we've had interest from deep tech, tech investors, generalists, health tech as well. Well, congratulations. Where's the name from?

Starting point is 00:40:49 I got to know. Yeah. Yeah, so call credit goes to my wife. So Adelante stands for a couple of things. Our scientific approaches are known as radio frequency, adibatic pulses. And my parents saying is, Sigip Adelante, which in Spanish is keep going forward.

Starting point is 00:41:04 Also tattooed on my forearm here. There we go. And so Adelante is very close to Adelante, which means go forward in Spanish. And it kind of goes that no matter what happens to us, we're just going to keep on making progress every single day. Very cool. Thank you so much for coming on the show.

Starting point is 00:41:19 Logo chest piece. No more. We don't need any more. Have a great rest of your demo day. We'll talk to you soon. Yeah. Thank you so much. Like that.

Starting point is 00:41:33 Up next, we have a good buddy of mine, Russell Smith from nine mothers. He's back in YC. I went through YC with him in summer of 2012, 14 years ago. Overnight success. There we go. but I'm excited to catch up with Russell Smith.

Starting point is 00:41:52 Introduce him to everyone. Tyler was, what, one or two? He was probably seven years old or something. Just a boy. Yeah. But he's working on AI mission systems for defense. Can we get Tyler in here for a little bit? Focused on counter drone protection.

Starting point is 00:42:08 Yeah. Yeah, in between the next guest, let's slot in some Tyler for sure. Anyway, we have Russ Smith from Nine Mothers in the TVPN Ultradome in at Wysademaday. How are you doing, Russ? Hey, dude, good to be a long time. Been way too long. Thank you so much for taking the time. I'm glad we can make this happen.

Starting point is 00:42:26 For those who don't know you, introduce yourself. Give us a little bit of the YC history, the entrepreneurial backstory, how long you've been an entrepreneur and then take us through the current company. Yeah, I mean, I was an entrepreneur before I did YC, so like in the UK. But got into YC in the same batch as you. That's how we met in summer 12. Yeah. The batch that basically broke YC at the time and made them.

Starting point is 00:42:49 re-change it. 70 companies. Look, wait, how did it? So YC had been like 30. Yeah, yeah, you can tell it.

Starting point is 00:42:57 Yeah, it's scaled and then they decided to change how it worked. And it's the genesis of how they do groups now, right? It was basically a free-for-all at the time where the loudest companies or the most, like, intense ones would get the most, well, not deliberately the most attention, but would,

Starting point is 00:43:13 would. Yeah, it's sort of natural. Yeah. Get the most love. It's funny. Because it was the YC batch that broke YC. Everyone said,

Starting point is 00:43:19 Oh, I see Zover. It's, you know, it's gotten too big. And then it was like Coinbase, Airbnb or Coinbase, Instacard, Zapier, a couple other companies that came, became huge. Clever. Clever, yeah. Benchling.

Starting point is 00:43:33 Like a bunch of, there were a bunch of great companies. A bunch of decacorns and unicorns. Yeah. It was good. So. And then soil as well, right? Yeah. As in, I remember going to your, going to your tenderloin office and carrying bags of your

Starting point is 00:43:49 early product out in little bags. $1,500 a month for three people, live in San Francisco. I remember that. It was a bargain. Didn't you move in there when you had like $18 grand or something? Yeah, I think we had like $20,000 left in the bank. And you're like, good. It's the last us like 18 months.

Starting point is 00:44:06 Yeah, yeah, yeah. We paid our rent a year in advance, owned our laptops. We just had to pay internet and food, and that was it. It was pretty good life. And that's where Soiland came from, right? Yeah, yeah. It's like, just eat that. It's cheaper.

Starting point is 00:44:17 Anyway, we're not here to reminisce about the old times. Tell us about nine mothers. Yeah, so we make AI mission systems for the DoD, and our first product is counter drone. Okay. And we make a very small robot. So it's 35 pounds. We sell it for about 150K.

Starting point is 00:44:31 Okay. And it's capable of defeating multiple fast-moving drones day or night. Kinetically? Kinetically defeat. So, yeah, it's basically a gun on a robot. And then we build all of the models for day, night, uh, builds acoustic sensors, build models for that. And then basically,

Starting point is 00:44:49 capable of shooting down small, fast targets. Interesting. Can we pull up the video? Because I have no idea what this robot looks like. Yeah, let's pull up the video. You should have it with me. I have it. Oh, yeah, yeah. We'll have the team pull up the video.

Starting point is 00:45:02 I'm curious about the decision around acoustic models versus video models. There's so much energy around. We do all of it. You do all of it. So, like, the key here is that no sensor is perfect, right? And so multiple sensors gives you more chance at finding something.

Starting point is 00:45:20 And then, for instance, acoustics is bad on some of the projects we're doing that are around turbines, right, on Chinooks or other vehicles because it gets drowned out. But then radar is bad in some environments, and vision is bad in some. And so the fusion, the ability to fuse all that together into coherent picture and also adapt to new sensors means that we can stop more things, right? And the key here is that this is kind of like the same problem that you saw in Iraq or Afghanistan down with IEDs. These are just now flying IDs and all the demos you've seen existingly almost all of slow moving targets. So like sub 10 meter a second, sub 10 mile an hour, in some

Starting point is 00:46:02 cases drones, people are struggling to shoot down. And that's just not reality, right? As in we've been shooting and demoing as we shoot faster internally, but demoing 65 mile an hour small drones repeatedly. And that's way closer to reality, if that makes sense. Where does the name come from? Dude, it's North mythology. So we're nerds, and then our product names are all around the same mythology. And so naming's hard, but yeah, that's where it came from.

Starting point is 00:46:33 That's the story. That's good. What does it actually shoot? Is it shooting like 5-5-6, like some commodity round? We can adapt to anything, as in we can do 5-5-6. but the thing that works the best is 12 gauge. And so we make our own, we use a gun off the shelf at the moment. We're also developing our own belt-fed shotgun system,

Starting point is 00:46:54 like as in the world's first belt-fed. And that will be first firing in the next three weeks. Yeah. We also make ammo, right? You make ammo? Yeah, it's basically closer to match grade, which is like tighter tolerances and then slightly higher pressure. But owning everything, like the full stack,

Starting point is 00:47:12 that means we have better results. It's a short version. So we do raises and blades for defense, right? I mean, you're obviously very experienced founder. Where's the company right now? Do you already have a facility set up, staff hired? Like, I imagine that you're pretty far along.

Starting point is 00:47:28 For YC, we're a little later on. So we are, we've sold like $1.6 million of stuff to the DOD as in direct sales of units, not research and dev stuff. like actual sales, delivered units to DOD, and then we're expecting to sell over 100 units this year and 1,000 next year. Have multiple believable paths to both. But yeah, aim is to do over 100 Mille of RIF next year, and it's looking completely believable. Should hit 50 this year. Super impressive. Question for you. How, let's say there's a front line and you set up, you know,

Starting point is 00:48:05 I imagine, you know, let's say the U.S. military sets up a bunch of these drones are flying over. how do you manage friendly fire? Let's say a drone passes the line. There's actual troops back here and you want to take out the drone but you're effectively shooting rounds into the air. It's a great question.

Starting point is 00:48:27 And it's a complicated answer. And it depends on the mode the system's in. We can ignore stuff that's flying away from us and only prioritize stuff that's coming towards us. We can also take signals off networks they have if they know which drones are theirs. Really the common thing is that they don't actually have that kind of data. And so the operator can choose to not run it full automatic and then choose which targets. But mostly they will shoot them because of blue-on-blue risk. If it's not

Starting point is 00:48:57 drones, they are physically flying in the same group, right? That's the problem. Is it a friendly drone that's misidentified you or not? If it's on a network and we know it's not, has it still misidentified you. That's where you get into the problem of this, at this kind of range, right? Well, congratulations of the progress. Really exciting growth. Great to catch up. Congrats to you on the show and we'll talk more soon. Have a great one. Yeah, great to catch up. See you soon. Goodbye. Let me tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stopped breaches. Our next guest is Peyton Case from Dispatch.

Starting point is 00:49:38 I accidentally introduced him a little bit too soon, but we got him sitting down in the YC Demo Day. What's going on? How you doing? Yeah, it's good to be here. Good to see you guys. Good to see you, too. Introduce yourself.

Starting point is 00:49:52 Introduce the company. Tell us what you're doing. Yeah, so I'm Peyton. I'm the founder of Dispatch Space. We build satellites for manufacturing products in space, and these are for the most advanced materials in the world. that can only be made in microgravity, and we bring them back down to Earth.

Starting point is 00:50:06 What specifically are you building? Because the space economy has gotten so fractured at this point. There's multiple launch providers, there's satellite bus providers, there's impulse space. Tom Mueller has an extra propulsion system that you can bolt on. Like, what is the key product

Starting point is 00:50:26 that you're focused on for this early stage of the company? Yeah, we build the reentry vehicles. So this is the satellite is in space, You've made your product and now you need to get back down to Earth. You're flying at Mach 25. You're surrounded by a plasma that's 3,000 degrees. And we build the heat shields and satellites that can survive that environment. Who is doing stuff in space?

Starting point is 00:50:46 I live in the world of Varda very closely. Yeah, yeah. Elephant in the room is that we are very familiar with the catatroy because Delian's our friend. And it seems like obviously they're going vertically integrated, building the reentry vehicle and also doing the contracts with biotech companies and pharmaceutical companies. But it does seem like the industry is expanding. So who are the other buyers that aren't working with Varda that you have an opportunity to work with?

Starting point is 00:51:16 Yeah, the biggest players are semiconductors, pharmaceutical, and biotech. Varda is very vertical and that pharmaceutical application. We're the biggest bulls on the semiconductor space. We think that the regulatory environment makes it so that, you can come to market a lot faster, and the applications of growing those gallium nitride and indium selenide crystals in space are massive. And that's where we're bullish on it. Massive. I'm down. Everything that touches semiconductors is massive. You say semiconductor, you're a trillion dollar company. But timelines, the timelines, like, it feels like we're a decade away

Starting point is 00:51:52 from making a chip in space, making a chip on the moon. Like even the big SpaceX bowls are saying, Okay, actually doing the mass driver, that's maybe more than 10 years away. So maybe another question is like, what traction are you talking about this demo day? Yeah, yeah, yeah. Yeah, so we have a pilot mission that we're flying with a paying customer to grow semiconductor crystals in space. And we're not talking about doing like the lithography or assembly and iPhones in space. It's about it's earlier in the supply chain. So wafer.

Starting point is 00:52:20 Wafers actually come from these cylindrical crystals. And when you grow those in microgravity, you get up to a thousand times fewer defects. so you grow those crystals in space, bring them back down to Earth, and then you cut them into wafers. And that is actually happening now on the ISS. And these companies are looking to scale. They just don't have the vehicles to do it. Oh, very cool. How did you make your first dollar?

Starting point is 00:52:41 The first dollar was the deposit on the pilot program. No, no, no, in life. In life. I want the founder origin. Oh, yeah. It was probably working in a smoothie shop that I totally loved the smoothies of, wink, wink. I made a few tips. That was those about it.

Starting point is 00:52:59 You realize you weren't cut out for the smoothie business. You were meant to go to the heavens. Yeah. Fabs and space are a little easier than getting tips at a smoothie shop for me. That's amazing. What's the state of the space investors that have attended demo day? Are you seeing traction? How's the fundraise going?

Starting point is 00:53:22 Yeah, the fundraising is going great. Obviously, IPOs are very useful for us. That's been great indicator of where the entire market is going. We're getting close to closing out that round, which is super exciting for us. And really, that goal is getting back to work as quickly as possible and using that money to go build great things in space. That's fantastic. Dan in the chat said from smoothies to a smooth opera. That's great.

Starting point is 00:53:45 Well, thank you so much. Great to meet you, Peyton. Coming on the show. Have a great one. Thanks so much. We'll talk to you soon. Goodbye. Our next guest is Akshay from Madrone, building cooling systems for data centers that are more power and water.

Starting point is 00:54:01 I'm assuming he's reducing water consumption in data centers so that almonds can water. Yeah, yeah. Yeah, this guy's an almond industry plant for sure. Anyway, let me tell you about railway first. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web app, servers, databases, and more while Railway automatically takes care of scaling monitoring and security. Thank you, Jurydy. Let's bring in Akshay from Madrone.

Starting point is 00:54:27 How are you doing? Welcome to the show. Thank you for taking the time. What's up? Hey, guys. Huge fans, so appreciate the opportunity to talk to you guys. Fantastic. Glad to have you on the show. Introduce yourself, introduce the company. I have tons of questions, but let's start with the basics. Sounds good. So I'm building my drone. We're building hyper-efficient cooling for data centers. So specifically, we've engineered a thermonatomic process that works really well in hot and dry climates, which is where most hyperskilling is happening right now.

Starting point is 00:54:53 And yeah, we have our demo that's writing live at Demo Day today, so we're actually really excited to show people in person. But the big problem really is that data centers are limited today by power and not by chips or land. So cooling and power are two sides of the same coin. If you can save power on cooling, that's more power for your flops and that's where our real value unlock is. So people can install more GPUs, given the same grid permit, which takes years to get.

Starting point is 00:55:16 To give you a sense of the scale of the problem, last year in Texas, semi-analysis said, that there were 150 gigawatts worth of load requests to the Texas Grid. Texas Grid on a good day, 60 gigawatts, and only 1.5 gigawatts worth those requests were actually approved by the local utilities. So it's a crazy scrambling time. Elon Musk was probably the first person to kind of come up with a method of putting engines on the back of trucks, rolling them onto data centers for power. You know, Boom Supersonic, who's ultimate goal is to be building

Starting point is 00:55:43 supersonic jet planes, has had a slight pivot to selling Super Sonic Jet engines, and apparently they're crushing it. So it's a crazy time and it's a first time in a long time where a new company can come into the market and build a thermal industrial company. So what is the bottleneck for you? Is it figuring out where your system, you know, what data centers your system can fit in? Is it actually just how you have the demand and you're trying to meet it? Like what does your supply chain look like? Like what's standing between you and let's say not this round right now, but you know, the next one. Yeah, it's a great question. So we actually scaled our one kilowatt prototype a hundred times during the back. to 100 kilowatt prototype, we're going to be deploying our one megawatt version very, very soon. The 100 kilowatt prototype that's actually running live at Demo Day right now is going to be running our

Starting point is 00:56:26 first pilot in two months. So really, it's going to be between the 100 kilowatt and the one megawatt prototype, it's going to be figuring out how to elongate it and make it taller. So we've done the really hard work so far, but really it's just building a physical big version of it so that we've talked to hypers, their feedback is they just want a bigger versions. Because they have such a huge demand. They're not really interested in 100 kilowatts. They're interested in 100, 1 megawatts.

Starting point is 00:56:48 Yeah. Okay, so one megawatt will be enough to get your first real, you know, orders, revenues with hyperscaler type customers. Yeah, all the data center campuses would say, I'll take 10,000, please, because I'm building 10 gigawatts. I mean, basically, you chain them together, right? But this is, I mean, we saw in the video, it looks like a shipping container. And essentially, the input output, the output is just cold water, takes in water. So you are agnostic to the design of what happens in the data center, I imagine. and you're not dependent on any particular chip or rack design. But then what energy is actually going into your system? Yes, exactly. It's a great question.

Starting point is 00:57:28 So the energy is used to be the fans and the pumps, so actually circulating the fluids through the system. If I may brag about my co-finder a bit, we don't just design the physical box. We actually, he actually designed the thermodynamics, the mechanical structure, the electronics to power it all, and then also the software to control it. We do this all in-house.

Starting point is 00:57:46 we're big Tesla fanboys. All our competitors are outsourcing every single level of this engineering out, and they're just integrating that and selling that as a product. So we've been able to get a lot of efficiency gains by just designing this all in-house with two people so far. We're fundraising, and we're going to hire some more people, hire our friends in the Bay Area and building the one megawatt one with our own hands. Fantastic.

Starting point is 00:58:03 Amazing. And so there's already a lot of energy on these campuses that you can draw from. So whether it's natural gas powered, powered from the grid, powered from nuclear or solar, or whatever's going on, you're drawing from that, but your goal is to just draw more efficiently and take less of the power so that can probably just go into the chips, right? Exactly, yeah.

Starting point is 00:58:25 You get a lot more inference value out of redirecting that power to your flops. Okay. And then in terms of how you're pitching yourself at Demo Day, are you on the path to LOIs with hyperscalers? Are you working backwards from, are you doing business with neoclouds? Like, what's the walk crawl run to actually create the economic flywheel

Starting point is 00:58:44 to raise money and then go build this thing? Yeah, great questions. We've already started conversations with two hyperscalers. We're actually talking to her third one this week after Demo Day, which is super excited about. Let's go. One of our investors, actually, we closed the check today. They're actually building $20 billion worth of data centers in Texas. So that's been really nice.

Starting point is 00:59:03 But the really first, yeah, fantastic. The first thing for us is really going to be focusing on that pilot, just demonstrating that we can plug into an AI data center at scale, do it reliably, and then scale with our investor that we just partnered with today. Yeah. And then what type of efficiency number actually draws the attention of a potential buyer? Is 1% enough? Is it 10%? Is it 50%? Like what's reasonable? Because I imagine we're 99% more efficient. Everyone's like, yeah, that's not real. But is is one, are we doing basis points here? Like how much of an impact do you need to have to actually move the needle? because some of these projects are so big, their bill for electricity might be, you know, hundreds of millions of dollars. Yeah, exactly.

Starting point is 00:59:49 Like, guys, it's so crazy. My competitors are 100-year-old companies that have been designing the same thing for decades. This is a thing called the mechanical chiller. It's literally the same thing as your refrigerator at home, and it costs $500 million per gigawatt. So people are really desperate to get away from these chillers and look for different types of cooling because that's actually eating up your grid permit. So that's a thing that actually costs 30% of your grid permit. And we can do it for cheaper and we can actually have better.

Starting point is 01:00:12 OPEX as well. So really any percentage is okay. I think that like you can't quote like 0.1% gain, but like, you know, people are thrilled if you have a 10% efficiency gain. And then also we're just in a supply crunch. So even if you just said, hey, we have the exact same thing, but it actually will ship in the next couple months. A lot of people might jump at that. So that's the benefit of being in what were you doing before this? Yeah. So I was actually scaling manufacturing at quantum scape. So I was doing computer vision working directly under the CTO Tim Home. A founder's CTO took it all the way from Inception to IPO. And I also did grad school

Starting point is 01:00:42 of Berkeley studying material science. There's a materials choice to this company we didn't get to talk about today but we're doing something cool. Very cool. Very cool. Well, thank you so much.

Starting point is 01:00:50 After you raised and it was great to meet you. Yeah, fantastic. Thanks guys. We'll talk to you soon. We're going to bring on some YC partners in about four minutes. Stay tuned.

Starting point is 01:00:59 In the meantime, we can go back to the timeline. Back to the news. Talk about specs. This is from Evan Spiegel. Yes, and the Germinator says the snap specs are official. $2,195, all-on-one AR glasses

Starting point is 01:01:14 that Evan Spiegel has dubbed the next computer. $2,200. That's almost Applevision Pro numbers. That's expensive. It's going to be a lift. I feel like it's a lot. It seems like a lot of money. It's got to be very...

Starting point is 01:01:36 Apple Vision Pro starts at 3,4.4. $499. So a bit of a jump. But, but this is a product that seemingly is trying to compete in the realm of like a, it's a mobile device. It's naturally a mobile device. Unlike the Applevision Pro, which obviously some viral images on launch of people walking around with it. I'm just thinking about like, like there are a lot of Apple fanboys who buy every. Get this. What? Get this guy on right now. What? Pull them up. Hey. What do you say? Are you buying the Snap Specks? $2,200 are you spending your paycheck on that? I mean, I think they actually do look really cool.

Starting point is 01:02:14 I don't know, like, I was looking at some of the product demos earlier. Some, like, the game features look really cool. There's, like, ping pong. Yeah. 2.2K is, like, a little pricey personally. But if you can watch Lawrence Arabia from start to finish. Yes, yes, yes. If we bought it, if we traded you your phone for a pair of these for a week,

Starting point is 01:02:36 would you demo them for a week? phone. Can you do, like, texting on it? Can you call? You don't need a phone. You can go without texting for a week. You have Lawrence Arabia. You don't need a phone.

Starting point is 01:02:47 That's true. I can watch Lawrence Arabia like, you know, five hours. During the show. What is the snap specs battery life? Spex battery life. I got to, I got to know. So what happened? Wasn't there some conversation that specs might spin out?

Starting point is 01:03:00 Up to four hours. That's Lawrence Arabia right there with a little intermission. If I remember correctly, there was some like rumors. that specs would be spun out. I remember at least one article. That has not happened. I don't think so. And so the stock is down 7% today.

Starting point is 01:03:21 I don't know. The trick is that you have Apple fanboys who of, you know, they have a lot of Apple fans, but some of them are wealthy. Some of them buy every product. They buy a new phone, a new iPhone every year. They buy the top of the line MacBook. They have the Mac Studio and the Mac Mini. just for fun.

Starting point is 01:03:39 And like, that's a whole class of consumer for Apple. And so when they come in with something that's a little crazy, like a $10,000 gold Apple Watch, like they'll sell a couple of them when they come in with a $3,000 VR headset.

Starting point is 01:03:51 Like a couple people will just be like, yeah, I'll give it a try. When you're talking about a new company entering hardware, meta, I think did a better job coming in with like, yeah, it's just a pair of sunglasses. You need a pair of sunglasses anywhere.

Starting point is 01:04:06 We put a camera in it. We put a camera in it. It's $100 for the version that you know and love, the Rayban Wayfarers that don't have a camera in them. Ours are add a couple hundred extra bucks. It's still something you could give as a gift. And I can afford to lose money on every single device effectively forever. Totally. Yeah.

Starting point is 01:04:23 And so it's a lot easier to get into that like meta ecosystem. Just saying, yeah, throw a camera on my, on my raybans. I'll give it a try. Maybe I'll churn. But if it's collecting dust in the, in the cabinet or in the sock drawer, you're a lot less like, ah, I really got burned. And like the Apple fanboys, you know, a lot of them took them back. But they're like, yeah, that was still like a cool experiment's mix up me.

Starting point is 01:04:48 Because I got to watch. Well, Tyler, once you're back in town. What was that? Lord of the Rings extended cut. I think that's even longer than. Is that over? You can't watch the whole thing. Well, let's do this.

Starting point is 01:04:59 Once you're back in town, let's pick up a pair. Tyler can live with them. We won't say a week. We'll say 48 hours. Exclusive. Exclusive. Do everything. I think you need the phone to do anything.

Starting point is 01:05:11 I think a lot of the compute happens. 48 hours, I think it's actually not very hard. That's just a digital detail. That's like one weekend. You're down? I could do a week. I could do a week. That's like 10 showings in Lawrence.

Starting point is 01:05:20 I'm not asking you to do a week. If you want to do a week, we can do it. Okay, he's doing it. I'm here in four weeks. I'm here in a full month now. Four weeks. All right, it was great to see you, Tyler. We miss you.

Starting point is 01:05:31 Great to see you. All right. I'll see you guys. Talk to him's not the same without you. Yeah, really is John and I were kind of messing around. Kind of quiet in here. And we realize when Tyler's not here, there's nobody super close by to laugh at our bad jokes. Yeah.

Starting point is 01:05:45 It's a bit rough. Spiegel called the specs both very, very wearable and highly capable, which he said give snap an opportunity in a market that's focused on either very large and clunky headsets that are super capable or very lightweight glasses that really don't do much. So it is interesting. I was reading the coverage of in, there's some videos here in upload VR. and it shows sort of the process of this. The HoloLens one from Microsoft 2016 was 580 grams. Snapsbacks are 132 grams,

Starting point is 01:06:19 so they are light. Now, the meta-rayband display is 70 grams, so about half as heavy. So these are heavier, but they're nowhere as heavy as a hollow lens or a real VR headset. And it checks a lot of the boxes on field of view, and we'll see.

Starting point is 01:06:37 We'll see what the developer ecosystem's like. I hope someone comes up with some interesting thing. I hope that there's a YC company that builds on top of this exclusively and has some sort of breakthrough user interface experience, some killer use case that comes from the community. Yeah, got to wait to see how many of these. Yes, that will be very... You're going to need...

Starting point is 01:06:57 I mean, even with a million sold, it's not a very big user base. It's a lot of money. Let me tell you about console. Console builds AI agents that automate 70% of ITHR and finance support, giving employees instant resolution. for access requests and password resets. And without further ado, let's bring in Diana from YC. She's the managing partner.

Starting point is 01:07:14 Diana, welcome to the show. How are you doing? Good. Thanks for having us. So much for taking the time. How is today? Yeah, how's Demo Day going? The energy and vibes are immaculate.

Starting point is 01:07:27 What is it, how does it compare to your YC experience, your entrepreneurial journey? Tell us a little bit about that and then take us through how you're talking to today. today's entrepreneurs about what's changed, what's the same? It's kind of fun that you were just talking about Snap because my company was building augmented reality SDK for game devs back in the era of HoloLens. So my company went through the batch back in summer 17. And back then, it was different. It's way smaller than what it is now.

Starting point is 01:08:04 and AR still taking time to get to the headset. And when we went through it, Demo Day was back in the Computer History Museum, back in the Mounted View. And now we got our campus here in San Francisco and have way more investors and companies are raising a lot of amount that probably was more Series A when I was running my company. What was the mood like around like platform risk building on top? of something else. Obviously, a lot of the augmented reality companies like Niantic built on top of the iPhone, on top of smartphones. And then now we see AR sort of, you know, trying to break in through new devices, specifically spectacles, headsets, wearables. But was that a risk factor that came up in pitching VCs? Or was there sort of like, okay, there's enough of an ecosystem here

Starting point is 01:09:04 there's enough of an enterprise play that you'll be able to be a winner, no matter if there's one platform that breaks out. I think it's a very good question. I think it's the same question right now that a lot of the AI companies are wondering whether they're on the path of the roadmap for the big labs. Yeah. We had the same issue when one fun fact, maybe not so fun, but on the first day of the batch when we went through it,

Starting point is 01:09:34 we basically had built the same platform that Apple ARKids exactly same. It was the same SDK. And on that first day of the batch, we actually freaked out because you have this giant company that have infinite amount of money that could just crush you. And I think what that forced us to do was to really accelerate a lot of our roadmap

Starting point is 01:09:59 because we have thought of building basically this SDK that would enable anyone to build code once and run on any hardware. So we were the first SDK that could enable running it across. You build the code once, and it was ship on Android and iOS, and eventually also got it to headset. Then later on, our company had an exit and got acquired by Niantic, the makers of Pokemon Go, which was the biggest user of AR back then. And then I ended up running a lot of the AR and engineering team

Starting point is 01:10:32 for Nibantic. We had Andrew Lee from Tasklit on first, and I really appreciated how he just straight-up said. He's like, yeah, we are in the path of Anthropic and Open AI, and we're just, we are competing, but just sort of accepting that. And workflow, automation, what they're doing, such a big category, can obviously, with great execution, build a big company, even despite of that.

Starting point is 01:10:59 What are some of the key themes for the batch? How does this batch compare to the last batch? What kind of trends are you seeing overall? I think this batch is following a similar trend to the winter 26 batch, where things that have been happening, a lot of these companies, are the first one that are basically riding the wave with agent decoding working. It has become the norm where company, well, the norm. It's become more common that companies can go to zero to seven figures in revenue within just three months,

Starting point is 01:11:36 which when I did the batch, that would have been impossible. I don't think any companies did that. The best companies back then would grow, let's say, 10% week on week, which is what, let's say, Airbnb did back then. And that would get them to a couple hundred thousand in ARR, which is a very good, very, very, very impressive accomplishment back then. But now, companies get to seven figures because the companies are able to build a more mature product a lot quicker with a gigantic coding. If they find the right user and customers on sales and nail and that, people can really speed through this right now. How often do you find yourself talking to a company in office hours that you think shouldn't be optimizing for that?

Starting point is 01:12:25 because there's a lot of companies that we've talked to today, like Madrone, for example, like very important sort of product category, clearly massive TAM, and they should go as fast as possible, but if they just optimize for revenue traction right away, it would take them on the wrong path, I imagine. Oh, absolutely.

Starting point is 01:12:47 So this is where we give very custom advice depending on the type of company. If you're building, let's say, core infrastructure or defense type of applications, it's more important that you deploy deeply into few customers. And it doesn't matter if the revenue is small, but it has a lot of opportunity for growth. And it really depends on the journey. So this is not advice for all the companies, for some of them. That would sense.

Starting point is 01:13:19 YC has gone through waves of like consumer boom in the early days with Airbnb and a lot of consumer companies and then sort of swung to business to business products and companies. And there's been demo days that we've been to where it's a lot of stuff that's deeper in the token path, which is a great place to be enabling agentic workflows deeper in the stack. There's a lot of hard tech this time. But I'm wondering how you're viewing the opportunity in consumer, in gaming. We're just seeing so many vibe-coded games that are remarkable demos. And it feels like with a YC team on an actual deadline to get to real functionality, real adoption, revenue traction. The tools are there now where previously to make something that

Starting point is 01:14:13 looked anything like a breakout game, you'd see these companies and you'd, and you'd hear, oh, well, you know, Epic Games was grinding in the dark for five years building their engine before they released a game or, you know, so many companies, Roblox, it took a long time. Now it feels like you can move faster. And that might make it more, the end state might be that all of this turns into, you know, there's no fundraising at all and it's all just indie devs. But it does feel like there's an opportunity here, but, How are you seeing consumer, either in gaming or just broadly,

Starting point is 01:14:47 develop over the next couple of years at YC? I think it's still very early. We haven't seen these companies break out yet. I mean, this whole era of AI right now, which is one of the most fun areas in tech right now, there's still so many patterns that we're figuring out. And with regard to consumers, I think there's going to be a lot of big category winners

Starting point is 01:15:10 that we don't know what the shape of them could be. It's sort of like trying to predict the Airbnb or DoorDash back in the 2010s would be very hard. And with regards to gaming, I haven't worked in gaming myself and know a lot of these AAA studios that you mentioned that take a long time to get a game out there because they build the engine and there's just so much technical complexity. A lot of that, yes, the barrier of that is entry is lower. but I think the thing that's still in short supply is the ability to encode fun. The thing why you could build lots of games today,

Starting point is 01:15:51 which I've tried and vibe-coded some games, but it's hard to build actually game loops that people keep coming back and have fun. And those are new patterns that we haven't discovered. I mean, you could imagine now in there where a lot of games could be on an agenic loop and based on a lot of the metrics of the usage, they could self-improve.

Starting point is 01:16:13 And it could be more fun. But I don't think people figure out how to build those. Because you have to be able to measure fun, which I'm not sure if there's a easy way to do that. It really depends game for a game. Yeah, yeah. That makes sense. Anyway, thank you so much for taking the time to come chat with us.

Starting point is 01:16:31 Have a great rest of the time. Great to meet you and great work this batch. There's a lot of fun companies. Yeah, we'll talk to you soon. All right, guys. Have a good one. Cheers. Bye.

Starting point is 01:16:39 Um, womanitis is getting himself in trouble for again debating the world's population. He says population is max 500 million global earth is max 15,000 years old. People are asking if it's bait. He's going back and forth with a bunch of people. He's pointing out that there are some economic incentives to certain countries to misreport population numbers if they're getting if they're getting economic incentives or distributions based on that. He's having a fun on the timeline. You can go dig into his post and make a call for yourself.

Starting point is 01:17:18 Anyway, over in China, one of the most populous countries in the world, allegedly, the final boss of ADHD has been spotted. The guy is simultaneously watching TikTok, chatting in a messenger, and playing a game. This is what a foldable smartphone is for. There's pushback against smartphones now. Imagine what foldable smartphones will do. Is it going to be even worse? Who knows?

Starting point is 01:17:44 This is an absolutely crazy setup to be rocking. And is this a single player game that he pauses or? Yeah, it kind of seems like he's larping. Yeah, this seems performative, right? This is a larp. This is a larp. I don't think this is real. I got in trouble for my game knowledge.

Starting point is 01:18:06 apparently used min-maxing improperly, although there was some debate over whether or not min-maxing has transformed into the definition that I used, whether I was using a more modern interpretation of that term. But anyway, you can go and dig into Satrini's post and make your own justification on where you sit on token maxing, token minning, token min-maxing. The clear point is that you've got to focus on ROI, like any other business process. Andrew Gao sharing some data. into our cognition. He says, this isn't public, but Stanford CS degrees

Starting point is 01:18:39 dropped 42% year-over-year. Berkeley down 61%. All of the propaganda, like, you don't need to learn to code anymore. It seems to be working. I mean, it's got to dramatically change the experience and, like, the fun of being, like, part of the fun of taking a computer science course

Starting point is 01:18:58 is doing an exercise and working through the code and, like, solving the puzzle that is the code challenge. And when you can just one shot that with any frontier model, it gets a lot less fun. And so you have to imagine that people moved over to other more amorphous, more rewarding disciplines. But it does seem silly because I'm sure that if you go really deep in a CS degree and your AI enabled as well, you can both find a job and also have a really, really interesting and rewarding experience. but certainly a reaction to the fact that the existing path of just like, someone tells me exactly what to implement,

Starting point is 01:19:43 and then I go and implement that, and I make six figures forever, is less and less clear. And so people are reacting to that by going to other places. I want to see the data on where these students went. What are they studying? If not, CS, did they bleed into everything else? Because sometimes you read these and it's like,

Starting point is 01:20:01 oh, they're not studying computer science anymore. they're studying computer engineering now. And it's like, yeah, that's kind of the same thing. Like, you saw that, like, computer programming declined, but computer science jobs increased or something like that. Yeah, they're studying. Computer engineering. They're studying AIML.

Starting point is 01:20:17 Yeah, yeah. Yeah, I mean, a lot of the AI track has always been symbolic systems anyway, not CS, at least at Stanford. That's what a lot of the early Facebook employees, a lot of those folks studied. There's a lot of other folks that study physics or math or all sorts of things. And it's got to be more fun to study a really hard topic like physics or math because you have an AI tutor and you can use AI to close that gap, to bridge that gap.

Starting point is 01:20:45 But we should ask Harsh Tagar from YC. He's ready to chat with us. We can ask him about what you should major in if you want to build the next grade company. What do you think, Harsh? Did you study computer science? I didn't actually as an undergrad. I started in law school and then I dropped it. out and then I told myself to program. A lawyer being coming successful. Anything is possible.

Starting point is 01:21:08 Yeah, it's true. It's a bad stop, but it's remarkable. Yeah, no. No, obviously, it's a fantastic background. It speaks to the diversity of backgrounds that lead people to success in Silicon Valley. What do you make of this decline in CS degrees? What are you seeing in the YC founders that join the latest batches? Is there, is there, is there, are you experiencing? more diversity of backgrounds. It feels like YC's always been pulling from all over the place. The best case and point is Airbnb, it's a bunch of design students. And that was not a feeder school for, it was not a unicorn factory, RISD, and yet we got Airbnb out of it. So what's going on at YC in terms of educational background these days? It's good. You know, it's actually this weekend

Starting point is 01:21:53 was with a group of college students that we were meeting and hopefully they'll do YC one day. And this was definitely something they said is that they are seeing their smart friends. enroll in computer science less. People want to do more physics and math and first principle stuff. So that definitely, I think that's a leading indicator. So I would definitely expect over the next few years we might see more of the smart founders come in with like a math or some sort of like thinking stemmy background that's not pure CS and then like I use the coding models to actually write the code. Yeah, yeah. It doesn't, there's one read on this where it's like, oh, no one's going into tech. And I think it's like, no, people are going to be like even sharper about using these

Starting point is 01:22:36 tools and even more, you know, ground truth in what matters, in what's not changing in the math and physics stuff. Anyway, other trends that you're seeing. Yeah, yeah, I wanted to get into like the psyche of this batch. If you zoom out, so many companies are having like this sort of like often, you know, if their later stage company midlife crisis, if they're a younger startup that's kind of in the path of the lab. They're wondering. It feels like even the last, you know, few weeks, like the narrative is kind of flipped and we're seeing, like, a lot of the application layer companies say, like, hey, we have an opportunity to sort of orchestrate a variety of models and maybe get efficiency for companies that were token maxing. So what is the psyche of the current

Starting point is 01:23:19 batch? Where are they finding opportunity? I'm sure, you know, you had a few office hours where founders were saying, like, is it over? Should I just, you know, pack it up and join a lab? And of course, the YC way would be to say, like, no, there's always going to be opportunity for, you know, smart, focused people. But take us through it. Yeah, I mean, it definitely ebbs and flows. I would say when Claude code first launched earlier this year, I feel like it really did flow towards the, oh, wow, like, the labs are going to take way more of the value and, like, are they just going to own everything? And I think that was a little bit in the in the psyche. I feel like now it's flowed back towards, you know, you don't, you want to try and

Starting point is 01:24:03 have a little bit of thought around, are you going to be in the path of labs or not, but mostly I just think the founders here are doing their thing. And they're not, they're not currently, like, worried about that as much so much, so much more. And I think, like, largely they're trying to figure out what are kind of maybe the boring but valuable things that the, um, the labs are just not going to focus in on. And I think there you see a lot of, it's, it's funny, like we used to talk about like making stuff that people want. Now I feel like it's a lot of making things that agents want. And so especially at like the infrastructure level, it's like how you set up sandboxes for agents that go into the nitty-gritty detail of giving them all the tooling

Starting point is 01:24:39 that they need to use to actually perform tasks and actually to do things. And it's the kind of stuff that the labs themselves are just unlikely to go into depth on. Yeah, it's funny. That's somewhat of, it's like somewhat a new framing. Think about what the big companies aren't going to do is to off the beaten path for them. But if you go back, I'm like, well, the hotel industry wasn't thinking about Airbnb and the big banks weren't thinking about Bitcoin and you get Coinbase. And so there's something there that's like always been true and is just sort of resurfacing and is actually maybe a ground truth that you need to pull back to constantly.

Starting point is 01:25:15 But it is, it is potentially more relevant than ever. Has the, has the Gavin Baker meme of being in the token path taken hold this back? this idea that you want to be deeper in the bottleneck that if you're if you're not in the downstream path of the labs the safest place might be upstream a deeper in the supply chain we talked as a founder who's working on cooling for data centers that's clearly something that's very aligned with the broad trend but what's happening there in terms of token path thought I don't I don't at least I don't think that has trickled down to like the founders starting these companies and figuring out

Starting point is 01:25:54 what ideas to work on. I do think some of that framework fits. Like you're certainly seeing things further into like the physical world, like actually, you know, building, um, building physical things. But I think that's actually more to do with being inspired by Elon and, and Parma Lucky at Andrewill on the founder side that people want to work on like physical, like work in atoms. So we've got a team in this batch building like a nuclear reactor like that, that I think is just inspired by, hey, like we want to build real things. And then I think on the other side of the marketplace investors typically really didn't want to invest in hardware companies. So we used to see this.

Starting point is 01:26:29 I always used to brace the hardware or the hard tech companies. Hey, you're going to have to be ready for it because no one's going to want to invest in you. You're going to have to grind it out. It might be years before you raise the money that you need. And now, I think, again, since Claude Code, there's a real shift of, oh, like, SaaS is dead. Maybe software is not going to be as valuable. Suddenly investors all want to invest in hard tech stuff, which is, you know, I'm happy about it, but it does also just show like how fickle that that side of the marketplace.

Starting point is 01:26:54 can be as well. Have you seen any change to average burn rates with, uh, or any, anyone in the batch token maxing so hard that you have to say like, hey, you should probably, you know, I expect you to raise, you know, after demo day, but, you know, maybe, maybe don't. No, not yet. You know, I'm a bit reluctant to say this. I don't want it to change, but I don't think I have that much control. Like, I actually think on the ground, the founders feel like the, um, the subscription plans from the big lab company. are actually very generous and so they're able to actually get as much token usage as they want largely out of those plans is sort of being what I've seen. We do internally talk about this, like we do wonder if there's some

Starting point is 01:27:39 training we need to do to like push the founders to spend more on tokens and that maybe they're thinking about tokens are something you should conserve spend on whereas like I think I and these a bunch of the other YC partners really think about it. It's a lot more like rent. Like you don't say I'm not moving to San Francisco to build my tech startup because the rent's expensive. Like, yes, it's expensive, but it's the place to be. Like you don't conserve on that. You like want to be, you actually want to spend money on that. There's other things you save money on.

Starting point is 01:28:07 And so I do feel like because so many of the founders, especially if they're younger, they just don't have the budgets to use the latest models and be like cost insensitive and token max. So they're just used to conserving and trying to do more with less. And I think that may well change, but at least currently, I don't feel like the burn rate is going up because people try to get to product market fit are spending way more on tokens. Yeah, it's interesting. It's like, you know, maybe you can delay hiring the incremental engineer, which actually can be very good because it's way easier to pivot if you have just the founders, right? Yeah. Versus like, you know, once you start adding. What about trends on the business model side? You know, last year I feel like the, you know, classic VC blog post would be like sell the work, not, you know, the product. We've seen it

Starting point is 01:28:59 some companies get to the point. There's like, you know, a legal startup that's like actually just doing the, doing the contracts. They basically, they sell executed contracts effectively. But it hasn't maybe been as popular as I think a lot of people had imagined, at least so far. But what are you seeing? I don't think the business models are, fully switched over to, like, sort of paperwork yet. It's certainly pay per usage, but I don't think it's pay per work yet. And I think that's just because the agents as yet haven't quite been good enough to actually consistently deliver, like, the final contract in a way that makes the startups feel good about charging that way. I do think we're right on the cusp of that changing,

Starting point is 01:29:46 though. Like, this batch really does feel like one where it's, it's not just sort of agents to review things or human in the loop or getting stuff teed up, it's really starting to be more end-to-end and you're getting to the point where the agents are just doing all of the work. And at that point, yeah, you may well switch over. But I feel like maybe that's going to be a batch or two away. Are we going to see a revenge of the network effect? It feels like in a world where, you know, software is mutable and maybe like the million lines of code that you and your team built or not as defensible as you thought. The original SaaSpocalypse thesis,

Starting point is 01:30:25 network effects businesses should do really well. I don't know if more founders are talking about that, but what's your interpretation? Because it is sort of the hardest moat to construct, happens sort of randomly sometimes. But if you can get it done, it feels like you'll be in a very strong position. You know, you mentioned like the late stage companies

Starting point is 01:30:46 having maybe midlife crises. That's where I hear network effects talked about more, honestly. it's like some of those companies, if you're whatnot or Kauci at the moment, like, yeah, or Coinbase, like you're feeling pretty good about the fact that you are in marketplace business with real network

Starting point is 01:31:03 effects and you're not worried about sort of someone vibe coding a competitor to you. So I think that's where I hear it. I will say like, yeah, again, on the ground floor here on the batches, I don't, we haven't seen as many founders like build pure marketplace businesses as

Starting point is 01:31:19 you would expect based on that thesis. And I think it's just because, obviously, people are really excited to work with AI and build AI tools. They're not really thinking about it as, hey, like, what's got like a network effect moat and how can I build that business? Yeah. What about gaming? We talked to Diana a little bit about this, but it feels like we, you were seeing incredible demos of vibe coded games. When I see them, I'm like, I would love for a team of YC founders to work on that for three months to get it actually to a place where it's a real game instead of just a demo. But the underwriting for the game industry has always been different than the traditional SaaS venture capital flow.

Starting point is 01:32:00 Is there a world where YC is more open to just pure games companies? We're just going to make a hit game. We're going to scale it into a AAA game studio. We're going to develop IP. That's a very different motion. But are you open to that? Is there momentum there? Or is that something that you think is farther off?

Starting point is 01:32:19 Yeah, we would totally be open to it. Diana's definitely got the gaming background and the gaming expertise there. But I think we would be totally open to it and just like the reason you typically, as a startup investor, I don't want to invest in games is it's so hits driven. Even if you get one hit, how do you know you get another one? What's replicable? But yeah, like, you know, these days, like if you can build some recipe for actually consistent in producing games.

Starting point is 01:32:45 And I feel like Zinger got the closest to that. that arguably that was more about, like, they sort of productize the distribution mechanism and the viral mechanics more than like the gameplay. But I know that seems plausible in AI. It seems like if you can actually build an AI or an agent that can consistently produce high-quality games, and also you just be able to run so many of those experiments in parallel, so many more experiments in parallel than you could before. I think that would be really interesting.

Starting point is 01:33:11 Again, with us, it's the same thing. We need to see founders who seem like they've got some insight at that level. Like they're thinking about it beyond just, hey, I really love playing third-person shooters, and I want to build like a really cool third-person shooter game. We'd probably have to hear that in the interview. Like, hey, like, I'm thinking how this would, like, generalize and build out. And here's, like, where the technology could be interesting. I'm wondering if, I mean, I'm wondering if there's a lesson from, I mean,

Starting point is 01:33:34 Hollywood is probably behind YC in terms of the forward-lookingness. But there's this interesting case study with backrooms and obsession, these YouTubers who went and made, like, classic, Hollywood films, low budgets, but real budgets. And they put up huge numbers and they're, and their massive successes in Hollywood now, driving a lot of attention. And,

Starting point is 01:33:56 and it's interesting to think about like, what is the next category that gets unlocked, that's more of like indie dev world, but now has the ability to scale through like the YC venture capital model. Yeah, I think, yeah, I think that's going to be really interesting. I think that's where you might end up seeing, I don't know if we'll see like the one person, but you might see the one person unicorns, but you certainly feels like you're going to start seeing the one person, like,

Starting point is 01:34:22 $100 million revenue companies in like a bunch of articles because you've just got like agents doing the core business functions. Yeah. Okay. So have you ever had an office hours where where a founder comes in? They're like, all right, like, I think I'm, I think I'm that company. I think I'm the one person, $100 million company. Because like everyone's been talking about this forever, but, like, there's such a much stronger incentive to just, like, hire the incremental person if it can make you, like, I just don't know that it will ever, I just like, maybe, maybe it, it, it just want, but yeah, there's such a strong incentive, like, do I want, do I want to do I want to be the company that I got to have a friend around? Yeah, like, do you want to, do you want to,

Starting point is 01:35:03 do you want to be, like, company that gets an award that doesn't matter, or do you want to win? Yeah, good point. Or not even a real award. It's just like, it's, it's something that you can say on the internet and then people will debate it and be like, well, you hire this freelancer, like, you know, you have a lawyer. Does that really count? So, like, I feel like it will never actually happen because it will always be debated. Sure, sure. I personally have not worked with a founder who's, like, come in with the dead set, like, I never want to hire anyone. I kind of want to only work for myself. I mean, my guess is someone like that is not going to gravitate towards, like, a community product, which is essentially what YC is. And so it could happen. Who knows? I, I, I, the

Starting point is 01:35:41 the thing that I'm seeing of 100%, you can get to, like, Series A, you can get to a million or two million dollars of revenue without hiring anyone and being way leaner than you've ever seen before. But after you've raised the A or the B, it does seem like hiring is essentially the same as it's always been. Like, who knows whether that's because, like, there's just more work to do or whether it's like, like, you know, once then you have, like, your investors and they're, like, running the same playbook. But I definitely am not seeing any evidence yet of, like, those companies really slowing down on hiring. Can you give us the lore on the Paul Buckeye $100 billion question?

Starting point is 01:36:22 That's a funny. Yeah, for sure. I mean, what's the law? Like, you know, we did this thing at YC called Group Officers where it's just sort of a couple of the partners and a group of the startups. And it's sort of like a, I know, it's sort of like a group discussion. Hey, like what's going on? What are your problems?

Starting point is 01:36:37 And sometimes we'll prompt with, hey, like, let's think about. about this. Again, everyone could have, like, chime in. And Paul always had this great prompt, which was, don't just think about how you're going to be, like, a billion-dollar company, where you get to, like, $100 million in revenue. Think about, like, what's the next act from that? Like, how do you get to $100 billion? You know, like Google started off with search, and then became so much more. Metas the same. And so it's sort of like a mind-expanding exercise, and it's useful to do at this stage, because you get a sense for how much ambition and vision do the founders actually have. And sometimes it helps them get more excited about

Starting point is 01:37:12 their own company when they realize, oh, wow. Like this is like, you know, I was thinking about it as like, like, how do we like build this one product that gets us to like, you know, 100 million revenue. But actually, I didn't even realize there's six other things we could do. I just have to like get there. Yeah. Is it too small? Do you need a 10 exit? We got trillion dollar companies now. It's true. It's inflation. Even Elon. Small time. I see. Just wants to sit around just knock and base hits to 100 billion. Are you swinging with the fences or not, harsh? Maybe I'll bring it back and call it the trillion dollar.

Starting point is 01:37:45 It has a nice ring to it. It's my thing. You should, you should. Sorry, P.B. Jordy Lowe's last thing. Elon was saying on Sunday, he would be surprised if they don't have a trillion of revenue by 2031. I agree.

Starting point is 01:37:59 I agree. What's the path? The market space time, so he can't beat that. Yeah. That's right. Anyway, we're going to hop on with Gary. Thank you so much for. hopping on the show.

Starting point is 01:38:08 Great to see you. Congrats on demo day. We'll talk to you soon. Cheers. We'll see you guys. Have a good win. Let me tell you about public.com investing for those who take it seriously. They got stocks, options, bonds, crypto, treasuries and more with great customer service.

Starting point is 01:38:22 Absolutely. We got to hit Gary with the flashbang. Right away. He comes in. Gary, how are you doing? Welcome to the show. He's now here. It's demo day.

Starting point is 01:38:34 Watch out. It's flashbang. Great to see you. Great to see you. How are you doing? Take us through it. How's the day being? Oh, it's awesome.

Starting point is 01:38:47 I mean, it's always the big show here in San Francisco. I mean, it's a beautiful, gorgeous day. And, you know, who doesn't like to go check out a nuclear reactor in our back parking lot? No uranium, don't worry.

Starting point is 01:38:58 Whoa. This is like the second or third nuclear company, though, right? I mean, I think we kind of need all of them, right? Let's do all of them. What's stopping them from going critical in the parking lot? Oh, well, luckily no uranium. No uranium. No gyver counter needed.

Starting point is 01:39:14 That's great. That's great. How much is hard tech booming? Do you have any stats? It feels like every year there's a slight, slightly more people taking the hardware isn't that hard. They say hardware is hard, but I'm going to try it. And it feels like that's been a trend that's been growing steadily.

Starting point is 01:39:33 Am I correct to read that? to the trends over demo days over the last couple of years? Yeah, man. I got to call out my buddy, Brett Gibson, over at Initialized Capital. I mean, right when Code Gen first started happening, Brett was my first friend

Starting point is 01:39:48 who said, you know, obviously, you got to fund the bangers, and it's not like they stopped doing B2B SaaS. Yeah. But, you know, initialized was one of the first people to do the pivot to say, you know what? Like hard tech is the way it's going to go. And, you know, things like flock safety

Starting point is 01:40:04 were sort of the first things we did that really, really became huge. And, you know, computer vision is in. And then now with AI, like everything is open. It's so much easier. I mean, you can do mechanical engineering. You could do electrical engineering. If you don't understand it, you can build it. You can find every vendor.

Starting point is 01:40:24 Like the supply chain management can happen right there. All in an open claw now. I mean, we're just in a new age. Yeah, I mean, we've even seen people like, like, optimizing their home Wi-Fi speed. Oh, yeah. You know, different models.

Starting point is 01:40:39 It's crazy. A lot of the hard tech boom, I feel like, has been driven by Elon Palmer Lucky, just sort of breaking the glass on like, it's possible for this to happen. And that's led to a lot of like defense tech startups.

Starting point is 01:40:55 But I'm interested in the intersection. What are you seeing and how much conversation is there around hardware and AI, data centers, stuff that's in the token path, but not in the direct path, not one-shotable by a frontier model.

Starting point is 01:41:13 Oh, my God. I mean, honestly, you should probably get Ash and Kutcher's here today. You should go talk to him. He's got a whole thesis around AI, solar. He's actually pretty bearish on fusion, which is interesting.

Starting point is 01:41:26 You should ask him about that. Okay. I think that's news. That's going to hit. Ashton Kutcher, You're bearish on fusion. And we love Ashton. He is like really, he went all the way.

Starting point is 01:41:38 Like he met everyone. So, I mean, Ashton has been an incredible investor, especially for hard tech. But anyway, I mean, I think what we've been seeing, there's like more, there's more defense here in this batch than the entire last year. Yeah. And there's also more geopolitical conflict. So there's more energy to purchase and the government's sort of changed a lot of the stuff is procured.

Starting point is 01:42:04 And so the old adage, you need to be a billionaire to start a defense tech company. The Elons, the Palmer, the CARPS, software singularity, and then hardware. Who?

Starting point is 01:42:14 Who? It's Emil Michael. And it's our friends who are actually in the Trump administration who are like changing everything. Like the DIY, like it's a new department of war. Like this is never,

Starting point is 01:42:28 you know, I think that there is actually an awareness that cost plus, You know, Cost Plus was originally designed to be something that created more innovation. And, you know, here we are 50, 80 years later. And, you know, what was needed for 50 or 80 years ago, it's the wrong thing now. You know, you have the defense primes that are just not actually innovating. They can't.

Starting point is 01:42:51 Like, it's sort of structural. There's sort of these, you know, it's the equivalent of big tech bureaucracy. Like, I think, you know, I think of iPhone how stupid Siri is and how terrible it is. or even like Alexa over at Amazon. It's like unbelievable. How incompetent, you know, I mean, these are very competent people put into a system that is structurally unable to deliver technology

Starting point is 01:43:11 that is actually what people really, really want. And so when you talk about consumer technology, like, you know, it's not that big a deal. Like, you know, my life is 20% worse, 10% worse because I can't use a Siri that actually uses large language models properly. You know, years after it was designed. And I usually, you know,

Starting point is 01:43:30 when I go to DC, I like to tell politicians that. It's like, hey, like, here's this thing that doesn't make sense in society. But, you know, it's consumer tech. It's no one's going to really, like, probably very few people have died because of it. But when I really, what makes me mad, though, is when you think about, like, there are service members who are going to die in service of our country to defend liberty and they're giving their lives because defense primes are not doing the job that they're supposed to be doing, right? They're not innovating.

Starting point is 01:44:00 They're not bringing these new technologies to four, and it's structural. And it's not like individual people trying to do it. It's structural, right? Like, just as the people who work on Siri, like, actually, we funded a number of them who have come and done YC. Delightful, wonderful people. Like, the people who work at these companies are incredible. Yeah. I never want to attack the engineers working on this stuff.

Starting point is 01:44:20 It's actually structure. Yeah. Like, you have big tech and you have, you know, big anything, right? It becomes structural where here's a thing that should have. It does not happen why. You know, I think cost plus is one of those things. Like it's just systematically, instead of trying to create something that's better, it's like how do we maximize shareholder value?

Starting point is 01:44:40 And, you know, I get it, but also why are we doing it this way? And so, you know, credit to the Department of War. Like, they are actually, for the first time in, like, decades, easily. Like, they're actually open to a team of 10 or 20 people, whether it's, you know, in Boston, in Austin, even down the street, in San Francisco in gundo. Shout out to our Gundow bros out there. Yeah, we talked to Dan Durskow about this,

Starting point is 01:45:06 about just like having range days where any startup can come and pitch and they'll just buy the best product right there. It feels like it's an entirely new day in terms of procurement. Switching gears to software is like I could prompt that with one prompt, I could build that, or it's just directly in the path of the labs,

Starting point is 01:45:27 it's so easy to build. Is that the new, like, Like, what if Google builds it? Like, I have a suspicion that we might have over rotated on the, on the, oh, like, it's easy to build that particular product or vibe code that particular product. And it's like, yeah, but you didn't. And then this team did. And they went and did all the other hard things to get customers and understand the user and actually build a great product. And so, yes, you could clone it, but they have momentum.

Starting point is 01:45:57 And so maybe they won't be a trillion dollar business, but there might be some beautiful businesses that are accelerated and they look like SaaS and they look like they could be cloned. But there's something secret there that's still going on. How do you think about that? Do you push back or do you agree? No, I think that, you know what, like being in San Francisco and being friends with people at the labs and being able to see like the inside, like, you know, what codex and cloud code is. being able to see the inside, like, you know, being able to see the inside of, like, you know, what codex and cloud code is going to be incredible, right? And then these are all 200 IQ people building 200 IQ models.

Starting point is 01:46:35 And, you know, it's going to keep going. I mean, it's going to be, you know, true AGI. The takeoff is happening. Sure. And the thing you have to realize is, like, everywhere else, like, well, those 200 IQ people can't really interface. way that they need to with the 150 IQ people and the 130 IQ people and the 100 IQ people and like the massed humanity room temp you need a room temp guy which is fine no you talk to a lot of entrepreneurs and it's like yeah this person just worked really hard and found an interesting niche

Starting point is 01:47:05 and just deliver down to my level room temp but like we're joking but on the you know on the other hand like the world actually cannot interface with these things like the models will be complex The UI will be too complex. Like, the more complicated you make any given solution, the less likely someone's actually going to be able to take advantage of it and implement it and use it. Yeah. And so, you know, that's where all of the value is in business.

Starting point is 01:47:37 It's like, how do you take this new tech that nobody has? Very few people understand. Yeah. And then bring it to them in a way that actually helps their lives. And, you know, I mean, obviously on a five or ten-year basis, with super intelligence becoming what it is, you know, all bets could be off on like a long enough time frame. But certainly for the next like 18, 24, 36 months, that's not happening. And for now, hey, man, we're going to fight for the little guy and for, you know, the person who has nothing

Starting point is 01:48:09 sitting in front of a recycled MacBook Pro to be able to make anything to make prosperity happen. Like that's what that's what this logo is, right? I love it. Yeah. That's what, that's what we're about. about like this is, as long as this is a flag that flies, like, I will, to my dying day will fight for the startups and for startup founders and for people to be able to create something from nothing. And, you know, I think it's going to be a while before we really have to worry about, like, are there going to be jobs and things like that. I completely agree. What,

Starting point is 01:48:41 uh, out of the current like request for startups, is there any category that comes to mind, Are you working on a new request for startups list? Yeah, absolutely. I mean, we're always doing that. I don't know. I'm trying to convince Ashton to do the call for solar and re-industrialize America with me, so maybe you'll see that.

Starting point is 01:49:04 Joe Gebia, Chief Design Officer in the United States, he sent me the one about agriculture. You know, I think that we're realizing that, like, YC has a real convening power, and people really shouldn't have to just go and work on, you know, nothing wrong with B2B SaaS and we love it, but also like, you know, if you've got something in you that isn't that, like we can actually try to harness like all that brain power

Starting point is 01:49:31 and all that AI power into basically everything. I mean, that's my, the number one philosophy that I think people have not absorbed yet enough and that I want everyone to hear and I will shout from every hill is that it's time to boil the oceans, guys. Like everyone else is out there. there trying to figure out like, oh, let me like lay off all these people and, ha ha, like, the machines are better than humans. I was like, you know, we live in a broken completely just, it's like a backward world, right? Like, if we could accelerate technology fast enough,

Starting point is 01:50:04 it would solve a lot more problems. People would have a lot more prosperity. And now, like, the biggest, one of the biggest things, the blocker is not even money. It hasn't been money for some time. It's definitely not right now. Like, it's just pure agency. Like, like, do you have an idea? Like, can you prompt it? Like, can you actually, you actually know how to use this stuff? And what do you want? Like, what do you want to manifest in the world? And if we do that, and we do that, like, 10 or 100 times or a thousand times more than we have been, and we have, you know, and then going back to what we were saying earlier, like, it's actually about the systems, right? A startup is the defining system where either you do it or you don't. Like, either you have

Starting point is 01:50:41 a customer that loves it or a customer that, like, doesn't buy it. And, like, most of the time, frankly, like people love to bag on startups. It's because those startups don't make something people want, right? That's why our t-shirt is literally make something people want. It's not more complicated than that. We actually give away the secrets on our freaking t-shirts. It's great t-shirt that you get on your first day at YC. But like, it's surprising how hard it is to do that. And then, but if we did that, you know, instead of trying to replace jobs, instead of trying to say like, you know, ha-ha, I could like lower my cost by 90%. Like the thing is, that's why I love markets. Like, if you have, you know, just to take, take like a big tech swing again, when you have a

Starting point is 01:51:22 moat that is so unassailable, that is basically like regulatory capture, like, you basically, if the government says, like, actually, Apple, you can just not have multiple app stores, then, you know what, the government just granted? The government is okay with the monopoly. And when you have a monopoly, you have no choice. And then that's what we have. Like, we do not have choice about, uh, we do not have choice about what messenger we use. Like, why is I message so stupid? If you had LLMs embedded in that, you would have, you know, we could actually literally save time by having that level inside our phone. But I guarantee you- Personally, I like when I'm searching for like a very specific message that I know I sent and I can't find it. Personally, I just- Where is the bar for big tech now?

Starting point is 01:52:08 Where's the bar for big tech? Because SpaceX is now a $3 trillion company. If you're just hanging, it out of just one trillion, are you little tech now? Are you a hundred billion dollar little tech company? Like, Toddler tech? Where, where is the line? Honestly, it's like after people are like 10 billion dollars valuation, I'm like, God bless you. Okay. Please don't, please don't, you know, mess it up for the next, yeah, please don't steam roll in the next generation, you know. Decker corn is the graduation. That's just my experience. Yeah, no, that makes sense. You become a Deckercorn, you start hiring these Google PMs who come in and, you know, the same people who closed off the platforms on app stores and things like that, you start hiring those PMs and then it's

Starting point is 01:52:46 over guys, you know, which is fun. Like as long as, the thing is like, as long as there are markets, as long as there's choice. Yeah, of course. There's going to be a check on that, right? And so that's, that's my fractal. I'm like, hey, man, let's boil the oceans. Let's make so many more people. I mean, it's, you want to get political? I don't know. I have a thing about that. Not particularly. Okay. No politics today. No politics today. No politics today. I would rather talk about gaming. You're running.

Starting point is 01:53:17 That's it. You're running for president. I'm Canadian. Sorry, guys. Oh, interesting. I didn't know that for some reason. I was born in Winnipeg. I'm naturalized though.

Starting point is 01:53:29 God bless America. What I actually want to know about is G-Stack. I want to know about G-Stack in the context of OIC. I feel like YC, the whole goal, the whole system is built to allow the founders to talk to their customers, to build something people want. And that's why there's bookface with investor reviews, because you could see a founder getting distracted and being like, I need to optimize my fundraise. And so I should go build a database and interview everyone about what funds are the best.

Starting point is 01:53:59 They don't need to because it's been done for every YC company that's been all gone through the batch. And is there a lens where I can view some of the research, some of the work, some of the experiments that you're doing around vibe coding, around G-Stom? as let me sort of like do the unhobbling, but fall on the sort of like optimizing systems and experimenting so that that burden doesn't go onto every single YC founder that's token maxing in the batch.

Starting point is 01:54:26 Yeah, absolutely. Okay, so GSTack was sort of my project from like January through March. I'm still maintaining it and fixing it. But like my new thing is GBrain, which I actually wanted to be like the Postgres for agents. So the thing I realized is that like, you know, when an agent, you know, a human can only keep like seven plus or minus three things in their head, right? Like a phone number is about it, right?

Starting point is 01:54:49 That's generous. Yeah, that's generous. I can keep hundreds of things, but yes, I understand for most people. That's very nice. Some of us are special. John, you're very special. And, but a computer with an LLM, it can keep about three Harry Potter books in their head. And so, you know, that's what context is.

Starting point is 01:55:06 I mean, that's like a stack of books. No, no, no. I understand. It's just way better when you say, like, it can contain, like, 75 airdish problems or, like, some complex. But, Harry Potter. But, yes, I understand. The context window is significant and especially when compared to a human. Yeah.

Starting point is 01:55:24 And then when you think about, like, what most computer systems are, like, actually, you should think of, like, the Library of Alexandria, right? Like, thousands of books, tens of, maybe millions of books. Like, actually, it's even bigger than that. It's, like, the whole internet, right? And then that's what, like, sort of, like, Carpathie's known. knowledge wikis are, people are talking about company brains. Like, you can basically take all of the relevant info about customers and everything that you've ever, any person that like any one of the company has ever even met. Like, that's what a CRM is. You can have that in like a hundred thousand

Starting point is 01:55:57 or a million markdown files that like comprises everything that that business is. So that's basically what Gbrain can do. Yeah, I think people misunderstand that like YC has been building software to help batch participating companies for over a decade at this point, I think, probably longer, maybe 15 years. And so that is in the lineage. We're building that in. Yeah, exactly. I mean, actually, I think, you know, now we have all these tools that are going to be

Starting point is 01:56:22 built directly into YC Bookface. So, you know, you're going to want to be a part of the community. Yeah, I've been seeing. It's been getting better. The pace of software development on Bookface and YC's internal tools is accelerating. It's been refreshing to see. that it's, people often have a window into it and they see like, oh, there's this demo, but there is an internal product that is evolving and getting better, which is great.

Starting point is 01:56:47 So the magic moment for Gbrain is basically like being able to take any book that exists in your entire business and then making sure the three books that really matter for the thing you're trying to do are loaded. And then that's basically ASI. Like you don't have to write software anymore. You can just straight up use Hermes agent or OpenClaw plus this. I am trying to make it work with a codex and, you know, co-work. But honestly, I think that those are like Honda's. And honestly, I mean, open plot of me is still the Ferrari, man. It's still the Ferrari, but, you know.

Starting point is 01:57:22 Open claw, that's good. Very, very fun. What are the trends that you're seeing in applications? Are applications increasing in the age of AI? More people wanting to do startups? or people coming from big tech? Applications as in like people. Applications to YC.

Starting point is 01:57:39 People are trying to start companies. Like it feels like in all this anxiety of like big tech layoffs and stuff, like the flip side of that is that there's more entrepreneurial opportunity because we see companies every day. I find myself just pointing people to the YC website

Starting point is 01:57:53 when they're maybe they're not a software developer and they decide to build their first app in the last six months, which there's been many. I'm just like, go follow this, even if you're not going to apply. Like, this is, this is what works. Yeah. Did you see a PG came out and said it?

Starting point is 01:58:11 He's like, if you're non-technical, just stop calling yourself that. Yeah. Because you shouldn't, I mean, why limit yourself, actually? No, I completely agree. Yeah, I don't know. I think we're seeing. Non-technical was always like, because obviously, like, it's not a binary. Like, like, I learned Python and a little bit of C++, and I could, like, sort of deploy

Starting point is 01:58:31 website, but I'm like levels below like a true, like, you know, incredible computer scientist who's like creating Python. But I would just sort of like, yeah, okay, I'm technical because that's good. But it's like, there's obviously levels to these things. And the more you can like, do you know, get away from that stigma? If you sit in front of cloud code, I guarantee you you could fix some bugs in Python and submit a PR that would be accepted. Yeah, everyone can. That wasn't true a year ago. Personally, I think John could solve continual learning. That would be great.

Starting point is 01:59:04 You could do it. It's just a matter of will. I got a podcast to run. I'm trying to be low agency. Low taste right now. Low taste. Low taste. Anyway,

Starting point is 01:59:16 thank you, Gary. Congratulations on Demo Day. Yeah, great. It's always fantastic. Catch up. We'll talk to you soon. Catch you guys. Have a good news to your day.

Starting point is 01:59:26 Did you see? Like the All Black. All Black. That's great. Did you see the Rolex open a new boutique on the Swiss Alps inside of a former telecommunications tower. It looks like a Modern Warfare 2 map. Look at this building. This is an incredible place to hold a boutique. I guess you have to hike up here. But look inside. This is a Rolex boutique. You go in there.

Starting point is 01:59:49 They show you some watches. But this looks like Modern Warfare 2. Hey Carter says, imagine being told they have nothing in stock here. You both know very well they have watches in stock. Yes. What actual resort is it on? I don't know if this is a resort. I think this is a telecommunications tower. But I got to know more about this.

Starting point is 02:00:08 We got to get someone to call in from this particular boutique because this is an absolutely wild, wild, wild scene. But what a great, like, brand activation. It does have a Bond villain aura. But I think it looks great. Imagine posting threads from here. Imagine posting threads. You know what?

Starting point is 02:00:28 Every time Conner Hayes comes on the show, he's like, you guys don't post on threads. I posted twice about this interview. So Connor Hayes get in here and tell us what's new with threads. What's up, gentlemen? What's up? Good to see you again. I'm doing great, man. Yeah, I appreciated the post today, John.

Starting point is 02:00:44 I love it. I love it. We have been getting better. We've been posting a lot more on threads. We're having fun. And also, this is the weird thing, is that like, I think, and you'll tell me, but I think what's driving thread success and the growth is the differentiation. like I go there for car community.

Starting point is 02:00:59 I go there for all sorts of different sub-communities around TV shows and different things that aren't really, it's not work for me when I'm there. When I'm on X or YouTube or even Instagram now, my Instagram, my personal Instagram is just my personal brand, my work product. But threads has, it's more fractured, but in a good way, if that makes sense.

Starting point is 02:01:21 But you tell me, what's driving all the growth? First of all, can we just, can we just hit the gong? Absolutely. Absolutely. For half a billion. I was kind of waiting. Wow. Congratulations. That was a big swing, too. That was a big swing. It's a warm-up swing. The real swing is coming later in the show.

Starting point is 02:01:39 Yeah, so we announced 500 million monthlies today. Huge. I mean, John, I'll get to your question in a second, but the other thing, which I know we've talked in the past when I was on the show about growth coming from promoting the app and Facebook and Instagram. Yeah. You know, another thing that we were sharing today is the proportion of people that are coming to, the threads not from those promotions has been going up quite a bit in the last year, and that's been a big focus for us. And we're doing it while deepening engagement.

Starting point is 02:02:04 So we're up like 130% year-over-year on time spent in Japan, 80% year-over-year in Korea. Part of the playbook that we've been running is to pick these verticals and communities, kind of one out of time, and just blitz them from a go-to-market perspective and a product perspective altogether. And sorry if you guys get background noise, I'm literally like at a team party for this right now. Huge. Amazing. So we've got the entire threads team surrounding me while I'm talking.

Starting point is 02:02:30 Non-family of apps referral traffic. Is that links being shared on the open web? Is that screenshots? Is that referral like invite your friend or share a link in a DM? Yeah, often actually the initial new user install comes from those. And then what we try to do is get people to come back to the app without a promotion the second time and third time, etc. Got it. So that's been the thing that we've been most successful at.

Starting point is 02:02:54 And that's just through like making the content better, getting you better connections in the app, launching messaging, which sends you notifications when you get DMs and just having good content-based notifications too. Yep. Yeah, every, I've I've transitioned from someone maybe a year ago that would click over to threads when there was like I was almost getting click baited by something to now just like a lurker because there's like good. Oh, he's in the lurker category. I'm in the lurker category. Did you transition from a lurker to reply? No, but every reply guy starts as a lurker. Yeah.

Starting point is 02:03:30 I have to build up, I have to build up the confidence to move into that. But no, it's been, it's seriously impressive. And yeah, the car stuff is what I've been following most closely. Oh, that's so cool to hear. We've been doing a bunch of the F1 too. I don't know if you guys are F1 fans, but we have this new feature called live chats. So we're trying to do like, one of the things people love about threads is they call it the quiet app. Like you're in a world right now where every single app you go to is like a sound on video feed.

Starting point is 02:03:59 Yep. You go to threads. You can just like quietly scroll, read things that are going on. One of the things that we're trying to do is make threads the live second screen. That is still quiet while you're watching something on TV. So we had a like a Q&A with Lewis Hamilton and Ferrari last week. There's like him and a bunch of fans in a chat that thousands of people can follow along with, react to the messages, asking questions. We've done this for NBA finals.

Starting point is 02:04:23 So yeah, we're just trying to be the best platform for public conversation, and that just means we have to lean into these lab events. Yeah, so you're basically taking like what's great about the chat on, let's say, like a Twitch live stream, right? It's like this real-time flow of information and ideas, but then making that, like disconnecting it effectively from the programming. Yeah, tie it to a live event and put like authoritative voices in charge of it. we were talking John, I think, a little while ago about just how well we've done reality TV. Yeah. I don't know if you guys are into the summer house lore. I'm not.

Starting point is 02:04:59 Like lives on threads right now. Is it Mike Solana? I saw, yeah, I saw Mike Solana on threads discussing some reality TV show. Maybe it was the, the traders or the, is that the one where they play like mafia? Yeah, there's a couple of these. And it's just so interesting when you bump into someone who you know in a different context and they're on threads and they're just like they're sort of like wearing a different it's a more casual outfit basically and like the the posture can be uh very political on other apps or very

Starting point is 02:05:30 defensive or very aggressive or maybe they're a particular influencer and they're always you know this just killed this on a LinkedIn but then they come over to threads and they're just sort of just being a normal person in some ways if that makes sense it works I mean it's it's I think a lot of it's just people want to wear that hat sometimes I mean you were talking about it was like your stuff too. There's something very like human about building an app that's for conversation and lean into that. And I think in a world where like increasingly so, conversation is just very hard to have

Starting point is 02:06:00 or the person isn't human. You know, we're trying to make that the case on threat. So that's what fun. What is the KPI culture like? Because you're the quiet app now. It's working. It seems great. But like it feels like there's going to be a moment where it's like, okay, well, let's

Starting point is 02:06:17 A, B, test, vertical, endless scroll. video and it's going to outperform and you're going to be like, yeah, we're no longer the quiet app. We're actually just like the second Instagram and like it just maths out. But like, it feels like that's not the mission. That's not the goal. But like how, how will this actually play out? I mean, I think that's the difference between metrics and goals. Like if the, you know, the goal is for us to be the best app for public conversation. I know that I could, I could tomorrow pump time spent on threads by putting more video in the feed. But that wouldn't make us the best that for public conversation. So we try to find metrics that give us as best the possible proxy for

Starting point is 02:06:54 that goal. And a lot of times actually it is like how much is their back and forth reply happening on a top level post? You know, in terms of growth, the main API that the team works on is what we call threads driven out. It's like people that are coming back not because they saw a promotion. So my, you know, my job is to like set those goals, give them the best incentives possible, but then just know that the metrics are going to be imperfect. And sometimes I need to come in and, you know, make a strategic decision to do something that might be against the metric, like, to your point, with vertical video. We're not going to do.

Starting point is 02:07:25 We're not going to do a vertical video tab and threads. Like, that's not part of the plan. It's going to be a horizontal video. You will turn your phone and it will be the entire Lord of Rings extended edition. We're getting 10 hours. It's going to be a Netflix. NBA finals game four. Who know.

Starting point is 02:07:41 We don't. How do you think about building your own kind of culture within, meta. Like is that is that a thing? Is that something that's encouraged? Or is it or or is it like, you know, all big, you know, one big team? Or are there, do you guys like, do you guys have like a nickname for people working on threads? Like, they're not metamabe. They're threaders. Threads. Threads. Okay. Yeah. We have a very unique culture. I mean, I think like all good culture is it kind of like happened on its own based on the people that were excited to work on the app. Because I think if you're coming to work on threads, you want to be on a smaller, scrappier team.

Starting point is 02:08:22 You probably care a lot about culture and pop culture and the types of things that we spend our time on. We just had actually a team of people in here talking about how we do that. So I would say it's encouraged in as much as you're like building value around the culture. But like it's really important for us to not build the culture around the idea that we're different and special. because we get so much value from being at Instagram and that meta. Like all of the infrastructure that we use, all of the integrity teams that we leverage, the monetization that we do comes from an ad team that's inside Instagram.

Starting point is 02:08:56 And so it's like you're trying to strike this balance between making sure people feel like they identify with the app and the stuff that we're doing, but appreciate being inside the bigger company where we just get so much value. Instagram Pro recently launched $3.99 a month, I believe, for some premium. features. I upgraded in instantly some fun details. I haven't even dug into it. Who's viewing your

Starting point is 02:09:19 story twice? Oh yeah. That's one of the features that they show. I like the fact, I don't post on my story that often. I like the fact that I can extend it to 48 hours because I'm not a daily story person. I'm only sharing like news about the show on there when someone else tags me. And so just extending it a little longer. That felt like a valuable feature. But are you looking to that for premium subscriptions there? Are you more focused on the end? ad product on the modicization side, like, because it's a new surface area. It's sort of unexplored. Where do you want all that to go? More focused on ads. Like, we're not currently working on threads as a part of the meta subscription. We love ads. That's a really good high margin business.

Starting point is 02:10:01 I completely agree. No, I like the subscription too, but I see it as a yes and for sure. Yeah. Because the ad business should be so good. I think if threads gets big enough and we have good, you know, brings to people for something like this inscription, why not? But it's not something we're working on right now. Yeah. We launched ads globally four months ago and have just been like slowly scaling it in line with the engagement going up, you know, just trying to be careful about it. Yeah.

Starting point is 02:10:27 Yeah. What is different about a conversational or kind of a news driven platform like threads like acts versus like a more passive product like, you know, Instagram? People say lean forward versus lean back. The big question is like the question that people have always had with Twitter and X is like, why wasn't the ad product like amazing for advertisers? Why did it never really maybe scale to the level?

Starting point is 02:10:56 Like was it an ad matching problem or was it a behavioral problem? There's been this like, you know, the audience on Twitter historically was like so valuable. The user base was always smaller, but, you know, some of the most important people in the world are on there as like their primary. application. And so the big question was like, why was it never a great ads business? Do you have any theories? Like, you know, you could say it was execution or I could say that. Connor Hayes says. Conner Hayes doesn't disagree with Jerry. He sat silently. I don't think it's the format because, you know, we know from years and years of experience. I mean, think about Facebook ads before the app was so video-centric,

Starting point is 02:11:43 that feed format is kind of similar to what you have in like a threads or an X today. And that was one of the most efficiently monetized surfaces ever in the history of technology. So I'm pretty confident that we'll be... Yeah, let's go. I'm pretty confident that we'll be able to get threads ads into a really good place there.

Starting point is 02:12:02 The other thing that matters is like, you know, obviously it's like how much attention are people paying to the content that they see in a feed, but also how much content are they view? Like the more shots you have to show them something relevant, the better you're going to do. So you're going to monetize more effectively if people are viewing 100 things a day versus 50 things a day. So to your point on KPI's, John, like one of the big ones that our team has is just like how many impressions of content do people have in the app? And then we try to make that as deep as possible by giving them more relevant stuff.

Starting point is 02:12:31 And eventually some percentage of those will just be ads that hopefully are really good because they come from the Instagram and Facebook ad in them flows. like advertisers can just opt into their demand going to our surface, which will checkbox in the flow where they create all the IG and Facebook stuff. So I don't know. I think we're set up for that to do pretty well, but we're just, again, trying to be careful and not scale it too quickly while we're still making the product great. It's been under three years at this point.

Starting point is 02:12:55 So I think sometimes people lose sight of the fact that we're still a bit of a young app in the ecosystem. Totally. Will AI-generated ads be more difficult to crack on threads? I'm thinking that if I'm on Instagram and I have, you know, maybe it's Manis or some MSI or MSL like agent that can go take screenshots from my website, reinstatiate them and put that in an Instagram flow. Like, you're showing me the product. Job may be finished, might convert well. But if I'm advertising on threads and I need a really punchy hook and I actually need to think about like 140, 280 characters that are really, really high skill ceiling type rights, that historically has been hard, and that's why like brands come off as like cringy often on these short text-based mediums.

Starting point is 02:13:47 And I'm wondering, even though we have all this crazy acceleration in AI, I haven't followed an account where I'm like, oh, yeah, like this model is really good at writing tweets or posts that are 180 characters. They're honestly better at the deep research. Yeah. That's part of what keeps us popular, I think. You know, like there's, especially when you're optimizing for conversation, at least right now, humans are pretty good at telling when they're conversing with another human. And also, there's a lot of metadata, right? They have a profile.

Starting point is 02:14:17 They have a username. They might be verified. I think for ad creative, you know, on the organic side, posts with photos and them do really well on threads. It's a pretty high percentage of our top performing content. I think it's probably similar on X. And so my sense is that as AI generated ads get more prevalent, it'll probably be through the image path more than the text path. And I don't know, I'm not pulling off of any data besides personal experience on this,

Starting point is 02:14:43 but usually for me, the image is the hook when I'm thinking about buying something or getting more fond of a brand. Obviously, the content matters at some point, but I think the image is going to be what people are. How much do you think about what Starlink will do to global connectivity? There's a bunch of people that don't have Internet today that are going to have, like reliable, fast internet over the next 10 years, people coming onto the internet for the first time. Is that something that you're thinking about or is there just enough people that haven't tried

Starting point is 02:15:17 the app yet that are obviously taking more of the focus? I think probably more of the latter. We're not at the point with growing threads where we're worried about, you know, lacking more opportunity at the top of the funnel for people that are already online or maybe even already online and using similar services. The other thing is we tend to be pretty market focused. Like we, like I was mentioned before, I've been really focused on Southeast Asia, Japan Free Air to the biggest ad markets in the world. We are very U.S. focused. And so in these early stages, I think it's actually really easy, especially if you track a top line global number to see that go up, but not actually build an app that's like

Starting point is 02:15:57 meaningful and makes sense is a little big part of people's lives every day in a country. So I look at success more as like tipping individual countries into threads being the app of record for public conversation. Right now that's starting to happen in Southeast Asia. And then the U.S. is our biggest focus outside of that. So from that lens, I think in those countries, the connectivity is quite high at this point. So we worry less about that opportunity. What's the plan to get at Connor on threads? You got an underscore sitting there.

Starting point is 02:16:28 How did you get sniped? You're the head of threads. The craziest part about this is the at Connor is an Instagram employee, I think. There he was at some point. Yeah. So he's locked it down. Wow. Johnny come late late on our platform.

Starting point is 02:16:41 Yeah. That's hilarious. It's over. It's over. You're never getting it. Walk me through the process of spinning up a little like, I don't know if you call them like micro features or something. Like the year ago, you launched sort of like disappearing ink text, like sort of glows.

Starting point is 02:17:00 and you click spoilers, and these little features are, they seem so simple. They're probably getting easier to build because of AI tooling, but at the same time, your code base can be kind of a mess. 25 people want to do different things

Starting point is 02:17:15 with text at the same time. How do you think about those? What's the importance of that in terms of the growth? So we have actually done a bunch of work to get the process on this, right? I think, like, social products often, you know, there's so much work

Starting point is 02:17:30 that is just like getting the details right, executing well. But there's a bunch of work that's just like a hits business. Like, you know, look at Instagram, getting instance to work. It's just you got to take a bunch of shots on goal with small creative teams. So a lot of this stuff that you see in the app that feels like a little bit more fun, the music attachment feature we did, animated stickers. These are like pods of three or four, you know, technical but creative people that we just give a bunch of space to. We talk to them like once a week or once every two weeks about the ideas they have, give a bit of feedback, and then they've just been coming up with hits.

Starting point is 02:18:05 So the way that, and your point about AI making that easier is true. I mean, I think one of the fun things that we've seen is that design talent has been getting more technical, which has made these things easier to build. So whereas in the past, you would have two people sparring over how to build a feature. Now the really creative designer with great taste can actually develop the thing on the road. That makes a sense. Congratulations on hitting 500 million monthly active users. Fantastic. 500 billion soon.

Starting point is 02:18:33 500 billion soon. Hey, with Starlink, you never know. Economic imperative, solve the fertility crisis because you need more users and that's the fastest way to get more. Everyone should be having the 10 child policy rolled out globally. Yeah, for threads. Great to see you so much. Congrats to the whole team. We'll talk to you soon.

Starting point is 02:18:53 Have a good one. Incredible. Like Futures does it again. This is a post from Bronson-Showen. This is mid-20206, France wakes up. Europe's leading lab releases La Chaton Fact. Their latest open-source frontier model, it's state of the art.

Starting point is 02:19:11 The U.S. and China scramble wildly to catch up. The mainstream narrative around AI has changed from maybe the hype will blow over to a guess this is the next big thing. The U.S. enters talks with France to hopefully gain access, but without much success. I like a little twist on the China Wakes Up from AI 2027, a funny copy pasta at this point, but still, again, congrats to Mistral on launching Le Chetton Fat.

Starting point is 02:19:35 This is important before we're joined by author Luke Burgess. We have a safety alert from the U.S. Embassy in the Sao. Ambassador Herschel Walker has an important message for American visiting the Bahamas jet ski rentals pose a serious risk of injury, death, and death. U.S. government employees are banned from renting them. That's how serious they are. And you should avoid them too. Really taking shots at the jet ski industrial.

Starting point is 02:20:04 A lot of people were surprised to see Herschel Walker. He's a famous football running back. This is how they found out that he is, in fact, the ambassador. But fun story. Is the risk worth it? Jet ski? I don't know. Are you a big jet ski guy?

Starting point is 02:20:21 No, not really. I'm fine. I prefer scuba if I'm on the water. every time every time I'm on a jet ski euphoria good you like it's yeah it's it's incredible they're fun under

Starting point is 02:20:35 I think they're underrated I think they're underrated they feel like very highly rated no but highly rated but are people organizing their life around jet skiing it's gay it is nice that it's sort of the experience of a motorcycle with a lot lower stakes

Starting point is 02:20:51 right Gold Rock says jet skis and vibe coding are a golden combo. Yeah. Anyway, let me tell you about Cisco, critical infrastructure for the AI era, unlock seamless real-time experiences and new value with Cisco. Our next guest is the author of the One and the 99 forging identity in the age of social contagion. Ben, welcome to the show. How are you doing? Hey, what's up guys? What do you think about a jet skis? Yeah, what do you stand on jet skis? We got to start there. We got to start with Jeskies. As one who spends the summer literally on Lake Michigan, I'm offended by that warning, big time.

Starting point is 02:21:27 Oh, okay. You love jet skis. You need to generate some positive memetic desire for jet skis. Okay. There we go. See? Yeah. That's what we got to do.

Starting point is 02:21:32 I think they're still underrated relative to have fun though. Rent to them. Buying a jet ski is idiotic. Rent. Right. Why? I feel like when people say don't buy, they say don't buy the boat because the boat requires so much maintenance.

Starting point is 02:21:47 It feels like the jet ski, you just pull it out of the water. It's not, it's going to dry out. It's not going to need a million. in maintenance of year? It's just like a boat. It's like a boat. Yeah, I mean, it's just what everybody. It's still a relatively large piece of equipment that I don't want in my garage.

Starting point is 02:22:00 I guess, I guess you don't want to own it because you're probably not organizing your life around jet skiing. But if you are and you're going to be jet skiing 200, 250 days a year, 300 days a year, then you might want to own. My number one criticism of Malibu is like lack of jet ski launch points. Okay. Like it's, makes it like it's good for the natural environment. that there aren't, but it's tough to be a catapult from the bluffs to just launch you straight in.

Starting point is 02:22:28 Anyways, more important, let's talk about the book. Give us the thesis, but first, I mean, can you explain what you mean by social contagion? What are some examples? Why that phrase? Yeah, well, first of all, the book that I wrote four years ago was about memetic desire. So I want to congratulate both of you for doing more than anybody in the history of media to generate mnematic desire by talking about founders and who's investing in who and the size of exits. I don't know if you built your business that way, but in the gone. So you literally are on the medic machine. Yes, I did work at founders fund. So maybe some of it rubbed off. Who knows? Yeah, yeah. The people ask us like, you know, how do we book so many guests? And early on,

Starting point is 02:23:14 we saw a pretty extreme flywheel, right? It's like people see someone they know or someone they like or a company they'd like, and then it just created this really crazy flywheel. And now it's been, for the last year and a half, it's been more about, like, managing. It's perpetual motion machine. That's right. Basically. Yeah. Yeah.

Starting point is 02:23:32 Anyway, take us through it. So this book, I mean, in Wanting, I had one job. And my job, I think, was to explain the ideas of this obscure French thinker in an accessible way to the world at, like, an eighth grade reading level. And I think I did my job. But it meant that I had one thing to do, and there was a lot of things that didn't make it into the book unless my publisher was, he'd be very mad if I wrote a, you know, 600-page book. So I took that, all the things that didn't make it into that book, and I applied some of the

Starting point is 02:23:58 ideas of Mimesis and Mimetic Desire to the family, to the education system, to technology, to politics, and to religion, right? And one of the most interesting things about, and what is social contagion, you ask me? Social contagion is just the unconscious way that we catch the desires of other people, the ideas and beliefs and emotions of others, and we integrate them, into our sense of self and the way that we move through the world without really realizing that we caught these things from other people. With desire being a most important one. Yeah, is contagion bad?

Starting point is 02:24:31 Because it feels like contagious. It feels like I caught a cold or a flu. It feels like a negative twist on memetic desire. Desire feels like something. It could take you in a bad place, but also could be good. And just this idea of pulling something you want, what someone else wants. Like that can be good.

Starting point is 02:24:52 Someone else wants to go to the gym, be healthy, do something. Like you want to do that too. That's probably fine. But contagion feels like it's memetic desire, but only in the negative. Is that the way I should be thinking about it? No, no, no.

Starting point is 02:25:04 I mean, I'm very clear in the book. Like there's such a thing as good contagion, right? There's things, the desires that I want to catch, right? Like if I want to be more ambitious and take more risk, maybe I should move to Silicon Valley. Yeah, yeah.

Starting point is 02:25:15 So I think there's definitely good, good contagions out there. So it's not all bad. If you want a jet ski more, maybe you move to the Bahamas. You move to the Bahamas, right? Or yeah, move next to you. Yeah, totally. So, and one of the interesting things about AI in the book is I'm wondering if it's actually

Starting point is 02:25:33 contributing to contagion and Mimesis in a way that we don't fully understand, right? Like, it's, did you guys see that Tim Ferriss blog about what's happening to his nonfiction books? Yes, I know exactly what I'm talking about. Yeah, like he had an 80% drop in sales of his book. And it coincided with like 2022 when Chat ChbT came out. And he's saying like why would somebody read prescriptive how to nonfiction when they can go to, you know, chat Chbett or chat bot and ask, well, here's my life. Here are all the things that are going on for me. Apply, like what should I do?

Starting point is 02:26:06 And by the way, like summarize Tim Ferriss's book and make it highly personalized for me. Like, why would you read the book? You get instantaneous, highly personalized customized advice. So I wonder though if that is actually going to. lead to a form of contagion that we don't even understand because there's something happening in the AI. Even the engineers don't fully understand what it's doing. And it's giving us back something that feels really personalized to us. But the inputs are obviously being drawn by what, you know, large language models and what other people are putting into it. So I wonder if we're entering

Starting point is 02:26:40 like a pluribus, you know, the Apple Show kind of situation where the AI is actually, while seeming and highly personalized, it's actually sort of contributing to some form of social contagion. Yeah. I feel like we are in the sort of like, we're just starting to diagnose the social contagion that breeds on social media. And we're just starting to understand the effects of what seeing certain videos or political biases or even memetics just like you see a sports car and you want a sports car and that changes your perception of things. And, and with, AI, it feels like we're in this unknown, unknown period where it could, it could wind up creating more the pluribus. It also could wind up being more personal. And so everyone sort

Starting point is 02:27:27 of goes into their own world. There's this weird stat where, like, the, I think it's like, over 50% of high school students are a super fan of some creator that none of their friends know about. And so they are in this, like, niche internet micro-sum. celebrity world, nym cells, where they know everything about this one Instagram or this one TikTok or this one YouTuber's life, their repertoire, their whole catalog of work. And they talk to their friends and they're like, no, I've never heard of, you know, Markiplier or whoever is like their choice or you're, you're a Doug DiMuro fan and you run into a car fan and they're like, I've never heard of Doug DeMiro because you're in this like niche world. And, but at the same time,

Starting point is 02:28:12 you do have this like, you know, like ideas do spread extremely quickly. And we're just starting to diagnose that. And with AI, it's, it's somewhat similar where, but potentially even, even more removed, because you might be getting fed Tim Ferriss insights, but it's sanitized of Tim Ferriss's name. And so it feels like it's personalized to you, but you're actually doing what everyone else who's asking the same question is. It's very hard to diagnose. It's very hard to track the actual outputs.

Starting point is 02:28:40 What do you look for in the book in terms of like just level setting on like how things are going in America, the state of things, whether we're on the uptrend or downtrend on a particular category or even just like the level of memesis. That's something that I can't really, I don't have an economic factor for. I can't trace it to GDP or incomes or anything like that. Yeah. I mean, it seems like the book is basically about the relationship between the self and groups in various communities, right? And to what extent can we be in something without becoming completely of it? Whether that's a political party or a city, you know, if I live in, you know, Vegas, which I did, there are certain things that I really like about living in Vegas, but there's other parts of Las Vegas that I don't want to, you know, permeate me. So it's like, how do we be part of something and take the best from it and give our best?

Starting point is 02:29:37 without it completely forming us and determining everything about us. So when it comes to the state of the U.S., I mean, I'm finding that maybe it's because of social isolation, but I'm finding people over identifying with various communities, whether it's like national level politics. It seems like when people don't have like a solid sense of self, which is a term that I use throughout the book, like a solid sense of self as opposed to a pseudo self that is constantly adapting to expectations and conforms. as we walk into different rooms and move to different cities. It's kind of the malleable version of ourselves. It seems like, you know, to the extent that people don't have a solid sense of self, they're much more likely to be sort of liable to capture by the first group, party, whatever,

Starting point is 02:30:24 company that comes to them and tells them how special they are. And I sort of noticed this over-identification, right? That worries me quite a bit. And it's like groups want everything from us. They want complete loyalty, which, you know, like right from the very beginning, right? It'd be like meeting a girl at the bar and like, you know, she's asking for, you know, a full commitment right in the beginning or else you don't love me kind of a thing. It's like, no, we actually need to take time to like test things and get to know each other. And it's probably not a good thing to completely identify. We like to situate people and people like to situate us.

Starting point is 02:30:59 And some people own those labels and some people are kind of repelled by those labels. So the book is kind of looking at identity and the way that. that we, how easy it is to situate people and how we can become, you know, how can we become a little bit immune from being overly situated ourselves? And the most interesting people to me, I don't know, like a Teleb kind of a figure,

Starting point is 02:31:19 are those people that are really difficult to situate like politically, you never quite know what they're gonna say. They're the most interesting people for me to follow. Yeah, Renaissance man maxing. What do you think of the maxer culture? Because that feels like the, in many ways like the worst case of what you're describing where, whether it's looks maxing or fitness maxing or whatever

Starting point is 02:31:42 maxing you're doing, it's like everything else about the identity has to go by the wayside. And we saw this very interesting clip with clavicular, the popular looks maxer, where they ask him about politics. And he doesn't even have a read on any politics whatsoever because it's completely irrelevant to what he actually cares about, which is just looks. And it feels like there's the inverse of that, which is someone who's politics maxing or left wing maxing or right wing maxing and they don't have an opinion about anything outside of their political domain. And if you ask them about health, they would have to tow their particular party line. And that, it feels like in the competitive sphere, maybe that's something that's going on with maxing. But how have you been

Starting point is 02:32:25 processing that, that trend of maxing? Yeah, yeah. My colleague and friend Jordan Castro did a whole thing on Clavicular, who you've seen on the show, which is worth watching. I mean, my book is an anti-maxing book, and I think maxing is evidence of exactly what I'm talking about, because when you max anything, like, what are you giving up in order to do that? Like, you can't max everything, right? So I'm pro, like, maxing in my marriage. I'm pro maxing love. But apart from those things, like, there's really nothing that I want to max because it turns me into an individual that, I mean, what's interesting to me is that there's some people that just max one thing, and then they get exhausted and then they just look for another thing to max.

Starting point is 02:33:04 And that was me throughout my 20s, right? I was maxing on a company so hard that I didn't even, and I didn't even like it, right? And I would get bored with it and then move on and start another one and max on that one. And it's like maxing cannot be an end of itself. I went through this with health. Like there was a period of my life in college where I would, it felt like 80% of my attention and energy was just going towards being healthy.

Starting point is 02:33:31 and I eventually got to the point where I was like, okay, the point of, at least for me, the point of being healthy is so that I have energy to do other things and I can feel good day today, hopefully live a long time. And I just went from, I went from like 80% down to like 10%. And I probably got a little bit less healthy, but it just freed up all this energy to do other things. But the benefit of going through that maxing period is I learned like a lifetime's worth that I now still get the benefits of. And that's part of what probably keeps me higher from a health standpoint than my previous baseline.

Starting point is 02:34:09 So I would say for me, it's like going through periods of obsession has been healthy, but it's like how do you remove your, how do you eventually figure out the time to remove yourself? So it's fun to grind, but you don't want to get lost in the sauce. Periods of obsession is the key word there, right?

Starting point is 02:34:27 Periods of obsession. I don't know if anybody's, done some kind of empirical research study on this, but I would guess that a very low proportion of hardcore maxers have children. You know what I mean? That's been a game changer for me. Instinctually, I think everyone agrees with you. If you had to turn your book into a Tim Ferriss-style self-help book that would then be

Starting point is 02:34:52 summarized by an LM, what would the output say? Well, it's not a prescriptive how-to book, first of all. all. But hypothetically, turn it into one. If it were. No, but, but, but, but to me, like periods of obsession, not getting lost in obsession. Well-rounding, maxing and Renaissance man. Yeah, yeah. I mean, it would, it would probably tell, you know, I think that we're really bad at leisure

Starting point is 02:35:19 and especially like ambitious entrepreneurs and founders. And I've been one for most of my life. So one of the things that it would probably tell people to do is actually think of leisure as a skill that you actually have to get good at. I suck it. I remember being on a beach in Thailand when I was starting at one of my companies and I could not rest. I could not sit still. I could not have any fun at all. So one of the things that we do at the Clooney Institute, which Jordan and I run together, is we quite literally offer really busy founders and people the opportunity to come on like humanity's retreats and chill out for a couple of days and think about the human person and think about important questions

Starting point is 02:35:54 with the promise that they're going to go back refreshed and with a better sense of what they actually want. want to max out on for a couple of days actually. But like leisure is not a popular thing to talk about in this world. But, you know, learning to do it well has been one of the most important innovations in my life. All ties back to the jet ski. I do my best thinking on the jet ski. Yeah. You got to be able to turn off. It is really, really hard. Where do you stand on, on digital detoxes, dumb phones, black and white screens, those sort of like dopamine, like dopamine, like dopamine, fasting little mini techniques that people like. Sometimes they can become their own pursuit and their own obsession.

Starting point is 02:36:38 But where do you stand on all that? Like what is the secret to like unplugging, I guess? Yeah, I mean, I think you've got to know yourself. The micro mini little tools have never worked for me because I'm, I guess I'm like an all or nothing kind of guy. So I have to like go hike the community of Santiago for 30 days to witness any change whatsoever, right? For better or worse.

Starting point is 02:36:58 That's like the kind of guy I am. If there's like some incremental n plus one change that I'm making to like, you know, I don't know, the background of my screen, it doesn't do anything for me. So like, but to each to each his own on that stuff. Well, thank you so much for coming on the show. The book is the one and the 99 go pick it up wherever books are sold. Thank you, Ben, for coming on the show. Yeah, let's do it again soon.

Starting point is 02:37:24 We got to do this again soon. This is a really great conversation. We'll talk to you soon. Have a good rest of your day. and let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents.

Starting point is 02:37:38 And without further review, our final guest of the show, And Agraška, from Notch. That's right. He's the co-founder and CEO, Onda. What's going on? Welcome to the show. How are you doing? Great to see you. Hey, guys. I'm good. Good to see you, too, Jority.

Starting point is 02:37:52 Good to meet you, John. Good to meet you. It's an honor to have you on the show, on a big day. Thank you. the news. But first, introduce yourself for everyone. Introduce the company a little bit. Yeah, thank you. Hey guys, my name is Anda. I'm the co-founder and CEO of Notch. Notch has been the leading digital intelligence and optimization platform for Fortune 500 brands for the past six years or so. And our news today, coming live to you from the Guggenheim,

Starting point is 02:38:16 actually, where we've taken over the Guggenheim for the day, is that we just announced ACE, which is infrastructure that makes websites convert both humans and AIDS. agents. Okay. So what is the, what was the first big client that you brought on board, the first big logo? What did you actually try and move the needle for them? How much of that is focused on e-commerce conversion versus brand building, broadly marketing. Like, where's the surface area of the impact of the product? So when we started going to market, we actually moved from Silicon Valley to New York with my co-founder. And some of our first clients were the biggest banks. still are. If you can think of the biggest bank in the in the world, there are a client.

Starting point is 02:39:02 Our service here at your impact, look, we, we track how audiences journey through digital experiences and we help brands understand what's resonating with these audiences, what's not, and how they can ultimately gain more customers by making better content and better digital experiences. With ACE, this feels like a product that people have realized that they need really acutely over the last year. We had Matthew Prince on. I believe it was maybe the middle of last week, and he had recently shared that there's more web traffic from agents now than humans. That flip came even faster than he was expecting, and so now there's a bunch of Fortune 500s that you guys are working with that are kind of waking up to this, realizing that they need a solution.

Starting point is 02:39:51 But I guess talk about kind of the origin of the new product. What we saw across the biggest websites in the world, which we measure with notch, what we saw was the same thing. There were more agents starting to come to websites than humans. And about a year ago, every single CMO became obsessed with GEO or AEO, which is how do I make my website readable to agents? But ultimately, LLMs cannot be gamified. And the tricks of GEO, which is publish another 100 blog posts and 10 other product. F-A-Q pages and whatever, it ended up really destroying the human experience. So in the last six months, we've actually seen crazy high immediate bounce rates,

Starting point is 02:40:35 like human beings that actually arrive to websites and are getting a horrible experience because they're presented with this agent-ready, agent-first content. So A's, the origin story was really, how do we build a system that can personalize for each of these audiences without taking away from the experience of the other? And that's making the agent side invisible to humans and vice versa. Like how does it actually work? Yeah, so let me tell you how it works. So ACEs, first of all, we have to ingest all of a brand's knowledge from currently it's sitting in silos.

Starting point is 02:41:07 The website is just one area where that knowledge exists. So we ingest that knowledge alongside brand guidelines, governance, et cetera. That knowledge gets vectorized and atomized, and then it gets semantically tagged. This step is really important because what you want is to enable both the human, humans and the agents to retrieve this information through natural language. Then we start to reassemble the information based on implicit inputs. These are kind of the data signals that we've been capturing through notch for the past six years. And that reassembly enables for a really personalized human experience as well as a person agent experience. So for example, I'm a mom. I've

Starting point is 02:41:45 been looking for an electric card that doesn't look like a mom van. And I've been doing my research in LLMs and I know nothing about cars, unlike my husband. And so I type in electric car that doesn't look like mom van, and it's telling me to go to, you know, GM.com, Ravin.com, et cetera. When I get to that website, which I will still go on because I want to validate the information, I want to see, like, is this brand the brand that represents me, what I get is really static, really generic, essentially a dumb website that doesn't actually speak to me. And so it's a very jarring difference.

Starting point is 02:42:16 And LLMs have really conditioned us to expect conversational, like. experiences and the ability to really command these experiences through semantic intent. So what we've built with ACE is the ability to give GM the capacity to allow people to just copy-based their research from an LLM, put it on the site, or just simply say, I'm here to find an electric card that doesn't look like a mom van, and then the site reassembles with dynamic atoms in response to what you just said. So the whole website comes to the homepage. There's no more pages. It's really just dynamic reassembles. assembly based on your semantic input. When an agent shows up, the same thing happens. It's the same

Starting point is 02:42:55 knowledge base. It's just visualized differently. It's a different modality. It's really just code. Very cool. Yeah. There's so many people talking about like generative software, software that's created on the fly. But forgetting that like websites in their current form mostly have just been like effectively digital catalogs that are kind of, you know, there's just one, one version of it. But that's very cool. I wonder, yeah, the idea of like, you know, it, and that's, that's like, in some ways I've, like, stopped going to a lot of websites because I know it's easier to get the information that I really want. But the thing is, like, then the brand is losing, like, full control over how they're being displayed. Maybe they want to be showing, like, images of different

Starting point is 02:43:41 points that and obviously the chat app or whatever consumers using is, you know, they don't have any control over kind of that whole service area, which is rough. Yeah, they really don't. I think the website as we know it to date is definitely dead. I think we have to replace the word too. Like I'm starting to play with knowledge site because I think what really matters in tomorrow's world, whether or not humans still come to your experience, right? What matters is that you have to, have that knowledge in a knowledge base that can speak to agents. Right now, what brands are doing is they're just cramming a bunch of crappy content into the same website.

Starting point is 02:44:20 And it's not personalized for humans and it's not personalized for agents. And so the experience, of course, is going to turn people off and people are going to stop going to website. So we've got to enter a new era where websites are dead and long-lived a knowledge site. What's the long-term... Oh, yeah. No, my immediate thought is like you want the, you want like, the source of truth. and then you want to be able to effectively just prompt the

Starting point is 02:44:45 and be able to get the information that only matters to you. But what were you going to say, John? I was going to ask about just the, obviously, you've seen a ton of success in the enterprise. Are you thinking about going down market? How do you think about that? What are the decisions that would lead you to work with smaller companies or is there just so much value in such a unique problem in the enterprise

Starting point is 02:45:09 that that's where the company will be firmly placed for the long term. We're going to go everywhere, John. We have big ambitions. I think this product that we built can serve everyone. And actually, if you go to our website, it's running on ACE now. And we've built it in the past 48 hours because, you know, we're a startup, so we leave things last minute. But if you go to our website and start prompting it, I'm sure, like, I just said this, and so many people are going to go try to break it.

Starting point is 02:45:35 So, you know, knock yourself out and send this feedback if you manage. But yeah, I believe that any company needs to have a knowledge base that's accessible by agents, that they can control and maintain in a really structured governed way. I also believe humans are not going away. Like every time there's been a new tech trend, whether it's social media or something else, we've always been like, guys, this is it. All we need is a Facebook page. I don't believe in zero-click marketing. I believe humans like choice. I believe especially for high intent purchases.

Starting point is 02:46:06 We really like to do our research. And so I believe websites are going to be around because it's kind of a yes-and situation with marketing. It's always LLMs are just another channel ready to opt-a-mic bar. Yeah, I just went to notch.com, as you said, and tried it. And very cool to have the bar on the bottom. You can just... And then they basically just creates a section

Starting point is 02:46:28 that's just like effectively generating whatever information you want, which is like, John's talked about this before, like, Ramp, you know, who we work with. like it's nice to have access to language models within the application, right? You don't. Sometimes you want to take it, you know, back to your preferred agent, but having intelligence where you are already makes a lot of sense. And I wouldn't be surprised if this is like, you know, hopefully you guys capture as much

Starting point is 02:46:55 value as possible with this like new paradigm, but I'm sure a lot of other people will happily follow suit. Well, Dirty, I think there's actually a great use case for TBPN as well. So if you guys want to be customers, I'm all. years. I mean, you guys create so much great content and you're truncating it in a really great way, making it accessible in a way that enables you guys to, you know, just give people the information they really came there for would be cool too. So let's talk. Totally transcripts. I love it. Well, thank you so much. So great to finally have you on the show. Okay, go check it out

Starting point is 02:47:25 and get the demo today. Thank you so much coming on. Thank you guys. We'll talk to see you. Have a great rest of your day. Let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents, whether you're writing code, analyzing data, creating content, or automating business workflows. Codex helps you move projects forward from start. Over on X, Daniel's advocating for banning social media for over 65s. Yes, yes, the inverse of the UK's project or the UK's initiative to ban social media for under 16, over 65. I think that would ruffle some feathers. That would probably be pushed back on more by.

Starting point is 02:48:04 the social media companies because the purchasing power from 65 onward is so high. There's so much wealth with the boomer generation. That would actually be very devastating to Facebook and YouTube and TikTok even. The young folks, I think all the social media companies have an incentive to get those users as early as possible, keep them on the platform. But I think more than anything, they want a level playing field. So Connor Hayes can duke it out with every other social social. media site as soon as someone turns 16, try and acquire them, try and get them on board.

Starting point is 02:48:40 But we'll see as that rolls out and where the guardrails are set. We'll close out the show with Sean Frank. Everyone is doubting meta. Spenders are mad at bugs and model changes. Investors think they're missing AI, like the AAR saga all over again. But as a large-ish customer who will spend something like $100 million with them this year, meta is working and no ad space comes close. to their level of scale.

Starting point is 02:49:05 Don't doubt the Zuck. The fact that you can buy that you could have one SpaceX or two metas for the same price is pretty unbelievable and certainly not something that most people would have guessed even a year ago.

Starting point is 02:49:25 Yeah. It is a crazy, crazy situation. At the same time, rockets, they're inspiring. And social media is sort of... Totally. I mean, I can see how we got here, but it's still surprising. But yeah, obviously, fantastic business.

Starting point is 02:49:43 And yeah, people are probably need to, you know, reality check with Sean Frank, because he's in the trenches, he understands the business better than most. So go pay attention to him. Anyway, thank you for tuning into TBPN today. It's been an honor and a privilege. Leave us five stars on Apple Podcast at Spotify. Sign for a newsletter at TBPN.com.

Starting point is 02:50:02 We'll see you. We love you tomorrow. Goodbye. Cheers.

TBPN Live - YC Demo Day Lightning Round, New Snap AR Glasses, SpaceX Rips | Garry Tan, Andrew Lee, Stamatios Floratos, Hugo Frisk, Efraín Torres, Russell Smith, Payton Case, Akshay Trikha, Diana Hu, Harj Taggar, Connor Hayes, Luke Burgis, Anda Gansca

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.