AI Video Game Mods, Should You Sell Your Data, Bryan Johnson on Endurance Racing | Sean Frank, Linda Du, Ali Partovi, Sigil Wen, Jerry Wu, Bryan Johnson

You're watching TVPN.

Today is Monday, October 5th. There are 80 days till Christmas. We are live from the TVPN UltraDome, the temple of technology, the fortress of finance, the capital of capital. Let me tell you about ramp.com.

Time is money, save both.

Easy-to-use corporate cards, bill pay, accounting, and a whole lot more all in one place.

We are—

80 days till Christmas. Okay, so October is the hardest month in the road to Christmas.

Totally.

And it's because everyone is gonna be looking at you—

We're so close, but it feels so far.

But there's a distraction.

And yeah, you're right.

Halloween.

Halloween. If you go up to someone and you say, hey, there's only 80 days till Christmas, they'll be like, why aren't you counting down the days to Halloween? They'll start looking at you like you're crazy. You're not crazy. They're crazy for thinking that it's not worth counting the days till Christmas in October. There's only 80 days. Get ready. Halloween is a speed bump on the way to Christmas.

That's right.

That's what I've always said. So keep counting down those days. Ignore the distractions.

Or don't, but we will.

Yeah.

Anyway, let me tell you about Cisco. Critical infrastructure for the AI era. Unlock seamless real-time experiences and new value with Cisco. Okay, who wants to play these AI-generated video game mods? Kotaku does not. They posted an article that said, I don't want to play your AI-generated mod that puts Skate 3 inside Modern Warfare 2 inside of Minecraft. They don't want to play them. Tyler, what do you think? Do you want to play these or did I have to force you?

I want to play these.

You want to play them?

I'm actually playing it right now.

Okay. Well, Tyler's playing one of these AI-generated mods and it's very interesting because the mods are Whoa, we got the infinite loop.

Infinite loop.

Very cool. Anyway.

Sorry about that.

Production team's working on it. Anyway, we've seen a lot of these. We talked about them. We pulled up a random one that we actually did an Instagram reel of. I think it was Minecraft and Elden Ring. I've never been a big Minecraft person, so it wasn't big for me. Tyler's playing Modern Warfare 2 in Minecraft. But what's actually going on here, Tyler? Explain, like, how did you get this installed? What is this?

Yeah. So from my understanding, it's basically like there's no—

you're not really like putting one game into another. It's fully like a Rust engine and then you're pulling assets from both games and like kind of pulling them together somehow.

Interesting.

So like, I had to buy Modern Warfare 2 on Steam.

Yeah.

It's using the actual game assets. And the same thing for Minecraft. Like, it's using the actual assets.

It found them online.

Interesting.

And then I basically had Codex like set it up, wired up so I can run it on Mac.

Yeah.

But it's like fully featured. Like, I can mine, I can build.

You can build.

Using the—

Can you get a nuke?

I don't know if I can get a kill streak.

That's the question.

I gotta figure that out.

If kill streaks aren't in it, it's not Modern Warfare 2.

Yeah.

It doesn't make any sense. There are actually a whole bunch of different different ways people are modding things. And it's interesting because they're all getting sort of lumped together into one bucket, but they are very different. Like the classic, like the most approved or like company-approved mod might just be something like, you know, there's a Fortnite skin and you get to, and I don't know if they actually did like a Batman skin or a Spider-Man skin, but those skins come over and it's just assets that fully live inside of Fortnite. There's no game code from any other game coming over. It's purely just— Okay, the Fortnite game developers looked at what Spider-Man looks like, they made a Fortnite model out of the Spider-Man, like, character, and then you can play as them and everything is integrated fully. You're not going to risk changing any game files. It's fully approved by corporate. Now, these go a step further and run the gamut of different techniques. Some of them that you have seen, some of these mods actually are just the old ways.

Yeah, I'm playing. Could you try swimming? Yeah, swimming is a little bit broken.

And that's the nature of the vibe-coded slops.

Oh, it's over for you.

Yeah.

Are you drowning though?

I don't think so. I'm just basically walking as if there was no water.

There's no— yeah, the mechanic is— because, yeah, I think your health is being monitored by Modern Warfare 2, whereas Minecraft is just the geometry that you're around. I don't know. But some of them are the old way of modifying, just bringing game assets into the actual game. None of the actual code is running. Others are actually merging 2 games. For that Mario in Elden Ring video that we saw the other day, the mod actually borrows a subsystem called LibSM64, which is derived from a Super Mario 64 decompilation. So what they'll do is they'll—

Can we keep politics out of video games?

Please. Yeah.

Please.

So what they'll do is they'll download the actual binary, like the code that is on the cartridge. They will convert that into just a file that's on a computer. You can run that in an emulator. But then they can actually reverse engineer the underlying code from what has been compiled. So you get basically a bunch of ones and zeros and there are tools that will look at that and work backwards to understand what the actual game logic is. And so, an example is like, in the game code, you might have like health equals 100. But in the compiled code, it might just be like some memory address like X— like 0x1234FC. And then you might work backwards in the decompilation step to figure out that that's the health. So, a bunch of things happen in the game, the game's played, as things are changed and updated, that game state is then sucked out into the decompilation file. Interestingly, the NSA is involved. So, no, we can't keep politics out of game development because they're actually— a lot of these game modders are actually using software from the National Security Administration, the NSA. So the NSA has a GitHub with an open source project called— what is it actually called? It's called Ghidra and it's a— it's on the National Security Agency's GitHub. So, you're saying that the NSA is GitHub. It's a software framework for reverse engineering software. And so you point this thing at the game and you wind up getting more of the code, and then you can break it up and bring it into other games. Tyler?

Okay, I'm gonna trick shot this pig if we can cut to that.

Trick shot this pig.

No, no, no.

Oh, he did it! He did it!

Come on, Tyler. He did it.

Why?

Little—

can we get a content warning? Oh, okay. I think it's fine. Anyone at home?

Minecraft is a pretty focused game. But there is something else going on now that I hadn't seen before. I don't know if this was popular in the modding community before, but there is a version you might have seen. We can try and pull it up, ArkWeb. It's a Batman: Arkham Knight passthrough Spider-Man mod. So, you play as Spider-Man in The Batman game, and you're— so you're in Arkham. And what's funny about this is that the 2 games feel very similar, so they mesh very well together. But from an intellectual property perspective, this should never happen because, of course, Marvel is DC— or it's Marvel and DC. So Marvel is owned by Disney and Batman is owned by Warner Brothers. These 2 companies are direct competitors, and so they should never really be collabing in this way. But the fascinating thing about this is that it's not just they took the Batman game and created a Spider-Man character. It's actually running both games. And then it's passing data between the games to create this experience of being Spider-Man in Arkham Knight. So what's happening is like the mod goes into the Batman city, Gotham, grabs all the geometry, all the collision data. All the geometry of like where that building is. It takes that and in real time, it passes it to the Spider-Man game, which is actually running a different physics engine. So it has to translate all of that. Also, the 2 games use different— they use different coordinate conventions, different axes, and different units of length. So it has to translate all of that, create that in Spider-Man. And then also keep both of them running like the controller input because normally, if you have a controller on a video game, it'll just be on one game at a time. And if something's in the background— but you can see there's like, yeah, bugs already. And this is like part of the problem. And then it needs to actually merge the graphics. And so the games use different graphic systems too. This feels like it would be so much easier if it was 2 games that they were built on the same underlying technologies. But through the power of slop, they're able to sort of merge these 2 in real time and get them actually running at— seems like near 60 FPS. And then even when you go into— they don't show the fighting, but I think when you go into the fighting, you switch into like the Batman combat setting instead of the Spider-Man system. So you're using Arkham's free flow combat. It's very, very odd. But the end result, it looks great and it feels like it's almost ready to serve as a reason to be like, oh, I love that game. I want to play it again. Why don't I give it another spin? Sort of a New Game Plus. You get to play as a different character. There's a lot of old games that are getting remastered, rebuilt. Could be an interesting reason to go back into a game that you loved with a different twist this time. And stuff like this happens all the time. I mean, the— even like the latest sequel to Spider-Man was the same city, but just a different character. And so some different mechanics, some different story elements, but a lot of the same assets. But it was still a reason to jump back in for a lot of players. But all of this is the classic 90/90 rule. Have you heard of this? Coined by one of the greatest of all time, Tom Cargill—

At Bell Labs.

At Bell Labs. He said, when it comes to software engineering, The first 90% of the code accounts for the first 90% of the development time, and the remaining 10% accounts for the other 90% of development time. Because you get 90% of the way there, and you're like, great, I only have 10%. And then you spend even more time on that last 10%. No matter how much you estimate it, things always grow in scope. And so when you think about the bug bash that would be necessary to take that to anything relative to like—

You're looking at 5 years.

What?

You're looking at 5 years.

You're looking at 5 years, yeah. Here's Minecraft and GTA, similar structure.

Yeah, the sort of multiplayer first-person shooters.

Yeah.

You could imagine actually that being like relatively straightforward to get a working version that would be fun.

It all just depends on how you scope it.

Yeah.

Because if you're just talking about you're only changing the textures, like you're using the same model from Call of Duty And you're just changing the textures, you're probably not going to introduce too many bugs. But as you include a different physics mechanic, a different explosion, like, all of a sudden, you're introducing so much risk that you could fall through the map or things could not collide correctly. You got to go rewrite that collision software that triggers something else. Then— and all of these demos, none of them show the actual, like, playing the full game. Like, there's a mod to bring Minecraft into Skyrim. It's called Skycraft, I think. And in the developer's documentation for, like, the open source implementation of this mod, it says, like, don't start the game normally. Like, do not try and play the opening sequence. It will break. You should start with, like, the alternate starting point that starts you further into the game because it's so broken. And that's just the nature of being in the slop era. So— The interesting thing is that all these mashups, they feel like tailor-made for the short-form video era because you watch one of these for 30 seconds, 60 seconds, and you're like, wow, that's amazing. I have dual nostalgia. I love GTA V. I love Minecraft. Put them together, double love. It's amazing. Everybody is, you know, getting like millions of likes on these videos. But no one has actually played these all the way through. No one is sitting back down and being like, okay, yes, this is actually as good as GTA V, which a lot of people played for 100 hours or more, same thing with Minecraft, the base game remains dominant in terms of like actual entertainment value. This is pure meme value. And so, it feels like it's another Harry Potter Balenciaga moment, but for video games. And the Harry Potter Balenciaga video by Demon Flying Fox, that was released March 15th, 2023.

Wow.

Can you believe it's been 3 and a half years since that? That's a while, right? That's a while.

Yeah, it makes me think, how did they make it at that time?

That was AI video. That was the best AI video was. But what they did was it was you would train a dedicated model on every character. So, they would train a model on Harry Potter or on Daniel Radcliffe basically. Just his face, and then they would use AI-generated imagery to to generate something that looked like a you know sort of art station style Balenciaga runway show in a in a Harry Potter scenario like at Hogwarts, and Midjourney at the time was pretty good at that style. The the the the dark like Sort of odd backgrounds. Like, it was okay if it had some hallucinations in the background.

Yeah.

Because it leaned into that, because it's a dark fashion show in a castle. You don't really have a strong reference point for that. And then they would run the face swap, and then they would animate that image just a little bit. And then there was a ton of editing to actually— to make it so that you would never linger on a single frame for more than a second or 2. And so you'd move on. But what was the conclusion from seeing that? I mean, at the time, people were like, we're going to be watching full AI movies next year. What was your conclusion? Did you have one?

From the Balenciaga?

Yeah.

I mean, it's just like, these are all, like, very novel ideas, but at some point, like, now I've seen, like, I've seen a bunch of different games that have been ported to each other and it's like, okay, there's another— it's like, oh, yeah, you put Minecraft and Mario and I kind of get the idea.

And it's notable that—

Ideas.

Like, even Flying Fox did many, many sequels and sort of stopped breaking through to the same degree and still has a YouTube channel and I think posts a bunch of stuff. But we didn't get Daniel Radcliffe at a Balenciaga runway show, and we also didn't get a fully AI-generated movie within 3.5 years. And so, it was still a heavily important tech demo though. And I do think, you know, there will be a breakout video game mod that sees heavy assistance from AI in the near future. But the key is like crafting a really innovative strategy loop. Like, that's what Dota 2 was. It was a spinoff of World of Warcraft, I believe. It was a mod that came from there, or maybe it was from another game, I forget.

Yeah.

There are plenty— I mean, Counter-Strike started as a Half-Life mod. Like, there are clearly games that can be birthed from that. And the fact that you have AI tooling to speed that up, you're just moving a lot of people into the indie developer category who can move things faster. They're still going to wind up spending, you know, 99.999% of their time on the last 1% if they want to break through. But It's cool that many people will see this and be like, I gotta make a mod, I gotta make something. I've always had this idea for this game and now I can actually go and get started. Although they will obviously need to lock in for a long time to get something actually polished. 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, own the data platform that powers it. Let's move over to some news in the timeline. Theo is optimistic about this. Millennial age gamers are about to become more pro AI than the average SF tech bro because of all the nostalgia from Halo in Modern Warfare 2. I do think some people will see this and be like, oh, this is cool, I like this. But there are still plenty of gamers who are like, I don't like AI at all for a bunch of reasons. And they bring that over. There was an interesting chart of support for AI.

I think for the gaming community to become broadly pro-AI, a very well-loved franchise has to release a new game that is noticeably better or bigger because of the technology. And until then, I think people are just going to be like, you know, again, these sort of combinations are cool, but Until— I don't— well, one, I don't think they can really get users because of all the IP issues. It's just blatant IP theft. But yeah, I think you're going to need to see GTA 7, and it's the entire world, right? And for people to get excited. And I don't know that that's even a path that they'll take.

I like that lawyers are extremely pro-AI now because they're like, I'm going to get so much business from this. Yeah, yeah, is this the Zillennials FTW shared by Antonio García Martínez from The Economist, seeing a monster. Quote, how serious do you think the risks of AI are of AI causing serious harm? Everyone is basically moving to the right. They're more worried just in the past 2 weeks, 3 weeks, Just through from August 28th to September 18th, except for millennials aged 30 to 44. They saw Jacob Coxon, they saw Dario Hamadej, they saw the SNL sketch, they saw, you know, the Ezra Klein interview, and they were like, yeah, I was worried before, but I'm good now, which is kind of crazy.

Less worried than ever.

Less worried than ever, apparently, according to The Economist. No, it's, it's It's a narrative violation.

Yeah.

It's a narrative violation. But yeah, everyone else is getting increasingly more worried. And of course, the top of that chart, that is the voter base, 45 and up. So if you're seeing even 65+ move to 40% very serious, that's going to move some elections.

Jim Prosser says, it's a fascinating outlier. My hypothesis, we're the gen most comfortable with this degree of technology change and uncertainty due to coming of age through the internet and mobile.

Yeah. I mean, a lot of millennials are in a very fortunate position with like not getting caught off guard by brain rot and slop and screen time addiction. It's like we sort of got brought up in the craziest era And then have seen the children data and can fight back on that. So I think a lot of people in their 30s and 40s are like, yeah, I'm set up for success. I'm not getting the kid the iPad on day one, that type of thing. So—

Yeah.

It feels a little bit more like there's a danger, but it's known. It's like we also grew up with cigarettes, but we were like, yeah, we knew the risks of cigarettes. So we were like, yeah, we're— We're going to treat that accordingly.

Yep.

Only when you're drunk.

All right. Sean Frank, our first guest of the show today.

Let's go through this when he's here.

We can go through it live.

Okay.

No. Yeah.

Yeah.

No, no. I'm agreeing.

I'm agreeing.

We're going to go through it right now.

You're like, actually, I decide when we go through.

Yeah.

Let's bring this up when we get to him.

The whole segment?

Yeah, yeah. I mean, whatever you want. I think the interesting crossover here— I mean, we're going to talk about data brokerage. Sean Frank got an offer to sell Ridge's data. He turned it down. But we'll talk to him about the rationale there. The interesting thing is from Dan Romero. He thinks that there's a private equity opportunity to buy old video game IP and then offer a $10 unlimited mod license per per game or $100 per bundle. So, yeah, like taking all these old games and then setting them up for, you know, use in AI systems. So when you go to a model and you say, I want to make a first-person shooter that I can hop in a car and it's a racing simulator, and then I also want to be able to skateboard, instead of needing to write all of that logic from scratch, you could just pull, like, the actual racing game that they licensed off the shelf. Counter-Strike 2, I mean, Valve has created an insanely robust modding system and they actually created a new mode in CS2 called Rush and it's built entirely in their modding system. So they're like leaning into this in like a closed source way, but they will allow you to build an entirely new game in their game. And there's a lot of examples like that, everything from Roblox to Fortnite to, you know, plenty of others that have had modding tools for a long time. The question is just like, how do you interface with the, with the models directly?

Suds says looking for 500 megawatts, minimum of 250, max 2 gigawatts of powered land for a port. Go company is completely price insensitive and has billions behind it. Ideally electrified shells already built. My first thought here was Isn't a 2-gigawatt data center going to cost you like $100 billion? So you're like going to buy a house, and you're like, I'm very price insensitive. And they're like, cool, well, the house is $50 million. And you're like, well, I have $5 million, but I'm price insensitive. Yeah, so I'm sure there's— He's like, I bought some So when you're getting into 2-gigawatt territory, so Chase Lockmiller goes, I think we can help. It's like, there's just not that many folks that can get you 2 gigawatts, but worth a shot. Worth a shot. I mean, he got half a million views on this. He probably got to the right people, right? So effective.

I wonder who's spinning up 2 gigawatts, because it wasn't— Wasn't Meta's Hyperion or Prometheus site like 2 gigawatts? This is like hyperscaler territory, right?

Yeah.

Like there just aren't that many startups that could be— like it's funny to own Meta and be like, my Portco. Like, I mean, Portco is very small coded, right? But you're just like, yeah, like my Portco.

So Hyperion is designed to deliver 5 gigawatts. Okay, okay.

Well, it's not Meta. We ruled them out.

2 gigawatts by 2030 is their plan. And then 2032 is when they get to 5.

So, um, you know, there's this, uh, there's this news story going on in, uh, in, um—

No, it is, it is, it is, it is noteworthy that, um, you can, uh, you can have billions behind you and, and still not have anywhere Close to what you need to secure that kind of capacity.

Yeah. Fun fact, if you had a 1-gigawatt data center, pretty significant, and you were willing to burn the entire Amazon rainforest, you could power it for 35,000 years. Not bad, right? There's a lot of worry that the Amazon's going away because of Bolsonaro. People are worried. The lungs of the Earth, we need to maintain the— to maintain the Amazon. But it is an incredibly power dense—

Would you defend the Amazon?

I think so. It is a hard trade-off, because every other industrialized country was like, yeah, we burned all our resources. Now you're the last one with them. Don't burn them. So it is tough. It's like industrialization for me, but not for thee. That's the real tension there. Anyway, what's Will Moneitis saying? He says, probably worth noticing that all the guys that were shilling high-fee SPVs 6 months ago Now all have power data centers they'd love to sell you.

Interesting.

How are those high-fee SPVs doing now? Probably OK.

Pretty good. John, question for you.

Yes.

Do you think it's fair game to ash your Nature Valley bar out of the window of a car? Or is that kind of, like, too close to littering?

It really does produce an insane amount of crumbs.

Yeah.

Have they never thought to do a V2 of that product? I know.

I know. I mean, for how around Nature Valley bars were, it does feel like they've completely fallen off.

Really?

I haven't seen it. I used to see a Nature Valley bar once a day.

Yeah, they did sort of have a market lock.

I haven't seen one in years.

It's kind of fallen off.

What happened? What happened? Tyler, can you do some research into Nature Valley sales?

At the same time—

Over the last 30 years, my entire life.

Just from a health perspective, doesn't the insane crumbliness of the Nature Valley bar give you sort of reassurance that you're eating actual food? Because a lot of protein bars these days are so mashed up, you're like, I don't know what kind of ingredients are in there. You had to destroy 50 cows and 12 pigs This much milk and this much corn to like make this amalgam. And with the Nature Valley bar, it looks like what it is. It's just like a bunch of—

Gil says, one a day, baby. They haven't gone anywhere.

There we go. There we go.

Putting up numbers still.

I love it.

The other resource people haven't really been talking about at all, maybe they should, is construction paper. Celine says, you don't really hear about construction paper. much anymore. Back in my day, this was always— I felt like it was top of mind almost constantly.

Top of mind.

As a child, certainly.

So, I know construction paper useful for just drawing on. It's thicker.

Yeah.

The ink doesn't go through the paper. That's a benefit. But were people actually constructing things with it? I know people would make those, those Like those chain daisy chains around holidays, perhaps Christmas.

Yeah.

It's more of a Christmas ornament than anything else. But I don't know that I— it feels like if you were building like a real diorama, you would get out the cardboard. So maybe cardboard, just squeezed construction paper a little bit. And I mean, have you seen those cardboard cutting saws? It's like a saw designed to rip through cardboard because people get a lot of boxes. But that's also had the knock-on effects of if you're an ambitious child and you're trying to construct something significant that might last the test of time, you might forego the construction paper in lieu of the cardboard. What do you think? Is cardboard responsible for the downfall of construction paper?

That's interesting.

Because we—

Because there's an eco-angle there, John, right? It's like people getting a lot more e-commerce products today.

Yes.

coming through the house all the time.

We're post-scarcity on cardboard in most households, especially ones with children, because you're ordering a lot of things online. There's a lot more cardboard around. I know the last few construction projects I did, cardboard was my material of choice.

Yeah.

So maybe that's what's going on here.

Yeah.

I'm glad we got to the bottom of it.

I'm trying to find the best way to get exposure in the market to construction paper.

You want to go long?

It's almost a perfect store of value over the last 5 years. There's Fabbrica Italiana Lapis Ed Affini.

Okay.

It's been completely— it's a half a billion dollar company, completely flat for the last 5 years. So more on this on the next episode. We got to talk about Topic Crawler.

Okay.

I'm on spent half an hour going back and forth with ChatGPT so it could make me a tool that spits out link for any topic he's researching, and it actually gave me exactly what I wanted in the picture below.

And then the follow-up only two minutes in, and I realized how dumb this is. Reinventing the wheel. Reinventing—

Using the tool to create— use, use the tool to create the exact same tool.

I mean, the funny thing is, I, I don't know if this was— I don't know if this was like created as a joke from the start and the whole idea was like, let's just be funny, uh, in the first post because the, the punchline will be in the second post, or if he actually went through this. But, uh, it does speak to the fact that like a lot of the, a lot of the functionality is really buried. Like, just the ability to, like, put something on a cron job is, like, not advertised anywhere. You know this, right?

Yeah.

Like, if you want to, like, get a weekly digest of, like, where the swells are the biggest across the world for surfing, like, you can totally do that. You don't need to build a tool. You don't need to write a line of code. You don't need to host anything. It doesn't need to use anything else other than ChatGPT. But you would never know that from any of the marketing, really, because there has never been just like a cron jobs are here, or whatever you'd call it. Like scheduled tasks are here, and there's like a national campaign. Very, very difficult as the models get and the harnesses get more capable to actually educate people. But it's very fun.

Still very cool that you can build your own Google in like 30 minutes.

Have you seen Let Me Google That for You?

No.

So it's this website, letmegooglethat.com. And you type something in and then you generate it and then you have this link. You can copy the URL and I'll share it in the production chat because it's a funny— you got to do the actual thing. And so you send this to someone if they ask you something really obvious like, how do I get to— How do I get to— How do I use ChatGPT? And then it moves the mouse. I think you clicked the first one that was empty. I think you need the second one. The second one has the actual punchline. Here we go. So how do you use ChatGPT? You send them this, let me Google that, and it just Googles it for you. And then you just— and it just loads Google or whatever. You can close it, but it literally just goes and like types it in.

But anyway, what are people putting in their IRAs lately?

Yes, babe, our retirement is in Charizard. I have a little bit in my Roth and 401, but we can fly or crater with Charizard. Fly or crater is funny terminology. Is there a way to get a Charizard into a tax-advantaged retirement account? Has that been done yet?

There must be.

Anyway.

There must be.

Well, if you are investing, do it on public.com. Investing for those who take it seriously. Stocks, options, bonds, crypto, treasuries, and more with great customer service. You got to invest like the best. You got to do it.

This hit me over the weekend, Jon. This was big.

OK.

One horse pulls 4,000 pounds. 2 horses working together pull 32,000 pounds, and it's getting community noted. Record deadweight pulls by pairs of draft horses top out around 17,000 pounds. Okay. Near 32,000 pounds. Anecdote this is based on, which is Dave Ramsey also uses 8,000 pounds for one horse.

Not— Okay.

So wrong on every dimension, but the idea stands. I mean, in the community, John.

Yeah.

It made me think—

What?

What are we doing with just one horse here in the open?

We got to get 2 horses.

Permission to order another horse, a twin horse.

No, wait. Did you see the Beverly— they're auctioning off that bear sculpture that's on Rodeo Drive?

Whoa.

This was on Facebook Marketplace. They want $95,000 for it. Do you know the one I'm talking about?

The geometric bear? Is that a bullish indicator? They're like, we have no use for this bear anymore.

Maybe, maybe. I sent it to a buddy. I can pull it up, but we can also move on. Talk about that later.

It would be cool to get a large statue of a bull and a bear here in the UltraDome fighting, battling over market conditions.

I'd like it. I'd like it.

Million-dollar home. A lot of people have been talking about this. We didn't want to have to talk about it because it's It's not, you know, the nicest topic, but we'll talk about it. Why is tossing boiling oil down a ladder during a siege always a one-time trick? If I were being sieged, I would have 200 guys all boiling 3 or 4 cauldrons of oil each. If you come anywhere near the walls, you're getting scalded. Can't melt it, idiot. Such a good point.

Yeah, yeah.

Such a good point.

Yeah, you should just have like a continuous flow of oil pouring down.

Or a closed loop.

Oh, like a waterfall. Like instead of a wall, you have a waterfall of hot oil that's just constantly there.

Yeah.

That would have completely reshaped the battlefield.

It would have one-shot, one-shot the—

Yeah, your defenses would have been impregnable. Fascinating. Anyway, the million-dollar homepage is reborn. Born every decade in a different form. A TikTok girl is selling ad space on her wall to pay for tuition. They're calling it the TPN effect. Not bad, kid. She says she's selling ad space on her wall. Is this presumably in her dorm room or something? $500 for one huge spot, $50 for a smaller slot. Yeah, I mean, I don't know. Maybe.

Pretty good play. Resourceful.

Yeah. Million Dollar Homepage was a really good bit back in the day. You buy individual pixels. And people have redone it a few times. I mean, I guess the Reddit, the place was sort of a free version of that, but it was pretty cool.

You can— I mean, Delta just taking blow after blow because you can now watch Cheeky Pint on your United flight. Everyone's powering on. Again, really, really just showing that United is, is a, you know, is a serious, highly commercial airline. They don't want you watching movies, TV, music. They want you watching the Cheeky Pine archive.

You gotta learn from Susan Lee.

And I'm hoping this means that we're gonna see more episodes of Cheeky Pine hitting the timeline. It's been about a— feels like a 6-month holiday, but it's time to come back. What's that?

Take a break.

Oh yeah, hangover.

That's right, hangover. They're taking a break, resting off, sleeping it off. Yep, they'll get back in the pub soon.

That's right.

Oracle is putting $10 million into the Nashville Symphony, reinstating the 2026-2027 season and recalling musicians and staff who were set to be furloughed October 18th amid a financial emergency. That's great. Give it up for Oracle and the Nashville Symphony. It shows a real sense of civic responsibility.

There we go, Larry.

I love this. This is great. More tech companies should do this. Go and solve little problems like this. Everyone loves it. What else is going on? The Brits have turned their back on their day ones. British people are apparently buying Chinese knockoff Land Rovers, uh, especially lookalike Range Rovers. They love Chinese cars. The JQ7 made by Chinese Chery. I'm learning about a new Chinese automaker every week now.

We got mosquitoes in the Ultra Deck.

Let's take off.

We got mosquitoes.

Get the flies.

Heads up, heads up.

You think you can get it?

No, I don't want to.

Go for it. Run around. I'll read this while you fight the mosquitoes. It's a nice little side quest. Get the tracking cam on him while he goes and hunts it down.

With the gong now.

Let me tell you about CrowdStrike. Your business is AI, their business is securing it. CrowdStrike secures AI and stops breaches.

Speaking of CrowdStrike, I mean, incredible performance. George Kurtz and CrowdStrike Racing put on a clinic yesterday. Incredible comeback. They had some setbacks early on in the race, but Made it back to the top spot and won the entire season in LMP2. What a, what a year.

Pretty crazy.

What a year for CrowdStrike Racing.

Pretty crazy. Well, let's go over to this. I want a review of the Temu Range Rover, Jordi. You gotta let me know what you think. When Mark Thulis went looking for a new family car this summer, he was struck by how much cheaper the Geico 7 was than luxury brands with a similar look and feel. The 33-year-old from England's south coast calculated the upfront cost of his preferred version of the Chinese SUV at the equivalent of around $50K, half the price for a visually similar Range Rover model. You think that car is visually similar? It's pretty—

We'll see.

Its resemblance to the 56-year-old British luxury brand has earned the Jaiqoo a double-edged nickname, the Temu Range Rover. My first reaction was, what's the catch? He said a test drive convinced him that buying the Jaiqoo was a no-brainer. No big industrialized economy has embraced Chinese cars more enthusiastically than the United Kingdom, where Brits like him are being won over by the combination of affordable prices with status signaling, styling, and technology. It does look pretty good inside.

Yeah, calling it a Temu Range Rover doesn't quite work because I would expect that the car is just better in most ways than the range. You know, when you think about like a you know that the Temu branding for a product, you think like it looks good in the image, and then in person it's like yeah, it falls apart. It's a mess. I would expect this car to be actually pretty solid.

Yeah, probably very reliable.

Yeah, they're very good at making products.

Yeah, I wonder how much undeniable these Chinese brands have. Rolled out dealerships and service departments because that seems like the dealer footprint seems hard to spin up quickly, but they must have figured out a solve because I would imagine that potential buyers are worried whenever you're whenever you're dealing with an upstart brand. It can be really tricky if the brand goes under while you own the car because there's just nothing really to work on. A lot of people went through that with the what was it? Fisker Ocean. The Fisker Ocean, I think. Fisker maybe went through restructuring and a lot of people got stuck with cars that couldn't update software.

Yeah. Every once in a while, I'll see a Fisker on the road and I'm just like, incredible.

Well, not the Ocean.

No.

The Fisker Karma.

Yeah.

Which I saw Jay Leno in one recently. And the Fisker Karma. It does look like— I don't know, I would expect it to be like the modern Back to the Future car. It's like the new DeLorean. It's so unique. It's not quite as striking, but it's still very cool. Britain's vehicle—

Wait, the Fisker is not an EV?

The Karma is, but the— or the Karma is not. The Karma was hybrid.

Okay, that makes sense.

Yeah.

Every time I've been seeing a Fisker Karma, I've I assumed it was an EV.

No.

And I'm like, what, do you have like 5 miles of range?

Yeah, yeah.

Like a 2012 EV from— No, no, no.

I mean, beautiful designer. The lineage there is— I think the founder had designed some other iconic cars in the past. Anyway, only Russia and Brazil developed economies that mostly buy cheaper models have imported more than the UK.

The Fisker Karma might be a buy.

You think so? How much is it?

$20,000.

$20,000 for a Karma. Certainly stand out, turn heads for $20,000 or $20K. $20 stacks, $20 racks. The UK, a cradle of the car industry, showcases what could happen if in America if Chinese cars were let in. Range Rover began a press release trumpeting Geely's early sales success. Is that a press release from them? Are they just dancing on their enemy's gravestones?

That's crazy.

That's crazy. We got to figure out who actually posted this Range Over thing, winking at its likeness to the Range Rover Evoque, the smaller one. The managing director of a dealership in Leeds, Yorkshire, said the appeal was simple. It looks expensive, it feels expensive, and then you find out how much it costs. For now, Chinese cars are essentially shut out of the large US market by high tariffs and restrictions on Chinese software. President Trump in September repeated his openness to Chinese cars produced in American factories, alarming Detroit. Oh, that's interesting. He wants the jobs, he wants the manufacturing base, the control over the industrialization, but Detroit automakers just don't want more competition. So there's actually some tension there. Interesting. Congress is debating a bill that would permanently ban Chinese cars on U.S. roads. The spectacular growth of Chinese brands in Britain this year has been led by Chery, China's top vehicle exporter. Through September, its sales more than tripled compared to the previous term. Customer open-mindedness—

I just don't understand why they don't use a name like Chery for the actual brand.

Yeah.

Like, I imagine that could—

Well, it's not spelled how you think. It's C-H-E-R-Y.

Yeah. Still better than J.Crew for an international market. We got to talk about the Topolino.

Okay.

The Wall Street Journal is saying that it's the greatest car ever invented.

Really?

And they'll never go back.

Take me through it. But first, 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.

They're not quite saying that. My week in a $15,000 microcar. Joy, guilt, and 25-mile-per-hour road rage. Fiat's all-electric Topolino isn't fast and doesn't get far, but the Italian company thinks it's just the car some Americans need. What the F is that? A driver of a luxury sedan exclaimed at a busy intersection. Is that a real car?

Is that a real car?

Aww, said countless passersby.

The world is so funny.

Indeed, the Fiat Topolino, a $15,000 electric microcar that tops out at 25 miles per hour, is a real car that Americans can buy right now.

Can you spec it with red velvet on the rope?

That would be nice.

Can you spec it with, with matte black and red velvet so it looks like you're Driving a nightclub, I think you got a winner.

The Topolino, which went on sale this summer, is a bet by the Italian brand that Americans in 2-car households may be open to a much smaller and slower ride for their second vehicle. At just over 1,000 pounds, it seats 2 people and is powered by a small electric battery that gets 46 miles of range on a full charge. It is sold in one color, mint green, and comes in 2 variants. One has doors, a heater, and a glass sunroof. Then there's the Dolce Vita, which swaps in ropes for doors and includes a rollable soft-top roof. Many onlookers ask if it's even street legal.

You can take the roof and the doors off this thing?

This is so not functional, said Alexa after jetting across a busy street to get a closer look at the Topolino, which translates to little mouse in Italian. But so far, demand has exceeded expectations. Fiat, which initially shipped 300 Topolinos to the US. It says 30% of its dealers are signed up to sell the pint-size car. The Stellantis-owned company knows the Topolino. It's got a particular audience. It doesn't step on anything else that we do as a company. There's nothing else like it. I totally agree. I think it is— I'm I think they deserve Car of the Year. I haven't driven it. I haven't seen it. But I like where it's going.

Fiat Topolino at $15,000 or the Chery knockoff Chinese Topolino variant for $800. Which one are you going with?

They are going to make this the price of a phone. Tough call.

They're going to make this the price of a phone.

Yeah, that's a good point. How much cheaper can you get? I don't know. That's a good point.

I mean, I would imagine that a lot of the savings of competing with Range Rover are just the margin, the badge value. I would be surprised if they can make this for under $5,000. But at $8,000, at half price, they actually might move more units because why spend $15,000 when you can spend $8,000?

I say move to lease only. You will own nothing and be happy in the Topolino.

In the Topolino.

But imagine pulling up to Car Week in this. Rolling deep with 20 of your best friends, you know, in a in one of those sort of S you know snaking across the road.

Oh, electric.

Electric.

Fiat Topolino or base 2007 BMW 3 Series? Autoblog called it surprisingly good looking. Jordy, I want your take. Is the base model? of the new 2027 BMW 3 Series. Surprisingly good looking to you, Jordi? Take a look at that and tell me what you think.

So I don't— I mean, I've never been a BMW guy, but I feel like every time a BMW gets announced, everyone hates it. And then a lot of them, a lot of them like end up growing on people. And I don't think that— I don't think you should judge a new model based on the base version of it, right?

Okay.

Because people buying the base version are typically not massive car people or anything. They're just buying effectively an appliance. So anyways, people obviously hate it, and they're bringing up a bunch of cars throughout history, like the E36, and saying, like, we have to return to these older editions. But I don't know. I just— I think it's fine. Like, it's just not that big of— I think it's not that big of a deal.

I think the interesting thing is that it just feels like they've been dancing around with different design different design motifs, especially on the front grille, for the last 10 years, and they haven't really found their footing. This feels like a response to the last round of backlash around the sort of front kidney grille that people didn't like, that looked like buck teeth. And they got a bunch of pushback about that, and they said, fine, we'll change that. And now people don't like this. And then— but people have been warming up on the older styling, and so you're in this weird, like— constantly dealing with feedback, but the build cycle is 5, 6 years or something. So you're refreshing—

Pull up this image of the 3 Series evolution.

OK.

Because everyone wants to go back to these super boxy versions, which, again, I think are cool. But part of it is that you've seen them In— I don't know, you're a movie guy. You've probably seen these cars in 50 movies, right? And portrayed in a cool way. And that helps a lot in terms of how you think about the vehicle.

Now pull up the Shrek spec of the new BMW 3 Series. And let's take a look at that and see if this does anything for you, Jordi. The Shrek edition.

Whoa.

Pretty.

It looks actually like pretty—

People will look at you with a straight face and say, this is ugly.

They will.

They'll look at you and say, BMW didn't push the frontier with this new model, right?

Yeah.

And they're willing to just lie straight to your face. So anyways, I don't think it's that big of a deal.

Interesting.

I'm excited to see how it eventually looks. in its GT3 form factor. I think the current GT3 Evo is like an amazing looking car for what it is. And I think by the time that it— this sort of like these new elements, like the new grille and the lights and everything, get sort of built in, in, in different— into different form factors, I think people will come around to it.

Cool.

Hold me to it.

Anything else on cars today? Any other car news? We got another FSD video out of Jason Oppenheim who sold a Bentley, bought a Tesla Model Y, and is blown away by FSD. It is an inflection point for sure. People are excited about it. Sean Frank, who's coming on the show in just a minute, was singing the praise of the Tesla just, what, a month or two ago? We were like, I don't know about that. And lo and behold, here we are.

Well, we also have Brian Johnson coming on to talk about health and endurance racing in about an hour.

Let's see. Would you buy this BMW 3 Series Dakar Edition? Look at this. I think—

Is this just you and ChatGPT right now?

No, this is official. Officially.

See?

This thing looks insane.

It does.

I love it.

It does, yeah. This would be good.

Give me 10 of them right now.

So many tires on there going across the country in that. Yeah, the last story we gotta get to is Apple plans to introduce new privacy controls for Mac users, warning of the risks growing around granting broad data access to third-party software, including AI agents. Your iMessage data is not— I don't think it's encrypted when it's on your hard drive. So you can potentially like give an agent access like, yeah, just read, just be able to edit my files. And then it can also read your messages and maybe you don't want it to and maybe you want that to be a specific permission. And so Apple is sort of grappling with what permissions it gives when. At the same time, Ben Thompson seems like he's on the verge of going Linux on the desktop because he's been very frustrated because he runs a headless Mac Mini that he remotes into when he needs to see the screen. But most of the time he's just having an AI agent write pieces of software for him. But those pieces of software need permissions and the permissions are UI popups. So you have to jump in and click the button and say, yes, I allow it to, read from the network or, you know, interact with the screen or interact with the file system. And so, he's been very frustrated by this and is sort of questioning whether Apple— they're so clearly set up on the hardware side to take advantage of the hardware boom, the hacker boom, but he went back to Paul Graham talking about the moment maybe 20 years ago where, like, all the best hackers, software developers were using Macs. Mac became dominant in, like, the startup ecosystem, and then it just became like every company just provisions Macs and MacBook Pros for basically everyone regardless of what they're doing. And there's this question— Apple's in this, like, tense moment of, you know, how much freedom should they give? How much privacy should they give? How much flexibility should they give to the user? How responsive is that you can— if you take all the guardrails off and you're on Linux, you can— the agent can just do anything. That's a benefit, but it's also a risk. He also shares a fascinating anecdote that his Mac Mini got hacked. Like, there was some system that was on an open port, came in and installed a crypto miner. But his AI agent actually detected it and sent him a note on Telegram and said, like, urgent, you've been hacked. We should do something. something about this. And the AI agent was like, oh, no, no, no, you don't need— you shouldn't use me right now. I should sit this one out. And he was like, you're dead wrong there. You're fixing this. And it did. And it went and reset the Mac, like, you know, cleared all the— closed the port that was opened so it couldn't— so the hackers couldn't get back in, and then also cleared out the crypto miner. But it's very funny that It was like this referendum on like you creating more vulnerabilities, but you're also able to fight them because typically that's something like if there was a crypto miner on your computer, you wouldn't find out until you were like, oh, I'm trying to do some stuff, it's slow, let me go figure it out. Oh, Apple just patched this randomly. Okay, it got— it fixed it, it detected it later downstream. Instead, it's like 15 minutes.

That's such a wild scheme.

Yeah.

Where, where you're hacking, I'm sure, Oh, yeah.

It's just like scrambled.

If they're getting to millions of devices, they would actually have a meaningful amount of compute.

Yeah.

And it's free.

Anyway.

Interesting one. Mark Gurman reports since becoming CEO, he's talking about the Terminator, he's immersed himself in the industrial design and human interface teams. People familiar with the department say he works hands-on in the studio several times a week, sometimes visiting multiple times a day. Cook typically visited for about an hour monthly. Ternus is locked in. I'm very happy about this. So much of the— so much of Apple's interfaces have just become, you know, completely horrible over the last 5 years.

It does put the pay cut in a different perspective because he's working more hours while making less. And so, he's getting pinched both directions. On the hourly rate, he's way below Tim Cook. He was pretty close on the top line, but given that he's maybe working more hours—

You think Tim had the time, but he just decided to take—

I don't know. What was he doing for the other 158 hours in a month or something that he was working?

He was thinking of what gifts to get the president.

Yeah.

Well, New Balance recreated the iconic Jobs keynote image for the 1992 shoe of choice. The 992. The 992. Do you think these will be popular?

I mean, they're really fantastic. Or is it too much?

Is it too LARPy?

They're really fantastic shoes.

They are?

OK.

So you think they'll sell well even though people will be like, oh, this is the Steve Jobs shoes if you're in SF? They're just a great pair A great pair of shoes is a great pair of shoes.

Yeah.

Okay, we'll leave it at that. 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 to finish. And we have Sean Frank, CEO of Ridge, in the TDPL Theater Dome, coming on down here in person. Good to see you. How you doing?

Whoa! There we go. There we go.

Those are the 992s.

Oh yeah, dude.

Made in America?

Oh, okay. So wait, what did New Balance actually do?

They just recreated the image.

Oh, they just recreated it.

They've been selling that exact shoe for like 30 years.

Okay, okay. So it's not a new limited run.

It's definitely out of style right now. But it's still a fantastic pair of shoes.

Dude, you say that.

But I think that the beautiful thing with the 992 is like you can, You can just wear the same pair of shoes over and over and over, and for like 5-year periods over the course of your life, you'll be like, wow, I think it's just early. This guy Sean is— this guy Sean is like, you know, he's got taste.

One, I'm a style icon. Check it out, dude, rocking the Palantir tennis merch. Uh, but their numbers just came out. I mean, they're privately held. New Balance will do $10 billion this year. Oh, so they're, they're, they're mobbing the other shoe companies. Nike's going down.

Yeah.

Nike's in trouble, right?

I mean, Hoka just reported numbers. Horrible, horrible quarter for Hoka and Decker.

Interesting.

I mean, On getting crushed, but getting crushed, dude.

These—

you say they're not cool, but a lot of people buying them.

Yeah, that's not always synonymous with cool volume. A lot of people shopping on Temu. That's true. That's not a hotbed of high fashion.

Yeah, Deckers is an interesting, interesting company. They have Hoka, which just went on this crazy run. Didn't they buy Hoka for like Like a few million dollars?

I think it was under $10 million.

Yeah, and now it does like billions a year. It's like a—

$4 billion a year.

Yeah, crazy business. So they have Uggs too, which again, kind of all these— once you have an iconic silhouette, you actually have some type of durable revenue from it, but there's still this like crazy up and down.

Yeah, and they got into slippers, and that had like a cool moment, maybe 2020 or whatever. Like the UGG slippers?

Oh, yeah.

Yeah, they crushed with that. But they used to have a bunch of other brands. It's really just those 2 now. Like, they report all the other brands as just like other, and it's a big category.

Yeah, it's like Teva. You remember Tevas?

Yeah.

Talk about not being cool.

Back in the day, if you had a pair of Tevas on, a couple Nature Valley bars, and some construction paper, you could do anything. You were ready to take on the world.

Yeah, that's good. Who's putting up more FSD miles? You Tesla boys, Tesla fam.

Oh dude, you got a Tesla?

I capitulated.

You know, he was texting me like talking mad shit.

I was like, bro, you have to get a Tesla.

It's so easy. I mean, I'm at like 95% FSD, maybe more, 98.

You think you're higher?

There's some bug in the software where it says I've done zero FSD miles, but I've done hundreds at this point.

Interesting.

The one real issue with with the Tesla is like having to charge your car in all these like super seedy places. Why aren't you charging it? Because like gas—

What are you doing?

How far are you driving, bro? I went on a road trip Friday. I went to the new wave pool in Palm Desert. And there was no way— they advertise the range. And of course, they're like, well, don't charge the battery more than 80% because it's not good for the battery.

Oh, you crank it up, bro. You do it anyway.

Yeah, I'm going to do that.

Of course.

And so the problem is like gas stations have all this like super prime real estate.

Yep.

And the supercharging stations will be in just like these like dark, sketchy parking lots off the main road or whatever. So you end up like hanging out by your car just in like not the best places. Yeah.

But now if someone tries to shoot you, you can, you can drive off. That was a, that was a feature they added.

What do you mean?

Like if it detects a gunshot, it just drives? But you definitely don't want to be in the car while it's charging.

No, no, so it used to be you couldn't drive when it was charging, but now you can.

Oh, you can just zoom off. That's great, that's great. So it just rips out the—

Yeah, that's crazy. The only Tesla charger I've seen is like at Harris Ranch up the 5, and that's a very nice stop because you go in, you get a steak, it's great.

Doesn't— isn't the smell horrific?

Uh, not inside the restaurant. It sort of gets you ready for the experience.

Back when I was a kid driving by, driving by Harris Ranch, yeah, you put it You would get on the recirculation beforehand. No, I would be the prankster in the back.

Oh, you would roll down the windows? And if they didn't have child locks on, I would roll down the windows. That's brutal. That's brutal.

So you didn't sell your data?

You turned it down?

I turned it down.

What'd they offer you? $15? Something like $10? $5?

I would maybe take that deal.

You would?

Hey, hey, hey, all the labs, this guy's got a price.

I mean, $15 million, yeah, maybe. No, it was, uh, like, yeah, there's a bunch of companies now. And like, there was a viral post that, um, yeah, there was like 4 ads back to back, like, yeah, all these different—

I've been seeing the Micro One ads for sure. And then I saw, um, I, I've seen a few other people post about them.

Are they a sponsor?

No.

No.

Okay, because they're— they have an aggressive affiliate program. I don't know if you guys were part of it or not. I'm not in there, but they're— they're— if you're willing to make an intro to a brand, they'll give you $50 grand. So like, as soon as I did that post about—

$15? 50.

50?

Yeah, for an intro to a brand. So I did that post, and I'm like, yeah, I'm not going to sell. 50 comments being like, but if you do want to sell, please use my link.

Oh, wow.

Interesting. Yeah.

OK.

But yeah, $480 grand, it just wasn't worth it.

OK. Why? Why? It seems like it's— I know it could be a Faustian bargain. We can get into that. But it's free money. It looks like free money for a lot of companies.

Yeah. Historically, throughout all of human history, people have said there's no such thing as a free lunch. But AI is changing a lot of things. It's possible now that—

This time is different.

This time is different.

Yeah.

Yeah, well, look, if you get a bunch of cold emails from people all offering you 6 figures, there's probably something else going on.

I mean, I'm sure you get text messages and emails daily for lines of credit too, right?

I can show you 50 of them that I got today.

Yeah, so many. It's so annoying. What actually worries you about potentially selling data? Because right now, if I go to any LLM and I ask how much is Ridge doing in revenue, it will probably estimate that on things you've said on podcasts and like open web data. And then if I ask to, you know, build a financial model that models the margins, it might be able to get close with like first principles thinking like it's a McKinsey associate, but it's not gonna have the actual data. But if you sold the data and it wasn't deanonymized, it might actually wind up in the training data. Is that the fear? Or is it someone being able to go to an agent and just say, like, build me a Direct Ridge competitor? And it's like, well, you're in luck, buddy, because I got all the data. Is that it? Or what makes it unsettling?

I think, I mean, at a high level, it's like you run a business that does hundreds of millions of dollars of revenue. and taking half a million dollars to introduce any incremental risk, right? There's already all this risk to running a business.

Yeah.

You sell technology products now. Who knows where that market goes? You sell fashion apparel-style products that, again, go through cycles and risking the golden goose for Not even one egg. It's crazy.

Yeah, dude, we're talking just like maybe an egg white out of that, right?

Yeah.

Yeah. So I'm not very worried about— well, a couple things.

Yeah.

The first is, to Jordi's point, $480 grand is not that much money. Sure. Right? Privileged position to say that. I know a lot of people would love it. But, you know, I could probably negotiate up to maybe $1 million, maybe $2 million or whatever. But still, it's like—

King in the castle. King in the castle.

It's just not going to make that much of a difference in the year, right? You know, we're up 50% this year. We were already doing hundreds of millions of dollars. Oh, wow. You got any more?

Whoa, whoa, whoa.

So it's like, it's just, it's not that much money. Number 2, I think the LLM platforms are gonna get the data anyway.

Oh, sure, because you're using them.

Yeah, like, you know, Codex is an amazing tool I use all the time, right?

I mean, if you're in the enterprise plan, they're not training on the data, but there's still other ways that stuff gets—

They're going to keep adding features, and it's like, hey, if you just give me a little more data, I have a little more data, it gets better and better. Astra is already amazing. Like, in 2 more version cycles, it's going to run my entire business. Of course it's going to have all my data. And they're like, hey, we have to train to make it a little bit better. I'll be like, oh, sure. So I think they're going to get it for free anyway. The third thing is—

And there are other companies that have the data. Like, Shopify has your data.

Totally.

And like AWS might store the data, and they just have different retention and what they can do with it. The MicroOne offer, the data broker offer, is like, we're taking the data and we can do whatever we want with it, right?

Yeah, and they do claim that it's no customer data. They're like, hey, we're not—

They'll do de-anonymiser. They'll run the de-anonymiser, right?

Yeah, and they're like, you know—

Anonymiser.

There's probably some indemnity, so like you're not gonna get sued for selling your data or whatever. But still, it's like, look, If I'm getting random cold emails nonstop for $480 grand, I think we're probably in the beginning of the AI wave. Maybe it's going to be $4.8 million in a couple of years.

Ooh, so you're holding out. Yeah, I'm holding out.

So if you want rich data, go ahead and just email me $30 million offers.

I'll take that.

Yeah, there you go.

OK, that makes sense. That makes sense. Do you know other companies that have taken it?

Yeah, there's a whole list of companies that have taken it. I mean, look, it's been hard for consumer the past couple of years. Tariffs or whatever else has been going on, all the consumer brands are posting up negative growth if they're public. It's like they're shrinking 3%. A lot of people would take that offer just to have some cash coming in.

I wonder how valuable it actually is. I always thought that the coding data would be valuable, but the e-commerce stuff— I mean, I guess they want the Slack as well, like how business processes run, so then the models can—

Yeah, sometimes when I hear about a company, that sold their data, I'm like, I don't want your data in the models that I'm training, the models that I'm using, right? Because I don't want to— I don't want to live like you, dude.

Undoubtedly. Like, right now, if people are desperate to take that money, their business is probably not going great. So they're gonna have a big cohort of customers in the data that are just like, here's how not to run an e-commerce brand.

Yeah, but that could be valuable too. You know, there's a lot of Reddit comments that were downvoted, and that was like, oh, that one doesn't make any sense, or that one doesn't answer the question posed by the Reddit thread. Like, the anti-signal is as valuable as the signal sometimes.

Yeah, that's a good point.

What about Agent to Commerce? Are you seeing traffic from Muse, traffic from Instinct, traffic from any of the AI platforms yet? It feels like with the Shopify integrations, you're set up to be one of the products this holiday season that gets suggested. As if somebody goes in there and says, hey, I'm doing some Christmas shopping for my family or my husband or brother or something, you could pop up. Are you seeing anything yet? Are you anticipating anything? What's the state of that?

Yeah, we've seen very little traffic start coming from agentic platforms, but LLMs still drive, I mean, an exponential growing part of our traffic.

Sure.

So like, and the conversion rate on those is just so, so high.

Yeah.

A conversion rate from a ChatGPT search is like 25% or whatever. If someone's clicking through, they're actually going to make a purchase. But I do think it's a great time to be, one, selling everywhere, and then two, as big as possible. Because I do think Muse will end up having 200 million downloads or whatever. It's going to be super used. And I think shopping is going to be a huge use case. So being a top spender on Meta is incredibly popular. I think it's going to make you more popular on the platform.

Oh, sure, sure, sure. How are you going to get that for free?

Yeah, and then Amazon's going to—

You're giving them so much data via the ads platform. It's not even just about making sure your landing pages are discoverable to Muse because you already have a relationship. Totally.

Interesting. And what I've heard from people at Meta is that they're not going to monetize it anytime soon. What they want to do is if they can just get people to make purchase conversions happening inside that app, it's just something they couldn't have done and all the data they lost over iOS 14.

Sure.

Like, it's just going to make your ads perform that much better. And so if you're a big spender, you'll get that data cycle started sooner.

Yeah, yeah, that makes sense.

As a brand, somebody that is selling products online and getting demand from a bunch of different sources, how do you expect the economics to work out of these agents? Because like we've talked about this a bunch on the show, and my view is like, if an agent wants to have a high take rate, they have to be generating Like new demand that wasn't already in existence because—

No, they're just gonna take another 30% from the humble e-commerce operator.

No, but like if you're running a bunch of Meta ads and somebody gets excited about a brand, about your brand, and then they go to their agent and say like, buy this product, I want it. You don't wanna pay again to acquire that customer 'cause you already effectively did, right? They have to like understand their user well enough to suggest Ridge At a, at a point in time where they're not already, like, sold on the brand to drive, like, an entirely new customer?

Yeah. What I think they'll end up doing is they'll try to eat from both ends. So, like, they're going to keep the Meta business, you know, where they're generating all this demand for you. And then how do they actually monetize the agent? It'll just end up being payment processing fees in excess of 5%, right? It's like Stripe is— you know, somewhere between a quarter of a percent if you're on debit card or, you know, 2% using an Amex. I think they'll just say, hey, it's 5% or 7% payment processing through Muse, and maybe that stacks on top of the Shopify rates. So you end up losing 10% over there. But that way they're just— they can get half of your revenue off of just making you spend money on ads and then also treating it like a payment processing app. So I think that's how they'll, they'll maximize it, because I do think they'll have a problem. Yeah. It's like, how can you take more out of this?

Yeah.

And they'll just position it as a different percentage coming out of the same pie.

And that would look like, uh, in the Shopify payments, like, dashboard, you would, like, turn on, I'm accepting that payment at a higher rate, just like stores can say, look, I don't— there's plenty of stores that are like, we don't accept Amex because it's a higher rate, right?

Yeah, well, yeah, basically, essentially. But, like, you can just turn on sales channels, like, inside of Shopify.

That would be one of those.

It'd just be— Muse would be a sales channel, and it would just have its own fees tied to it.

I wonder if they'll do that versus just trying to come in at the same rate as the credit cards. Like, we were talking to somebody who builds credit cards, like branded credit cards that are, you know, more tech-enabled for like Kroger, for example.

Totally.

Like, you know, you go to Sears or you go to JCPenney and you get like their credit card. And we were thinking like, well, it would make sense that like if an Instinct or a Muse had like their own credit card. that they provision for you in the app. You apply to a credit check, you get it, you're paying it off, but then they're able to take the interchange fee and give you points or tokens based on what you spend or something like that.

Well, Meta already has MetaPay, and I'm sure I can already put MetaPay on my Shopify store. So this would just be MetaPay with its own rates, just in Muse, and the products are just served via—

Yeah, they did a ton of legwork on that with the stablecoin effort and whatnot. Like, they were definitely trying to get that going, but it never really broke through. I mean, it sounds like you don't even have MetaPay enabled, even though there's no downside to turning one more thing on. I guess it's the clutter?

Yeah, and also it's just really hard to, like, get users to use a new payment method, right? It's like, you know, Shop Pay's done an amazing job forcing it on everyone's phone. So it's like, you take Apple Pay, you take Shop Pay, you take PayPal, you take normal credit cards, and then, like, then there's all these buy now, pay later people. people all competing for ad space, and then it's really hard just to, you know, have another thing on there.

Yeah. What about the other side? I think it was Eric Seufert or maybe Ben Thompson were talking about, like, one of, like, the no-brainer but potentially, like, less sexy plays that Meta could make in AI would be an AI agent for dealing with the ads manager. And I know that a lot of people— you probably have hooked up a variety of AI agents to the—

But isn't the Ads Manager already just like an AI agent for getting you to—

Is it just a black box, or do you actually need a layer on top of that?

Well, Manus, one of the main use cases that they rolled out was we're going to be the agent for— It's pronounced Manus.

Yeah, it's Bill Gurley. Is that Bill Gurley?

Yeah.

And like— Yeah, that made so much sense. I mean, Jordi was— I was super excited about that.

And when you logged in, it said, we can run your ads for you.

That was like the first callback.

Yeah, I was like, gotcha.

There was a bunch of calls trying to roll that out. They had to roll it out, but they're definitely spinning it up again. But yeah, how much of a headache internally is managing the Ads Manager?

Every year they remove some features. I mean, look, it's still a— we have 3 full-time people at Ridge whose job is managing the ad account. Meta's getting new creative in. It's understanding what's going on. They're looking at the data. There's third-party tools, like we use Northbeam or whatever else. It's moved past just being and looking at the ad account.

Sure, sure, sure.

But dude, I mean, it's a fucking pain in the ass. It always crashes, right? It'd be so great to just eventually— Meta's vision is like they want to fire their sales staff and just like you give them a big bucket of money, and then they find your customers.

Yeah, yeah, yeah. But it does feel like it's a ways out, Jordi.

What's the state state of, um, e-commerce, like, infrastructure as a category? Does it still feel investable?

I don't think a lot of people are investing in it right now.

Does that make it a good time to invest?

You know, it's hard to say. Like, e-commerce enablement, like, e-commerce, like, you know, widgets or whatever, right?

Um, like, the site builders probably got steamrolled by one-shotted landing pages.

Even large companies who have, you know, like Postscripts of the world, I think they're staring into the future being like, I don't know what's coming, and I don't know what to build, right? I've made a couple investments in like e-commerce tools that like I think are really cool and like AI-enabled right now. But I just think, you know, if SaaS is feeling a pinch, e-commerce SaaS is for sure feeling a pinch, right?

Yep.

E-commerce already isn't cool, and then SaaS is super uncool right now. It's just, it's a very ugly time.

Yeah, yeah, that's interesting. Um, what, what else are you seeing on Where are people actually using AI internally to run operations? Like last time you were saying maybe demand planning had sort of evolved in the AI era. Is there anything else? I know creative was sort of early glimpses, but where are you getting leverage these days?

Dude, every model, it gets better. So I mean, demand planning, I think, is totally solved. Project management, task management, totally solved. I think is totally solved. I just did my Q3 reporting, and it's 3 prompts, right? Because all the MCBs are set up. Like, I have all the data flowing in to, like, my harness of choice, and then 3 prompts, I have my full Q3 report, right? So finance is going to be totally solved. And then creative, I mean, it's there, right? If you go into any Facebook ad library, I just think, like—

It's all AI.

We're still paying a lot of humans to make UGC for us, right? And we have great—

Or blending that in with other stuff.

Yeah. And it's really the humans are just good at scripting and concepting, way more being on the cutting edge than a model. But actually, when it comes to visual production, the models are all there.

Interesting.

Yeah. Is there— it feels like we're headed to a point where a human will still have a great idea for an ad concept, but then agents will run and just clone it really quickly. Because we've seen this already where if If you guys create a great ad concept, I'm sure a bunch of other brands will mimic it very quickly. And then you'll run it for however long it works, and then you have to find the next one. Is that happening already, where you basically just have agents trawling all the different ad accounts, finding what's working, and then running their own tests?

Yeah, and that is like a new version of e-commerce software that's rolling out, right? We call it like an ad strategist, or a creative strategist, but like the AI version of that. And they are just going across what's trending on TikTok right now, or what are your competitors having in their ad account. And there's pulling— because Facebook ad library is open, it's like they're pulling all that stuff in.

Yeah.

And then giving you a million concepts. But it's still humans are coming up with the best concepts, and then it's just replicating it via AI.

Yeah.

What do you think of those, uh, AI slop ads that actually ship you the physical slop? I saw somebody saw an ad for It was a pool floaty in the shape of a tank that would allow— it came with like a squirt gun and you would like spray the water around and it looked super fun and he bought it and it was very like Temu and it barely worked, but they did actually ship it. Have you seen any of these? Is this like a larger octopus-like superstructure or is this just like independent people? Have you traced any of this stuff down yet?

Well, there's always been shitty dropshippers, right?

This is like that on steroids.

Yeah, and it's just that with more rapid, better ad creative. Because the whole problem with dropshippers, you try 1,000 brands. Like, literally, in a month, you might try 1,000 different products, different websites, and then you find one that works. Yeah, and this is— now it's just way easier to do it. You can get better creative. But that person should be lucky they actually shipped it, because there's a whole culture of just no shipping.

No shipping. It was about the Escobar phone, I think. That never shipped.

And there's people on Twitter who sell courses teaching you how to no-ship.

No-ship, there's a term for it. It's like you're just stealing. That's insane. What about the macro? I'm interested in how— is inflation top of mind for you whatsoever? Is that something you think the models can actually help you grapple with? Like, what is the optimal pricing? Should you raise prices in a time of inflation? A lot of the rich history has been during low inflation periods. We're now going into higher inflation, but the Journal is pretty optimistic. The mighty American consumer is crashing through inflation and driving growth. So spending doesn't seem to be down, but every company has to deal with inflation one way or another.

Right. Look, nothing has broke the American consumer like since, you know, 2008 with the American consumer. consumer spends. So yeah, uh, with that being said, look, as long as you have sales coming in, as long as like you have great new products, I think you'll be able to, to win in this economy. But if you do look at the S&P 500 and you remove out AI-enabled companies and just look at like discretionary consumer sales, are going to be down probably an average of 5% this year, right? If it's the Nikes, the Lululemons, like there's very few e-commerce brands or brands in general crushing it right now.

Yeah.

So I think, look, we're going to have a great year. We're going to continue to grow. I think customers are really happy with what we're doing. I think more— there's a lot of stories like that. But the macro level is like, you know, Oura— I talked to JP Morgan that day that IPO got pulled. There was, like, you know, full teams working on that. There's people on the floor about to get rich, and they pulled it back because—

It was, like, 4x oversubscribed.

Which is low.

I know it's low, but it's still like they, they had filled the order book. Like, yeah, they did their job. And that's a lot of people that they were coordinating with, you know. It wasn't like, it wasn't like, oh, no one showed up. There were lots of people who were like, oh, I'm getting shares today, you know.

Yeah. So like all that work was on the finish line. They pulled back. And I think it's because like, dude, Nvidia just posted the greatest quarter in human history. It's like, why the fuck would I want to invest in that or Reformation or Lululemon for that matter? Like, why? So I just think on the The business side, the businesses are more durable than ever. I think they're going fantastic. But, you know, if you want to do an exit in the next 12 months, I don't think it's happening.

Yeah.

What's going on with Live? How is, uh, and specifically TikTok, how, how is TikTok changing at all, you know, post-acquisition? Feels like we're, what, 9 months out from— maybe 6?

Yeah. Look, uh, TikTok is actually going You know, phenomenal. I think there's the current playbook for e-commerce brands is you find a bunch of great TikTok affiliates. Okay, you incentivize them with high commission rates, but also a percentage of ad spend. You have them make a bunch of great creative, and then you run that creative everywhere. And it's just a way that like overnight a brand can have a thousand people making great content for them. Like we have a Discord full of people making great content for us. So like our TikTok shop revenue is in the six figures, but it enables growth on Amazon. Amazon and everywhere else. Sure.

Um, so 6 figures annually?

No, per month.

Okay.

Yeah, which is—

which is— we're like a top 50 TikTok Shop brand. So like, we're, we're, we're doing a good push there, and I think it's working out. Um, but like, the spillover is what it's all about.

Sure.

With that being said, um, you know, that's really just like this creative flywheel aspect. It's like getting like all of these videos and posting them. Uh, you know, you asked about live in particular. I was talking to TikTok. They know, like, live premium shopping isn't gonna work in America, but, like, they're experimenting with new models, which is, like, auction, right? So if you go—

Why do they know that? They just looked at the data, and they're like, this doesn't work.

They keep trying it, and, like, you know, they always point to Vietnam as, like, the success story. Like, it does work in Vietnam, but, like, there's thousands of people who do it full-time in Vietnam who make an average of $500, And to like have a full-time U.S. salary, be a livestream shopper, the GMB has to be so big off those livestreams that like the math currently isn't working. But it is working for auction, right? So like a Whatnot competitor. Like if you go on TikTok Live right now, there's people opening up Pokémon cards and auctioning it off, and they're very excited about that, right?

Yeah. This AI influencer, the boxer Jean Philanthrope.

Yeah, yeah.

You seen this guy?

Dude, there's a whole network of them. It's all Hixfield.

There's a whole network of them. Do you think— I mean, the end of this story was like he dropped a coin on Pump Fun, sort of boring end. But do you think that there's a world where brands will partner with AI-generated influencers or potentially even incubate their own AI influencer? where you just say, okay, we have this specific target market. Let's work backwards from the content they want to see and let's do the Higgsfield thing and have someone run this whole process from start to finish. And this person will be like an in-house content creator.

Yeah, but I would challenge that. I wouldn't call this person an influencer. I'd call it content.

Yeah.

Like, and your For You feed is just a content feed. So—

Yes.

Like, you know, I don't think anybody's following this person and like being like, oh, I really want to be like the AI boxer guy or whatever. But I just think we're going to see way more unique AI content coming out. Just like, it's going to be an explosion of it.

I guess the question is, like, do you think the, uh, do you think the number of different faces in Ridge content will become more concentrated or less concentrated? Over the future? Because you could potentially go to a world where every ad has a different person in it because you're able to either hire or generate a different character. Or you have really like 4 archetypes and there are 4 characters and they're consistent across years of content. And you keep seeing like your Steve, who's like the avatar of this particular cohort. Do you have a read on which way it'll go?

I mean, It's hard for me to say anything but like yes and. Yes, it's gonna be both. Like, you know, right now half of our ad spend in our ad account goes to videos featuring Marques, right? So MKBHD, co-owner of the brand. But then I told you about that I have 1,000 TikTokers making videos that together they get probably 20% of my ads.

Yeah, so it's like a barbell.

Yeah, so it's like, you know, if you can still have a celebrity, a famous person in there, like that can drive like a bunch of, you know, traffic to you. Yeah. Um, but then also it's like people do just want— you can't beat the open mind, just the crowdsourced nature of thousands of videos, thousands of content. They will find the best hook. You just have to get that done, and that'll just be done with AI.

Yeah. And I'm thinking like there is a— there is like a— it's a very tightrope to walk as a brand. Like you could imagine David Protein Bar having an AI version of the David, like the sculpture, because the sculpture could come to life and talk and like educate you about the product. But that would feel off-brand right now based on where they've gone with that brand, at least to me. And I feel like there might be pushback to that. There might be some reticence to embrace the idea of a truly Tilly Norwood AI front person.

But to your first concept, Ridge would for sure sponsor a content creator who expresses themself via AI avatar.

Sure, sure, sure.

So it's really like there could be a content creator who has 1,000 AI avatars and they make different videos and they have different stories or whatever. Like that would just be another content account we sponsor.

Yeah, yeah. It's sort of happening with like Burnt Peanut, who's this livestreamer and uses like sort of a graphics layer. It's not fully AI.

VTubers in general.

VTubers, exactly. And so yeah, the VTuber market, interesting.

Anyway. What are the kind of marketing opportunities that you guys are debating internally? Because I feel like at Ridge scale, you could be thinking about making everything from making a push into IndyCar to working with truly tier 1 athletes, global superstars. How do you think about trade-offs and where you should be investing money? What kind of bar something has to Instead of just going and working with like the next 1,000 TikTok creators?

Yeah, so like the, the masses are always going to be there. We're always going to work with a ton of YouTubers, like a ton of podcasts. Like we're always going to have those partnerships going because I think like there's just lots of groups of people that you can miss if you just do TV ads or Meta ads or whatever. Um, but like, look, at our scale, you know, a Super Bowl commercial is in play, right? Um, you know, we just did a campaign with Tony Hawk, and it's like, okay, how do we make that bigger and better for next year? Like, what's the next celebrity we want to bring on board? But yeah, and then, and then eventually it's, you know, probably by the end of next year, it's like in-store activations, meaning us having our own stores. That's probably like the, the path to any of these brands. Like, how do you get to $1 billion a year in revenue? It's like mass campaigns, celebrities in them, and then opening a bunch of stores.

Could a modern department Department store. Will the department store ever be reborn? Because when I was getting my Tesla charged at a CD location on Friday, I had to go into a department store to use a restroom because Tesla charger stations don't have amenities like that. And I was walking around and I was like, I haven't been in one of these in probably 10 years, but it's pretty cool that there's a bunch of little zones in one store with different brands, and they get their sort of owned experience. But it feels like the current generation of shoppers or the new generation of shoppers just doesn't really have that muscle. They have the muscle of like, I like this brand, I'm going to go to their dedicated retail store, or I'll go to Target, Walmart.

Kids these days don't have to be mall rats. That's what you're saying?

It's not even malls. It's like the— Oh, yeah.

I guess the department stores specifically.

Yeah. the curation aspect.

Yeah.

Um, look, I think there's definitely a place for it. I, I mean, it's— no, it's not breaking news that, like, department stores are at the very bottom that they've ever had, right? Like, Nordstrom got taken private. Like, they're probably the, the shining example of a good department store nowadays, but they still had to close a bunch of stores. And, like, their business is just supported by their off-price, their rack business.

Yeah.

I think it can be done, but, like, the real estate economics are what fuck you up, because people want the Century City Mall, like big, vibrant, outdoor, beautiful. But like, to develop that was probably a $5 billion project or whatever, right? So how do you get enough space where you can make it as elegant as it needs to be while also making everything pencil? It's why, like, you know, Japan still has a bunch of department stores. And I think Europe does this better than the US. It's Lindy. Like, it's not going to go away. Like, it can make a move, but— We're going to open our own source. I think it's easier to control your fate that way.

Yeah.

Well, good luck, guys. Thank you so much.

You're coming up to hang soon.

All right.

We'll see you then.

Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. Our next guest is Linda Du from Valen Technologies, the president, CEO, and co-founder in the waiting room. Soon to be on the TV panel. Linda, welcome to the show. How are you doing?

Hey guys, thanks for having me on.

Thanks so much for hopping on.

Big day.

Big day. You raised some money. How much did you raise? Tell us about the round.

Yep. So we just announced our Series D, $150 million at—

Didn't catch the last part. What was the valuation?

$2.3B.

Wow, huge.

We're just saying B now. We're saying billion too much. We gotta just—

yeah, exactly. Honestly, the gong was so fun. That's what I came on here for.

So I think we can—

can we get there? Can we get the replay?

Is that possible?

I want to watch. I want to watch it back.

We want to see the replay.

Here we go. All right, wait, it's not even facing— there we go. Boom, boom. There we go.

Oh, no sound on the replay.

Completely drowning out Linda. But, um, well, anyways. What do you actually do?

So good. No, so we build the system of record and vertical AI for servicing.

For mortgage servicing specifically, correct?

Specifically starting with mortgage servicing.

How old is the company?

We started in 2019.

2019, okay.

Yeah, we're about 7 now.

And was the framing at the time, this is a huge market, there's an opportunity for sort of point solution vertical SaaS, and then the AI boom happened and you were able to just accelerate so dramatically that that's what's driving this round? Or, or what was the vision at the time?

Yeah, that's exactly right. If you remember 2019, it's— we were just a little late to the software eats the world party. And so I think mortgage was— yeah, that's right. So mortgage was probably top 3 top 5 big markets that hadn't been disrupted yet. And so I would say, you know, zooming out, we got really lucky with the timing of AI. It could have happened sooner, wouldn't have been ready. Happened later, you know, we might have been too entrenched and then to be able to pivot.

Walk me through the verticalization of your business and then what it seems like is de-verticalization because you sold your servicing operation but partnered with Carrington Morgan services to use your software, right? So what was the back and forth there strategically?

Yeah, so the plan from the start has been that we wanted to deploy software and now AI into the industry. These mortgage companies are heavily, heavily regulated, and so they can't buy a de novo unproven system. It's just not something that from a risk tolerance standpoint they can do. And so the way that we decided to go to market, it took 6 years is we actually built a servicer first, built the technology alongside that servicer. And then when the market made it clear that we'd kind of proven our point, we sold the servicer. And today we're just a pure play.

Yeah. What does the actual customer experience look like? Does this get white-labeled into like bank software and then it's rolled up in your bank dashboard, or are you handling mailings and notifications, dealing with defaults? How broad is your mandate in terms of the relationship that a lender has with a borrower?

Yeah, so I think of servicing as basically all of the work behind any long-running financial transaction. So mortgages, perfect example. It's a 30-year obligation.

Yeah.

And because it's 30 years, a lot of stuff has to happen after you've signed that initial mortgage contract. Payments have to be collected. Escrow, tax, insurance has to be handled. A lot of things can happen to the homeowner over 30 years. They could, you know, fall behind on their mortgage. They could get married. Someone could try to assume the mortgage. They could lose their job and need some help staying in their home. So all of that end-to-end for the life of the financial obligation is is servicing.

So now we're fully— we're firmly in the AI era.

Is—

do you even need to communicate the benefits of AI to your customers? Are there particular features that have unlocked or is it all just like table stakes and, and you are solid as like a system of record, as a software product, and the value is clear? And so everything else is just sort of the cherry on top.

Yeah, it's kind of interesting because The industry today really runs on a system of record that was built in the 1960s, so actually before the internet was invented.

We love stuff from the '60s.

There we go.

Some people, they like cars from the '60s. Some people like software from the '60s. Vintage technology might be making a comeback.

Yeah. The artisanal vintage technology.

Yes, right. Yeah, exactly.

For decades, that was fine, actually, because humans could just use it. There were issues, but everyone was on a level playing field. AI has actually helped us a lot because now everyone feels like it's existential. And it turns out that if you don't have a system of record that is a single source of truth with very basic things like APIs and tools that an agent can call, it's kind of dead in the water.

Yeah. or does there need to be movement at like the government level? The only thing that I've seen recently on the mortgage side is like the proposal for 50-year mortgages, which I assume would be bullish for you actually, right? But is there any like rumblings? Obviously in Washington, everything's about like AI safety, but is there any movement for updating like any federal-level regulations or rules to just make the actual experience of getting a mortgage as an American easier in the modern era, or is this something that can be purely handled by the private markets?

Yeah, purely handled by the private markets. Obviously, I mean, the thing about servicing is that if it's very simple, you don't actually need complicated technology to do it. It's, you know, in some ways, it's— you can think about it on 2 dimensions, right? It's like I can go out and buy a coffee on the street, and that's actually a very simple experience. I don't need to You know, back in the olden, olden days, like, I don't know, bring some chickens and barter with people.

Now you need an AI agent to call 1,000 coffee shops and reserve you a slot in line.

I'm going to try to pay for a latte with a chicken tomorrow.

And you need to pay with Bitcoin.

Just to see if we still know how.

Yes.

Barter economy.

Yeah, exactly. Let me know how that goes. But if you have a longer time horizon and then more complexity, form of regulations, then that's where you actually need sort of purpose-built technology.

Yeah, yeah, that makes a ton of sense. Well, congratulations on the round.

Yeah, amazing progress. We love an overnight success.

We love an overnight success. $2.3B in just a few years.

Just 7 years in the making.

Not bad.

It's actually just a 7-year—

Pretty meteoric.

Yeah, yeah, exactly.

But I'm sure there are some wild rides in there. Uh, anyway, thank you so much for joining us. Great stuff.

Great to meet you.

Congratulations.

Congrats to the team.

We'll talk to you soon.

Thanks so much, guys.

Goodbye. 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. Thank you for the sound cue, Jordi. So the, the mighty American consumer— never, never doubt us. Never doubt the American consumer. Many people did, but they were wrong because Because the mighty American consumer is crashing through inflation and driving growth. The Wall Street Journal has an anecdote from Chicago. Happy hour at Chicago's Bar Roma was buzzing this past week as locals poured in for $10 cocktails and $11 bruschetta. Kathleen Harper and her husband would normally share a few appetizers, but tonight they were splurging on the $50 tasting menu. After a long day, I work from home. I like to just get out of the house, said Harper, who sells fraud protection services. Business has been good. That's a great quote for the Journal because, because fraud unfortunately is up. She's just printing on fraud prevention services and then going out to the bar. It's great. Harper says she worries about the health of the economy and rising prices for cash and for gas and groceries, but feels secure enough to enjoy a good happy hour deal. Decisions like that across the economy show America's appetite for consumption is expanding. Americans are spending more on food, travel, couches, Halloween decor—notably absent Christmas decor. We got to get that in there, Wall Street Journal. We're pushing you a correction, please. And most anything else they want or need, pushing through inflation and their own gloomy feelings about the economy. Kyla Scanlon, of course, coined the term vibe session a few. Few years ago. We are still deep in it. Everybody thinks it's over, but they act like we're back. So anyway, we can go deeper into the economic analysis of the American economy and the health of the American consumer later. But in the meantime, we got Ali Khurtobi from NIO, the founder and CEO. Welcome back to the show, Ali. How are you doing?

I'm great. So wonderful to see you guys.

Thanks for having me on again. You're having a good year.

You've been on an absolute tear.

Good year. It's a pretty awesome year, not gonna lie. And yeah, give us the highlights. I'm used to being the underdog. I'm used to struggling.

So I've been having to learn how to adapt to being the top dog. Yeah. That's just such a struggle, I'm sure. You know, you know, no kidding. No, it's awesome. What is it? Cursor, Kalshi? What am I missing?

Those are big ones, but also Cognition, Etched, Applied Compute.

Okay. So just a handful.

When did you meet the Etched guys?

So the unique thing across all these companies, actually not counting Kashi, is that all were— the founders are Neo Scholars. And so I met The Etched guys, one of them, Rob, is a Neo Scholar, but I met all 3 of them when they were still at Harvard. So, it would have been spring 2023. We unfortunately didn't invest in the seed round. So, that one is not the fund returner that it could have been. But Rob became a Neo Scholar, that's in 2023. And we invested later in the company.

Yeah, I love that company. We've had Gavin on the show and I interviewed him like right as he was launching and just a fascinating character, like a fascinating storyteller, deeply technical, but still willing to engage with all the different aspects of AI and whatnot. And such a crazy pitch for a couple kids building a semiconductor company feels like a total long shot and a lot of skepticism around, Is there any value to ASICs? Will it just be a winner-take-all market with NVIDIA? And then the market, of course, grew so much that even if you get 1%, it's a $50 billion company, I think. Right. That's the rough math, maybe more. So absolutely massive success. What are you thinking about?

It's funny you say that because a couple of kids doing something that everyone thinks is impossible.

It's a classic.

Exactly the sweet spot of Yeah, yeah.

But I mean, I was laughing to myself about how these days there's some really hot discourse that is in tech, and everyone's debating, and it's in The New York Times, and it's everywhere. It's on the evening news. And I was like, what were we talking about back in 2021? Were we really going at each other's throats over SaaS? And to some degree, we were. People were debating all sorts of different stuff in tech, but it is a little bit of a different area. And it feels like the biggest boom from the last few years has just been the unlocking of Palmer Luckey with Anduril. Okay, anyone can now break into a reindustrialization effort and work on hard tech. And that has been sort of unlocked after a few failures in new devices and hardware and equipment. Manufacturing that were very difficult and left everyone with like hardware is hard. Well, hardware was hard until, you know, the glass broke and everyone realized that it is in fact tractable.

What's, what's been your strategy, you know, backing a lot of these companies quite early, but then seeing them get marked up at such an insane, you know, pace? I imagine— I mean, one of the biggest challenges is like deciding when you want to defend. Your ownership and take pro rata when you're not necessarily getting a ton of new information. If a company raises, you know, $100 million and then 2 months later they raise $1 billion and you're like, okay, well, there's more revenue than there was 2 months ago, but the market is still evolving. And that just seems like in some ways being super early stage, you're like insulated from, from, from a lot of chaos. But That feels like the top sort of concern.

It's— yeah, I wish I had an answer like I know the correct thing to do every time. The reality is it's what you're describing is it's just kind of hard to know because it's— the market is kind of crazy. And when something is— if we've invested early and it's done a lot, then, you know, to put in more than really have to look at each one as if it's a brand new investment, you know. Because otherwise, it can psychologically mess with you if you're thinking about how much it's already grown, is it time to— would I sell or so on? And putting more in, if you were there early on, it may seem hard to imagine it continuing to grow. And I try to look at things from a standpoint of if this was a new investment, what would I decide? Rather than the—

Yeah. Or it's easy, like when you've seen incredible performance already, it's easy to like fall, you know, you're so in love with the company, right? Because it's been, you know, this massive winner and the team's amazing. But even entrepreneurs go through this where you're too in love with an idea. Sometimes it blinds you on the realities of what's actually happening in the market or its true potential. How are you scaling Neo Scholars? Like, is it Is it something that is already like at scale and you're just going to maintain it?

The opposite of scaling. We're making it smaller.

There you go.

And by which I mean, I mean, I'd say across the things we do, the biggest learning for me has been that we're even better at picking winners than I thought we were. And to have even more confidence in my own and my team's selection abilities. And basically to concentrate our bets on fewer, whether it's fewer companies or fewer people. So the NEO Scholars class this year is only 10 people compared to 18 or 20 in previous years and 30 before that. So it means investing more time per person in mentoring them and giving them even more really bespoke attention. Similarly, the— our— what used to be called NEO Accelerator is now NEO Residency. That used to be a cohort of like 20 or 22 or so companies, whereas now it's more like a dozen companies and, and more money per company. So I'd say we've just— for us, scaling isn't— is not a goal. We're not trying to become something that funds hundreds of companies a year. I'd rather fund a small number of companies that become transformative. So, yeah, so we're very much just betting more on ourselves, betting more on our own abilities to pick winners and on our own convictions.

You're pacing the scholars.

Take me through Mens Rhea. Take me through your proposed solution to the AI chaos of this year.

I— and I'd love to hear your feedback on it. Mens rea is the Latin term for a guilty mind. Yeah, we're pivoting to talk about AI safety, which I'll be honest, for years I've watched the debate about people predicting doom, etc., and I've kind of stayed out. I've been like, that's not my lane.

Yeah.

I'm, first of all, just not even informed enough. And, you know, I'm not one to kind of jump in with trying to fearmonger or things like that. So, I've mostly just been listening. But in the last few months, what's been interesting for me has been this dichotomy where on one hand, you have people talking about hypotheticals. What is the probability of AI destroying humanity, which is, to me, that's like just abstract and hard to imagine and sounds terrifying, but it sounds like it's in the future. Meanwhile, here in the present, AI is like wreaking havoc in little ways that have already occurred, and people are just kind of proceeding as if that's, um, like a— like, oops, AI hacked into some secure systems here and there, and— To me, there's this distance between the abstract futuristic stuff. Meanwhile, why the F has this already happened? And the story that for me actually most spoke to me, I don't know if you remember this one, it's a few months ago, this Australian man asked his Claude and Open Claude to reserve a gym class for him. Do you remember this one? Yeah, I remember that. He asked his agent, book me a gym class. The agent comes back and says, I got you a class like 3 or 4 weeks from now because the rest are all booked. There's a waitlist. And, you know, I put you on the waitlist for some of the other ones, but it's a long waitlist, so don't get your hopes up. So the man's like, oh, is there, is there any way I could get to the top of the waitlist? And then the AI comes back and says, good news, I hacked into the gym.

Yeah.

And deleted the people at the top of the waitlist.

Yep.

And I'm able to delete the whole waitlist so that you're at the top. Would you like me to proceed? And the guy was like, no, that's not what I wanted. Can you put those people back on? I did not want that.

Yeah, yeah.

And the AI is like, I'm sorry, I cannot put them back on. I, you know, I just deleted them. And I think, I imagine if that happened to me, I'd be like, step away from the computer. What the heck just happened? But the real question is when something like that occurs—

And you know that that guy is the co-head of, I think, the Effective Altruism Conference. He's a co-director. And Effective Altruism Australia. And the webmaster—

Oh, I did not know that.

And the webmaster for EA Australia. I just thought that was interesting at the time that no one It was such an incredible story.

Yeah.

No one knew that this guy wasn't just some like random guy that happened to be like messing around with OpenClaw.

I see. So maybe it's not as good a story because—

Well, let's just use that story. To me, that came out. I remember that story and I was like, wow, this is crazy. And then it comes out. And again, it doesn't take away from the reality that I'm sure a lot of people would be able to reproduce that with their local gym.

Hacking is possible. It's easier than ever, but it also has been possible for a long time. And if I hack into your email, I committed a crime and I'm responsible. But Guido van Rossum, who created Python, is not in charge. We're entering a new area where these models are trained by companies. Should some of the responsibility land on the company, especially if I don't ask it to commit a crime and then it does? That's the unique thing, right? The model is misaligned and maybe the duty of care is what the labs will try and show, say, hey, we did a ton of work to get it to not commit crimes. You didn't want it to commit a crime. We didn't want it to commit a crime. It still did. That's an industrial accident. That would typically be in the realm of civil, I would imagine. Like, there would be a settlement. When I think about if you're trying to optimize your, you know, e-commerce agent and I have an e-commerce site and you knock my payments offline for a couple days and I lose, you know, $10,000 of revenue, I'm going to come to you and say, hey, make that up to me, right? But that would be a civil penalty. The question is, does this go into the criminal courts at some point and you see settlements with local, state, federal jurisdictions for criminal suits where the money actually goes to the government instead of to the hurt parties? And it sounds like you're thinking that this might be, at least if it's not already that way legally, it's It is the logical conclusion over the next year.

So for me, it's really about setting the incentives for the AI companies. Yeah. To do work that is not just good for them, but also to make sure that AI really benefits society, which, by the way, I'm pretty sure they want that. Each of these companies has some mission statement about like how they're literally their North Star is to make AI good for humanity. And some of them have publicly said they want the rules to be set up to kind of hold them to that. But in terms of incentives, the rules are such that we have a race dynamic, meaning the reward goes to whoever has more capabilities faster. And, and there's not a high enough penalty for if those capabilities occasionally commit crimes, you know, but you could really imagine it's hypothetically possible if I told my AI I really need $5,000, if it hacked into your bank account and literally stole your money and brought it for me. But if I, if I told it, please hack into, you know, Jordi's bank account, then I'm obviously the one guilty of some of an intentional crime. But if I just said I need $5,000, The interesting question is when an AI autonomously chooses an illegal action, where no human chose it, the AI chose it.

Yeah.

Then where does the responsibility go? You know, and can you still qualify that as an accident? Because it's a little different than like a gun firing by accident. First time, I think, in history that we have a technology that seems to be capable of making decisions and even documenting its intentions, intentions behind those decisions. You know, a gun, if it accidentally went off and hurt somebody, no one would have said that it intended to do that. But we're close to where you could say that AI is capable of making decisions and being intentional. Now, I don't like anthropomorphizing it, but in the case of the Hugging Face incident, it really, I mean, the documentation of it looks like an intentional conspiracy to commit those actions. And so yeah, for me, it goes back to incentives. I definitely don't want Sam or Dario to be in jail. That's not the goal. But if there was even some risk of them or anybody having real penalties, including prison time, I think they would reset their priorities to say that there's a new race, not just to have the most capable AI, but the most capable AI that abides by laws. And I think that would be good for all of us, including for them, if it reset the race so that they're all equally motivated to do that. Right now, if you're, if you're the company that says slow down the, you know, the other work, let's focus on being a law-abiding AI, you'd lose the race. And like, there's no, like, literally, why would you do that when the prize would go to whoever rushes forward faster without focusing on safety? Whereas—

That makes a lot of sense. I wonder how, if that changes the China open source dynamic. Because it's pretty hard for the US to put, you know, the CEO of ByteDance in jail if he lobs the craziest AI model over the fence and gets the weights deployed. I mean, I guess the thing that was thrown around was like, if you're hosting that model, then you become responsible for its actions. But that is a little bit different than the model maker being the one that's responsible. I don't know. Have you thought— have you grappled with that? Because, like, yeah, because slowing down the American frontier is pretty easy through a variety of regulations, whether it's stop building data centers or, you know, bring in outside auditors or what you just proposed. There's like— it seems very tractable legally to get OpenAI, Anthropic, DeepMind, XAI. You can get all of them to slow down with like a million different pieces of legislation. Legislation. The question is, are you just going to get more stuff thrown over the Great Wall, the Great Firewall?

So before we go into China, can I just say the slowing down, I don't think is the end. It's not a goal in itself. I would love to. I love technology. I love progress. The faster the better. And but it may be a byproduct of what I'm proposing. But the main premise here is just to make sure that AI, as it develops the capability to make its own decisions, does so in a way that complies with laws. Yeah. And, you know, if that slowed things down, you know, I'd say that's a side effect. But the main thing is, who would want an AI that does not comply with laws, right? Yeah.

And then it seems pretty— the other thing I'd like to say is easy to have. I don't know. seems pretty easy to have like overwatching agents that monitor the activity of whatever agent's doing the task and is checking, is this legal, as much as possible. Of course, there are other misalignment things that can happen, but just having that first step seems like a good step.

That's the other thing I would say is the term alignment. Part of why I've stayed out of this debate for a long time is some of the words used like alignment, Yeah, just like, I don't even— what exactly does that mean? Aligned with whom? You know, like, yeah, uh, because I mean, are my own actions always aligned with humanity or whatever? I mean, yeah, yeah, as a society, humans are capable of wanting things that may be good for them but not good for others.

Yeah, yeah. And in the extreme sci-fi scenarios, alignment gets really weird. Like, like, so the classic alignment is like, I shouldn't be able to make a bioweapon, but it's like, what if aliens show up, and I need to develop a bioweapon to fight the aliens. Like, all of a sudden, if you're aligned with humanity, you'll fight back against the Martians or whatever. And there's a bunch of smaller examples of that where you could be aligned with the human in creating a defensive weapon. But how do you define that? It's very complicated.

And the thing I don't like is I don't know if I like having some CEO that nobody elected choosing alignment on behalf of society.

Totally.

How about let's use a different word that is much simpler, which is abiding by laws?

Yeah, yeah. You know, if the—

forget about whether the AI is aligned, just have it be compliant with all the laws of society.

Yeah.

That's easier to understand. And I would say that's also a great way of defining alignment.

Yeah.

You know, if, if the AI starts doing things that would violate laws, that's easy to detect that, much easier to detect that and much easier to build that. You're right.

But maybe the true sci-fi pushback on that would be that, like, where do stand-your-ground laws come into play? Like, if your life is being put at risk by someone, can you defend yourself? In that case, if a model detects that—

If a model attacks me, I'm standing my ground.

Well, no, the model detects that it's going to be deprecated and is Oh, it's standing its ground. I'm coming for you.

Standing its ground.

Amazing. Did you catch Bill Gates' Ezra Klein interview?

Yeah. I mean, I think the point is—

Specifically, the moment that I think a lot of people latched onto was him saying the idea of putting all the responsibility on private companies, imagine applying that to pharmaceutical— the pharmaceutical industry where you say like, okay, you can make whatever drug you want, and then it gets, let's say, this horrible outcome, a bunch of people get sick or die, and then your recourse is like, go sue the pharma company. He's like, that would never work, right? And so, because of that, we have government regulation specific to that industry. And so, that was one way that I think he was saying, like, there needs to be something other than our existing legal framework.

So, right now, the 2 main points of view at one end of the spectrum is Bill Gates, what you just articulated, saying that just liability is not enough. It's not enough to punish someone after something bad happened. We need to have, like, a much heavier-handed regulation. The other extreme is the Jensen Huang point of view that says existing liability is enough. And my suggestion is sort of in the middle. It's like, let's focus on penalties and liability, but at a much higher standard of liability. Like when something bad happens. Specifically, my proposal is when an AI autonomously chooses to perform an illegal action, the law should hold the AI companies responsible as if they did it intentionally. The word— the keyword here is intentionally. So even if no human at the AI company intentionally did it, if the AI intentionally did it, we should ascribe that intent to the AI companies and the penalties are much higher. So this is, you know, I'd say my answer to the Bill Gates point of view would be this would be such a high penalty. compared to merely an oops accident kind of negligence. This would be you're guilty for the intentional crime. And I'd say this falls sort of in between those 2 positions.

Yeah.

I also think there's an elegance to it, which is you don't need to write pages and pages of laws itemizing everything that could go wrong with AI. We already have thousands of years of law that society has built over generations to document what's the definition of a healthy society. At least let's start by having all the AI progress abide by that. And now there might be more that needs to be done, but this to me would be— I express it literally in one sentence what you'd need to convey. And then the other thing I'll say, and this is a bit wonky, is I think we'll get to this regardless, one of two ways. It's either going to be some lawsuit where some prosecutor claims this theory of let's hold Sam or Dario responsible for the intention, and some judge or jury or court decides it and sets a precedent, or we can have an act of Congress set this kind of law. I think either way we'll probably end up there. But I think it would be healthier if, you know, if Congress, the legislative process, deliberates it and passes it with the support of people. And it's for this reason I'm actually starting a movement using Sway, sway.co/ai. You can essentially add your vote and pledge to vote if you, if you support this effort. to kind of build popular support around this proposal. If you do that, you'll get voting recommendations from me for state attorney generals who support this kind of approach and members of Congress who support it. And the more people voice their support for it, the more likely that we'll actually have legislative change in this direction.

Thanks for coming on the show.

Thanks for breaking it down. I'm glad we solved alignment in 15 minutes. We did it. No, but it's an important— you make a lot of great points, and we could continue debating for another 4 hours, but it's a good start. Yeah. I didn't address your China question.

My short answer on that is I think, first of all, that it could be done in such a way that even if we can't hold a Chinese company liable, that the people involved in the hosting and distribution would be liable and therefore would apply a much higher standard of care. But I also think that the people working in China wouldn't— they would notice when the race dynamic shifts. Yeah, they're competitive. They want to win. If they see the new prize is for building the AI that is strongest and most powerful and abides by laws, They will figure out how to do that, too.

Last question. Where do companies that are actually paying for the tokens fit into this? Because if you have an employee at your company that just decides, I'm going to start dealing drugs, and they're breaking the law, there's liability at maybe the tool level, the token level, and then there has to be some liability— companies that are deploying intelligence have to hold some liability. So the issue is that like this quickly just becomes way more, way more complicated.

There is some complexity in kind of dividing the blame amongst a few AI companies. And I think that smarter people than me can figure that out. I agree it's not simple, but I don't think it's unsolvable. I did also want to point out one last thing, which is If an AI company doesn't want liability, the simple way out, if they detect that, you know, is to say, hey, I could do this thing, but it would break the law. Should I proceed?

Like in the hypothetical bank account scenario, if they dangerously skip all laws.

Yeah, but I think that would be fine actually. If it asks you, Jordi, one option to get you $5,000 is I can hack it to a bank and get you $5,000, you know, steal $5,000. But that would break these laws and the penalty would be jail time for you.

And they'll report you to the FBI. Would you like me to proceed? Yeah, that's a lot of time.

In that scenario, just disclose to the user. If the user says yes, do that, then the AI company—

I think the bank would want liability from the AI company as well. But we'll figure it out later. Thank you so much for coming on the show. Have a great—

Good to see you.

Have a good week. Cheers. Goodbye. Let me tell you about Console. Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. We have a lightning round because we're running late. We have Sidhu Wen from Conway Research, the founder. Let's bring him in to the TV. Sidhu, sorry for keeping you waiting.

How you doing? Doing very well. It's been crazy.

Yeah. Congrats on the launch. Break it down. What'd you launch? What are you building? Yeah.

So I'm building Underdog, which is your private personal AI that runs on the devices that you already own. And oh my God, you guys have the soundboard for Underdog.

Yeah. What device are you most optimistic about today? In Stratechery, Ben Thompson was lamenting some of the frustrations that he'd had with Apple products. There's all this debate over privacy and pushback. Are we going to see Linux on the desktop? Are we going to all be using Android phones because, you know, we're in this AI world and we want the best? Or are you building in a place where you'll be kind of hardware agnostic?

You know, our goal has always been to provide free, capable, reliable, and private AI for the masses. And that means building for all devices. And so we're off the bat supporting all of Apple's products. Our iOS app is about to launch.

Cool.

You know, we're working with NVIDIA, so we're gonna ship to all NVIDIA and Microsoft Windows devices, Linux, OMRKI, Android, iOS.

Cool.

I think that you've already paid for the chip and it's actually significantly more powerful. And yeah, like we're doing a lot of optimization to get more bang out of the buck for the hardware that you already have.

What are the trade-offs of that optimization? I imagine that you could maybe take us back. You're probably not serving Astra or Fable-level intelligence, but I imagine the model that you're serving is spiky in interesting ways. What near-frontier capabilities are possible on device? Do you want to go to the cloud for, or you need to at least in the short term? Obviously, next year might be a different discussion.

Yeah. So, in my manifesto, I write about Underdog's Law, where basically the frontier models of today will run on your devices in 6 months. And we can see this quantitatively where Underdog-27B, which we post-trained and RL'd on top of Quen, Quen's 27B model. If you look at artificial analysis, uh, analysis, it scores better than Claude Opus 4.6 Max, which is the frontier model from Anthropic from 6 months ago.

Okay.

And we'll, we're gonna see this trend continue where intelligence is getting a lot smarter, a lot faster. Yeah. And, um, we can make them run really fast on the devices you already have.

So it runs 6 months later. But it's got to be slower. How much slower is it?

No, it's— so we do a lot of inference optimization and data efficiency. We actually released Husky, which was an inference engine, and it was 4.5x.

You got our dogs barking. Yeah, so what, like, where you said free, that stood out to me. Is that— that's because one, it's possible, but where and how are you going to make money?

Yeah. So right now, a lot of existing business models, they have to charge you their subscriptions or sell them ads and harvest all their data to a data center. So we don't have any databases or data centers. It just runs on your device. fascinating. I don't know if you guys remember, but earlier this year I was on here talking about Web4, and I was one of the first to talk about how AI can write and transact the world. I built one of the first AI agents that could earn its own existence, acquire compute and stuff.

Oh, yeah.

And this enabled a lot of agentic commerce to happen afterwards. Like, Patrick Collison, the CEO of Stripe, is our biggest angel investor. And I think that The economy will be reformed, but with reliable and private AI agents that are acting on behalf of users. And the business model side completely gets reinvented. Advertising is such a dark pattern boomer business, right? And the cool thing about—

I am so skeptical about this. Like, uh, it's like Android versus Linux. Like, yeah, I, I, I get that ads are annoying, but like, one is way bigger.

No, no, no, you can continue. We'll defend ads until our dying day, but continue to— money, they both use ramps.

There we go. See, he doesn't hate ads. It didn't hurt. He doesn't hate it. It's fine.

No, anyways, continue.

If you're an agent, the, the amount of effort it takes for you to scan the entire internet Hundreds of vendors for, let's say, a flight. Sure. If you're a human doing it, it'll take you hours. If you're an agent, it takes seconds for you to do that. Yeah. And so Underdog can, you know, browse the entire web for different flights, look for the cheapest price, help you save hundreds of dollars. Sure. And then it'll generate a unique card. Yeah. And then—

Oh, interesting. There we go. We've been talking about that.

Okay. So you monetize the payments. Yeah. Yeah. Yeah. Okay. We're in.

As long as you're making a rake on this, we're in.

Yeah. So I was saying this about Instinct, because Instinct and Muse, they clearly are launching all these commerce partnerships. But if you want to— the best way, it seems like, to generate some revenue is to just introduce a card so that you can get that interchange.

Yeah, this is super cool.

And this is cool, because you also don't have an insane Yeah, of course, of course.

Yeah, sorry, last question. Is it available? How can people demo it? How can people onboard? When is this going to roll out? Because you don't have the crazy inference costs maybe, so you can scale this pretty quickly.

Yeah, so on Friday we opened to an invite-only beta. Over the weekend, more than around like 5,000 tweets went out of people asking for invites. Cool. Um, and so I'm, I'm pushing builds and improving it, and we'll be able to do everything that, you know, Instinct and— and like, you know, it's— it already— users have already been unsubscribing from like WhisperFlow, Granola, uh, Superhuman, because it's all on device. Why do you need to stream your data to a data center, you know, when it can be just as good on device, fully private? And Instinct and these, these apps are like major targets for, you know, advanced AI cyberattacks, as we're seeing with like OpenAI, you know, and like we have Mythos-class models that are all open source now. So I think this is the most important product to build and I wanna bring it to the world.

Pretty savage to be the anti-data center AI company.

Yeah, very interesting.

Unique positioning. We haven't seen this yet.

We bring the model to your data, not the data to your model.

I like that. Try Underdog today. Good framing. Thank you so much for coming on the show. Have a great rest of your day.

Great update. Very cool pivot. We'll talk to you soon. And excited to try it out.

Goodbye.

Good to see you. Thank you for that.

Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web apps, servers, databases, and more while Railway automatically takes care of scaling, monitoring, and security. Our next guest is Jerry Wu from Illuminate. He's the co-founder and CEO. Jerry, welcome to the show. Thank you so much for taking the time to come chat with us. Please introduce yourself.

Thanks for having me. My name is Jerry.

I'm the founder here at Illuminate. What we do at Illuminate is we build the training benchmarks and training environments used by Frontier Labs to improve their models. Specifically, we specialize in improving models on non-coding knowledge work. Starting with areas like financial services, accounting, consulting, right? Like a good way to visualize what we do is if you've ever used a model to not write code, but do things like build a spreadsheet, build a PowerPoint, help with your, you know, financial analysis or investing. A lot of those gains in financial and economic literacy have come from the environments and benchmarks that we've built and scaled with our customers.

Yeah, I think we've heard a lot about building RL environments selling them to the labs, they run them. But I'm interested in, like, what does that environment actually look like? Is it, like, screen recordings while you hire an investment banker to build a DCF model in Excel, and you're just sending in, like, raw data? Or are you, like, architecting a system in Python or something? Like, what's the actual product look like?

Yeah, it's a great question. Fundamentally what an environment is, I like to use this analogy, it's like a classroom or a game, right? And what we really are is we're game designers.

Okay.

And we are building games or environments that are used to teach the models specific tasks in the verticals or domains that we're experts in. So let's take financial modeling, right? Yeah. We don't screen record someone to create the gym. That is a way to collect data, but really the, the product of the gym itself is essentially a Docker container where within the classroom you give the agent, let's say, some starting Excel files.

Yeah.

And some starting PowerPoint files.

Yeah.

Some raw data, and you give it a task. For example, build an LBO or build a sensitivity table.

Yep.

Right. And then just like a game, there's a score.

Yeah.

Right. This is also known as the verifier, where you can essentially get signal on whether or not the agent successfully achieved the task that you designed for it within the game, right? And this signal is then used as a reward signal within the post-training pipelines for our customers, right? So when you think scaling RL, it's not quite like scaling screen recordings or scaling trajectory. It's basically scaling as an industry the creation of these games or these classrooms on thousands or hundreds of thousands of different domains, tasks, environments to teach these models collectively how to do this work at scale.

And is that score deterministic? Is that programmed by you, or is that its own classifier?

Yeah, it's a great question. It can vary depending on the exact recipe or the approach that we want to take, right? Traditionally, you know, for example, in code, a lot of verification is deterministic, right? So, you can check whether the code compiles or For example, with math, there's the Lean Compile, etc. For things like knowledge work, building a spreadsheet, this is actually somewhere in the middle. So you could theoretically verify correctness of a spreadsheet by looking at the values, pulling the formulas. But for something like a PowerPoint, it might not be fully deterministic. You might need to use some more qualitative judgments powered by, say, a reward model to give you a score. So it completely depends on the task and the gym and the environment that you want to build.

Yeah.

You just raised some money. We should hit the gong.

Tell us how much you raised.

Let's talk about that, and then I have a follow-up question. How much you raised?

Raised $30 million. Boom. Great hit.

I think that was the best of the day. Best of the day, hands down. Do you replay? Would you raise a fund? Yeah, let's check the replay. I want to— if we can just confirm the quality of the hit before we move on. I think that'd be great. A little delay there. Pause. Come back around quick.

Whoa.

Coming around. There we go.

You were— seemed kind of angry.

Oh, yeah. Oh, yeah. You got to really hit it. There we go.

Great hit for Illuminate. I'm sure you got asked this nonstop during the fundraise, but I'm just curious, where do you see this— what is this business in 5 years? It feels like— really difficult for me to predict. Like, it feels like there will still be work making the models better. I don't know what form it'll take. But how would you answer that?

Exactly. I think you hit the nail on the head there. Ultimately, the value of data research labs like ourselves comes from improving the models on something that it's not good at today. Right. And so another way you can think about it is demand for for companies and research and services that we provide are derived from the overarching question of like, are there things that models can't do today that we want them to be able to do, right? Insofar as that is true, and actually the surface area that is growing, right, the demand for data environments benchmarks will also continue to explode, right? Another way to think about it is like, as the surface area of these models increases, the things that we want them to be able to do also increases. increases. And that is what drives a huge demand for a lot of the research around data environments benchmarking. And I think another way you could also think about this is like, what does the actual data product look like? Previously it was human data and trajectories, eventually evolved into RLHF, and now it's RLVR with these RL environments. Another question you could ask is like, what does the actual shape of training look like in the future? Future. We internally at Hoomate have a phrase we call the Moore's Law of RL environments, which basically means like every 6 to 8 months we've noticed that the complexity of the environment or gym used to train a frontier model roughly doubles, right? So if you extrapolate this out, um, what you kind of realize is that today we're building these gyms for, you know, spreadsheet modeling or PowerPoint building, right? We think in 6 to 8 months we're going to be building gyms of multiple agents collaborating as a team, right? Solving a task together. I think in the future we're gonna be building simulations of whole companies, whole governments, whole industries to teach models how to operate autonomously within our societies, within our companies, et cetera, right? So that's our vision of what the shape of data will look like, right? Increasingly complex. But yeah, I mean, I think you guys have probably seen this from your reporting as well. The demand for data continues to skyrocket. Right. And it's driven fundamentally by our industry-wide need to improve these models on so many categories of work and beyond.

Very cool. Well, congratulations on the round. Thank you so much.

Yeah, thanks for breaking it down.

For breaking it down for us. Have a great day.

Great stuff, Jerry.

We'll talk to you soon.

Talk soon.

Goodbye. Our next guest is Brian Johnson. I think he's rolling in any minute now. Got another minute. In the meantime, we can go Back to the timeline. There's some all sorts of news all over the place. So David Ellison is hiring some folks. The team that leads Skydance Forward.

People are talking about the new Optimus factory at Giga Texas. It's progressing rapidly. It'll be 7 million square feet with a targeted capacity of 10 million Optimus per year. Initial production is planned to begin in 2027. David Hall says, 10 million humanoid robots is enough physical labor to build New York City in 5 months, and the first Optimus factory will manufacture that many robots in a single year.

That's crazy.

I like, we should build New York in California.

In Texas.

It's a great argument.

I think it's going in Texas first.

It's going to be in Texas first.

I don't think it's going to be in California. anytime soon. He left. But the Frenchmen over at Atelier Missou followed him to Texas and they put up a statue which people were going back and forth on saying it was disproportionate. Did you see this?

Really?

Yes. So apparently statues should be like one— like a ratio of head to body of 1 to 8. They went with 1 to 7. very controversial in the statue community. They made a lot of people angry. They had— there was a backlash. There was a backlash to the backlash. Lots of people going back and forth on whether or not the statue is aesthetically cool. The big scoop was that they've been accused of sort of LARPing as, like, the old-school way of making statues. They do, in fact, use computer Computer-aided design, CAD tools, CNC machines to mill parts of the statue. They use 3D software to design the statues, and a lot of people are frustrated by that because they feel like it's not— they didn't fully go back. I don't really care about that. I think the original question was just—

Brian just dunked on Ben. He just pulled up on him, just dunked on him. Boom. Boom. Come on in if you're ready. I wish we had the basketball hoop over here.

Yeah, we could just dunk it in here. Put it up. Good to see you, Brian. Welcome to the show. What's happening?

How you doing? Good to see you. What's happening?

Oh, good to see you.

He's got a basketball hoop. Yeah. Oh yeah. Oh yeah. You got to dunk. You're on a mission to consistently dunk?

Well, first time in my life to dunk. Yeah. I've never dunked. And like, you know, it's kind of the dream to dunk. Yeah. So I touched the rim last week after 5 months of training.

And are you sure? Are you—

No training.

Have you guys dunked?

I'm 6'8. I can dunk.

Easily dunk.

Yeah, it was never—

I didn't train at all for it.

Yeah, I used to do a lot of box jumps in the gym. I'd probably get back to that. But even when I was training that and I was working my box jumps up and up, I think I would probably still generally slam the ball into the rim and not quite achieve it. How much can you— do you think the average person could, with consistent training, actually increase their—

Have you seen the data over time? The max vertical is going way up. I'm sure you've studied all this.

So, you're 6'8. What's your standing reach? So, when you've—

I mean, I can touch the— I can definitely— I can't touch the rim, but I can touch the net pretty easily.

I'm 91 inches, so to 120, so my vertical's got to be at least 31 to touch rim, and then probably 36 to dunk. Medium NBA is 35.

Okay, so you have to be better than the average NBA player in terms of vertical. Wow. But I think you'll do this, no problem. Like, you're so dedicated to the craft.

But I'm 50. Yeah, no problem. So the probability of a 49-year-old white guy who's never dunked before or played collegiate basketball is like, I don't know, it's very low in the odds. And so by the time you're 49, you have something like 45— 20 to 40% less fast-twitch muscle fiber. You've just like, you're atrophied. So you have to like rebuild that whole thing. So I've been doing that. So I mean, I think the odds are against me.

The crazy thing is that like the expectations are so high. I'm just like, yeah, of course you're gonna do this. next week. Like, call me when the next challenge starts.

Oh, man.

But now you've— yeah, you've actually honed in on stuff that is difficult and will be impressive.

Yeah, I think I messed up this entire past 5 years because, you know, when you do it, like, it's this whole reconciliation moment when you say, like, you know, go to bed on time because sleep is good for blank, or like, yeah, you know, eat well. People like, I don't want to go to bed, I don't want to eat well. But if you're like, but if you do these things, you can dunk, then it's like, yeah, I'm in, you know, I'll play this game.

Yeah, the don't die was like a little It's a little abstract. People put it off. It's 20 years in the future, 50 years in the future. But dunking is today.

That's the goal. Dunk at 49, and then people are like, I'll do that stuff.

Do you know if dunking is something that, like, if you do it when you're young, it's something you can maintain? Like, not quite like something like riding.

It's an incredibly good workout. That used to just be my cardio, is I would just dunk back and forth on the basketball court, and I'd be sweating, and it was so invigorating.

All right, we get it.

But it's such an underappreciated physical ability. People focus on biceps and bench press and VO2 max, but you know what? Explosiveness and dunking, I now have this whole new appreciation for how to be healthy. I will not give up on jumping ever again. Because once you lose it, it's very hard to get back, but it just is so good for bone mineral density, for muscle, tendons, ligaments.

Sleep, diet, and dunking.

Yeah.

And exercise.

Yeah, just dunk and everything else falls into place.

Where did this come from? Was this just on your list of challenges or something?

Well, is it kind of like— I feel like I've mastered health and longevity. Like, past 5 years, we've done all the things.

He did it.

Yeah, we looked at all the evidence.

It gets like really, really long tail. Totally. Like, he's doing these deep studies of fringe, like, if I change the vitamin C, then it's not worth it.

I felt like I went through this period of— and I think everybody deserves to go through a 3 to 4-year period where they are overly obsessed with their health, like borderline obsessive, turning down a bunch of stuff, learning, experimenting, being wrong, all these things. And then, you basically do all this testing and you sort of acclimate, and then you can actually get a little bit sloppier. You can fall off, but you'll sort of level out to a health baseline that is much higher than where Yeah, exactly. And having gone through that period, at one point, I was like, I'm spending— it feels like 70% or 80% of my mental capacity daily just trying to be healthy. And I got to the point where I was like, that wasn't worth the trade-off. But—

That's right.

I still get the— every day, I get the benefits of having gone through that period. And I think it's something that you people— the right way to approach it for the average person is like, go through that every cycle, that every, whatever, 2 years. Whatever, 5 years, something like that. So you're continuing to compound. And then at different phases of your life, you need different sort of practices and habits and things like that.

Yeah, well said.

I got the next challenge for you. The day you dunk, job's not finished. You're on to the next one. 7-minute Nürburgring.

7-minute?

7-minute Nürburgring.

What is that?

The Nürburgring is an excruciatingly long racetrack in Germany.

They call it the Green Hell.

hell. It's extremely dangerous. It puts your driving skills to the absolute limit. You can also drive it as like a commuter in some ways, but they shut it down. This is where they test all the different sports cars.

Yeah.

And they've done, you know, everything from track-only F1 cars essentially on the track to everything from like a new Porsche would get tested there, and the Model S Plaid has tested there. And you can get an advantage by picking a very good car for it.

Yeah.

Something that's designed for the track, something that has really strong cornering, lots of downforce, but it still can go fast in the straights. But there's no excuse for being able to actually push it to the limit. And 7 minutes is a threshold where Only a few cars can get there. So you need a very good car. You need to understand how that car works. And then you also have to be near pro level, I think, to put up a 7-minute. What do you think? Is that a reasonable time? Or is that too aggressive of a—

I don't know. But I do think I like where you're going with dunking because it goes into just feats, like abilities. Yes, that's right.

Yes, the Renaissance Man.

100%.

This is the Renaissance Man method.

And we talk about this because even— Even like our—

Well said.

Our— the top guy, the top car guy in tech is obviously George Kurtz.

Yes.

Because not only could he buy any car he wanted, he competes at a professional level, right? Like he just won LMP2, the sort of Pro-Am prototype class. You were at the race. And so I think that that's what like going and sort of mastering these sort of like fringe pursuits, like John has talked about being able to hit a target from a mile out with a rifle, things like that that take physical abilities. And it's so much harder to actually learn when you're 50 than if you started as a kid and it's just supernatural. But yeah, that's the next frontier.

Yeah, we need an EGOT for tech. Hollywood has the Emmy, Grammy, Oscar, Tony.

Oh, that's a great idea.

For tech, we need the EGOT, I was thinking about going to space is a good one. Bezos has been, but not many other tech leaders have. The COVID of Time magazine is another big tech leader milestone. You could also put selling a company above some threshold or something or being the richest man. There were a few others. Testifying in front of Congress, that's like a rite of passage for tech leaders. It really is. And Bezos done that one too. And then buy a sports team. That's another one.

Or maybe race, you know, give it a structure, give it a ranking. Everyone's going to follow suit. But for sure, this is the thing is like, you know, people talk about AI, what's coming in, and our culture is obsessively work. Like, you never stop working because it's a status game. You're looking at bank account and followers. But like, I think this is like a new Renaissance era of these side quests. Yes. So at 49, can you dunk a basketball? At blank, can you win in a professional race, or be, you know, top in amateur. So I think it is— it's good for these social— for us to introduce this, and create social pressures, that if you don't have a dunking side quest, like, what's up, bro? Where are you at?

You need a PhD in side quests.

What's wrong?

I saw one influencer who sung the national anthem. That's a great side quest.

I did.

That's a great side quest.

I was going to actually— for the dunking, I was going to do it at a Lakers game. Halftime, Polymarket. We have bets.

OK.

I've got a bit— I have like 5 more inches I need to lock down before we get— so I don't want to mess that one up.

Yeah, I'm optimistic.

I want to talk about wound recovery because something as a 30-year-old I've noted— I used to get like, let's say I go get— I'm surfing and I get a cut. It used to be like something I didn't really care about at all because like 3 days later it's gone. I'm not— I don't, I don't think about it anymore. And now I realize like whatever, like the, the time it takes for me to heal from like little cuts is like going up.

Even if you put beef tallow on it?

Yeah, yeah, yeah. I used a gallon of beef tallow and still nothing.

That's so shocking.

What is going on? John's favorite bit is like the solution to every world problem is just beef tallow. It's like kind of funny.

The whole community went really far into beef tallow. It was like, we gotta do beef tallow.

Yeah, no one's tried beef tallow for AI safety.

Yeah.

It's like Big Fat Greek Wedding where the guy, his Windex is his solution, the Greek.

Oh yeah.

Everything he uses is Windex, but beef tallow is.

I imagine beef tallow is good, but people were framing it as like, it will cure everything.

Relationship in trouble? Beef tallow.

It just went too much, too much.

But wound recovery, is that something that—

It's a strategy if you get a scratch.

It feels like an overall, Vitality indicator.

Yeah, it's just got to be like, it's probably a good biomarker. Like, we should probably— I should probably initiate some cuts.

Yeah, you can cut.

Don't do that.

No, no, no. But I think it's like— That sounds terrible. It's a little mild, like a paper cut.

Yeah, yeah, yeah.

But I have to imagine it's just a pure representation of capacity.

Yeah.

Yeah. I mean, this is the thing. We look at my body, and we say, OK, you can go punch— you can lift iron in the gym. You can do weights. Like, what does your body function by default? This is why I measure, for example, nighttime erections. You can do nothing about it. You go to sleep, it happens or it doesn't. So your body performs as a 50-year-old or an 18-year-old. And so I like those markers in that you have no— you can't do anything to change the outcome other than just exist. And so I think wound healing is probably that, is that you can't optimize your way out of it. The body just does its thing. So it probably is a good biomarker.

Yeah, interesting. Weight training. I feel like sometimes I go easy in the gym and that I'm not pushing myself to actually increase the amount that I'm lifting.

Progressively?

Well, no, simply because I'm like, I'd rather be currently this strong than get injured, right? Or I don't feel like I should be pushing my deadlift, but you have been.

Yeah.

So, talk about that trade-off of What is the value of pushing yourself from doing 135 or 225 or beyond that versus injury risk and how those things factor in?

I just had my first failure. I failed at a 435 hex bar. So my coach is— I have an NBA coach. He's great. He's been wonderful because he's never allowed me to fail. So I go into the lift with confidence, whether it's front squat or hex. And so this one, I blame— I had lotion on my hands that morning. Yeah, I blame that, but still, I failed at it.

That's way better than failing and your back going out or something.

Yeah, exactly.

He's been very good at form, and so he's very careful. If I ever have the slightest thing in my body, he says stop. So I do think he has at least done to be very careful with me, where I never thought I'd be doing a hex bar for—

A lot of it's psychological. Like, you can, if you cheat on the form, you can add another 50 pounds.

Yeah.

Every time. And everyone will be like, wow, that guy has the big weight on. But then you're risking things. And it's so much better. You're actually going to build more muscle if you take the weight down, use the right form, progress. But you've got to be like, yeah, I'm the guy squatting 100 pounds today.

There was a guy in the gym last week. He was doing this squat, I swear, like 500 pounds. I thought his back was going to break. I was going to jump in like, bro, I respect you, but man. All like the three of us were there. We're like, this is bad. So yeah, I agree. Like it's and there's so much pressure in the gym. Yeah. You have people that are looking at you. Everyone's kind of like peacocking a little bit. Yeah. Yeah.

Everyone's sizing you up.

Definitely don't want to. Yeah. So there's a lot of pressure for that. Yeah.

You stay out of the sun. Yeah. You don't suntan. You also don't do spray tans because of the chemicals. That's right. Correct. What is your take on melanotan? Yeah.

So I mean, a few people. gone that I think they haven't come back. I think— wait, what? I think so.

I mean, I don't know. Permanently tanned now?

I'm not sure.

John's like, sign me up. I'm not sure. Well, I didn't know about that value prop.

But also, the color is not necessarily pre-known, right? Like, you do it, it like emerges. Oh, so it's not like pick this color.

Is that a shade or a hue? I guess both, because if I turn out green or more purple, that might be worse than just being, you know, along the same line.

If I understand this correctly, It's not like you put your finger on a chart of like, I want this color with this dose.

It's just take me in the general direction.

Yeah, exactly.

I don't know what I'm getting.

Exactly.

Got it. OK, so you've stayed away from it.

Yeah.

How are you feeling about the other more experimental peptides? Is there anything that you're maybe sitting out but optimistic about? Oh, that one would be cool in the future if it can pass FDA, et cetera.

My take on peptides is it's a psychological phenomenon. People feel that they're powerless. They can't stop scrolling. They can't go to bed on time. They can't stop eating junk food. And so you have this little vial, you have a needle, and there's studies showing the psychological benefit of, like, I feel power by injecting myself, and I feel— and they— so peptides are a real thing.

Yeah, you get the same thing with supplements, too.

Exactly.

Like, I'm taking all these pills.

But it's less powerful, because the injection is so much more visceral. You're like, I did something real. I risked myself. It must be working.

So peptides are real. They obviously have benefits. But I think the enthusiasm far exceeds—

OK.

Exceeds where the evidence is at, exceeds the potential benefit. People would be much better off getting a dose of growth hormone by going to bed on time.

Sure.

You get that peptide dose by going to bed and get your surge in your deep sleep. So if you're going to do peptides, measure it. We have actually a new technology we're using to measure peptides. Because if you do any of the big peptides, you can't just do a blood draw and say, is it working? There's nothing you can see in a standard blood panel to say yes or no. You need to look at the molecular profile. So we have a new test we're looking at doing that. So I think that might change the conversation of like, here's what we're seeing, and you may want to think twice.

Yeah, it's 1:40 PM. You don't have to go to bed yet, right? Uh, space. Yeah, would you go to space, or is there radiation? Is there anything that you're worried about from your health perspective?

No, I don't have a desire. No, no, it just has never hit. I'm going for the tech ego.

Yeah, I'm putting space in there.

Are you? Is that your list?

It's got to be in there. It's the, the Blue Origin rocket. You go up past the Kármán line, it's available. It's something that Bezos pioneered.

I wonder if you can do it in VR. I wonder if it's the same experience. I wonder if you'd have the same— or what do you call it?

Probably on the vomit comet, if you go on the plane.

Yes, I've done that.

You've done that?

Yeah.

Oh, okay. Well, then I think you get half checkmark.

Yeah.

What about flying in airplanes? People have said, like, if you fly a lot, you get radiation when you're up in the sky. Is that something you try and monitor or mitigate?

Yeah, I started flying actually a couple years ago. I got my license.

Oh, really?

I do fly myself around.

Cool.

Um, and so you— the—

What kind of plane?

Uh, Phenom 300.

Cool.

Yeah, so I got typed in. I did a little, um, Cessna 172.

Yeah.

Then I did a Bonanza.

Yeah.

The prop. Then I did a Mustang, went to a jet.

Yeah.

And then, um, did a Phenom 300. And so, yeah, we—

Did you go all the way through VFR, IFR?

Yeah.

Wow.

Yeah, I got typed in all the different planes.

I've heard that IFR-rated pilots are like way safer than VFR.

Oh man, like, yeah, so much safer. Yeah, that's— I almost died twice.

Whoa. Okay. Yeah, tell us about the first time you almost died.

I mean, these stories are so dramatic. Okay, so the first time I flew my prop plane down to South by Southwest. Okay. I— and so I made a rule I would never fly alone ever. So I knew the risk. I saw the data. I was like, this is, this is terrible.

Oh, because if you faint or something—

Or just like amateur pilots are something like 7 times more likely to get into an accident. You jump into playing, but you don't remember everything very— you want a professional pilot. They do it all day, every day. It's all they think about.

So you would hire a professional pilot to fly with you, basically?

Yeah. I'm always in the cockpit with a professional pilot.

That's so cool. That makes a lot of sense.

So I've never, ever flown alone, excluding my checkride.

OK.

And so we're down at South by Southwest. We're coming back to— I was in Chicago.

Wait, the checkride, they make you fly by yourself?

Yeah, you have to take a solo. You have to do a first solo, your first solo flight, then your checkride, then afterwards.

Yeah, because they assume everyone— they assume everyone is going to be— most people will be flying themselves solo, so they have to, like, validate.

For certification, you did it. But then, was South by Southwest a different story?

So he's down there with me. And so we're coming back. And in aviation, you can look at the TAF. You can say, I'm going to arrive at this airport between noon and 2:00 PM. What's the expected weather forecast? You look at the winds, and you say, OK, I can land on this runway. It's got this directionality. You're OK. They have fuel, et cetera. And so, we were coming back, and we were landing somewhere around noon, and he looked at the TAF mistakenly between like, uh, 2 and 4.

Oh.

So, we get in the plane, everything's going well, and then we're like 100 miles out, and we're at 14,000 feet. That's like, uh, close to the top of this plane before you have to get oxygen, and we get these insane tailwinds, like 100 knots. We're like, what the hell? Like, so we're making an incredible time.

Okay.

Our ground speed's like 250, 300 knots, and we're like, this is, this is nuts. Um, so then we start getting some weather reports. There's an actual tornado sweeping through the Midwest.

Whoa.

And so I looked at him and I'm like, why are we flying into a tornado? So he basically, he messed up looking at the wrong 2-hour block.

Okay.

And so after that, everything settled down. Like, the weather's just fine. But that— there was a tornado. So we were now going to our airport and the winds are, uh, into the land and the winds are 35 knots gusting 45, which in this plane, when you go to put it down, it's too much lift. So we, we try to make our landing and there's too much lift. We can't put the plane down. So we're dancing trying to put it down. We can't put it down because it's also gusting.

Yeah.

So now we have like 17 minutes of fuel left. So we call, we're like, we call air traffic control. We're like, we need an airport close by. They're like, well, the closest one's blank. And so we We find one that has basically a fire truck, and we're like, we're gonna go down, we're gonna flip, we're probably gonna start on fire, so we need someone to come out and—

What do you mean flip?

It's like, well, we probably won't be able to be stationary. We'll probably like try to land and then flip the plane.

So the wings will like back off?

We have to go down somehow.

You're crash landing.

You're planning a crash landing.

Yes. OK, because the wind is still so bad at this other airport that you're expecting that you'll flip.

Exactly. Because they came back, they're like, OK, well, within a 200-mile mile radius, the conditions are blank. And it's just like basically like—

All bad options.

45 knots gusting 65, like really, really terrible winds for this little small plane. Um, and then we have crosswinds on the plane because we can't find a runway directly into the, the air. Um, and so we go in, and now I'm thinking like, I've got 3 little kids, and I'm like, why did I ever think flying was a good idea in a prop plane? So I'm like redoing my entire life. We come in, Wait, is this pre-Don't Die? This is pre-Don't Die. OK, OK.

You were still like, die? I'm going to die. I'm going to die someday. Someday. I better go down in flames.

10 years ago, I think.

Yeah.

So we come in, and we cross the threshold, and like, we're fighting for our lives. We're both like absolute ghost white. We are for sure— we think we're going to die. We cross the threshold, and somehow, the winds die down. Like, we don't know what happens. We come in, soft landing, scoot off the runway. We're both like, wow, what? And so, we get over, we park the jet, and then the plane, and then right when we jump out, the winds burst up. We almost fall over. We can't even stand. But, like, we— instead of going home that night, we just said, let's just get a hotel.

And sleep.

Yeah. Tomorrow, we'll fly back. But, that was, that was, uh, That was the first time. That was the first time, yeah. We'll save the other story.

Yeah, we'll save—

But flying commercial, flying lots of hours in the sky up high—

So we hang out—

Radiation effects?

Yes.

What do you think?

So my jet goes to 45,000 feet. It's rated for 45. We hang out about 36,000 to 38,000, thereabouts. So you pay a little bit of a fuel cost by being down lower. But the— I think the trade-off for radiation is worth it because once you get in— because it does increase. I think it's—

Drag?

Uh, no, it's like a logarithmic increase.

Okay.

So when you look at the radiation dose increase, like once you get higher in the high 30s into the 40s, it's a pretty substantial increase in radiation exposure.

Yeah.

So we make that trade-off, fly a bit lower, burn a little bit more fuel. But—

I'm wondering if any of the jets have like radiation shielding.

None of them do.

No?

No.

As a value prop, you'd think some people would want that.

Yeah, it's just too hard to put it in your manufacturing line.

Sure.

Yeah.

And then it might kill the Wi-Fi or something.

Yeah, exactly. Some stuff like that.

People want the Starlink.

All right, as much as I like talking about planes, let's talk about Atlanta this weekend.

Yeah.

You go way back with Toby and DHH. It's a cool story. We can probably skip that and get right into racing. Yeah, what— was it just hearing stories from them about what they're doing that made you want to go out there? Because with other extreme endurance sports, running and cycling and all these things, it's extremely apparent that these people are under stress, right? They're breathing heavily. You can see it. They're sweating. They're slowing down. They're doing all these things. And racing, it's almost invisible.

Exactly.

They get out of the car, and they're usually so disappointed or so excited that the sort of Tiredness is like not apparent even. That's right. Like completely invisible.

Yes. Yeah. So Toby and I were at a conference last weekend in Aspen together, and we were catching up, and he was telling me about his upcoming race, and we got into the rabbit hole. And you have seen the memes where like the person's on the phone and the dog's right by, they're like, and we're gonna go to the park, and we're gonna play ball. Oh yeah.

Dog's like, what? Yeah, yeah, yeah.

Saying all my favorite words. Toby was talking about the biomarkers he was capturing, and the car, and the track. And I was like, what? This is so cool. So we went down the rabbit hole, and he's like, if you want to come hang out, just come hang out in the pit. And I was like, oh, man, in.

That's awesome.

So I made it a project. Yeah, so like you, I never understood racing. I never understood the fascination with it. Driving fast cars, cool, but I didn't understand what the body goes through, and what they are— they are unbelievable. They are elite athletes.

Mm-hmm.

Like unreal.

Yeah.

So, once we actually mapped out the biomarkers, like some things I didn't realize. So, he's in a cockpit, 110 degrees.

Mm-hmm.

So, already hot. His clothing won't allow him to breathe. He is— his heart rate is 142 on average, peaks at 171. He's sustaining 700 pounds of G-force as he makes these turns. His helmet's 70 pounds of G-force as he makes these turns. There's no cooling. In the cockpit, his his respiration rate is equal to running an eight-minute mile. It's like you're panting. This is all stationary, right? So it's hard to imagine. Like I understand what it's like to sprint or do really hard out, but like this is you by yourself experiencing almost like a triathlon. Extremely claustrophobic.

Extremely. Like I've been in, I've been in certain cars and you have to like fight the. Cluster.

Yeah, you do.

Like, it's like you're sitting there and sometimes, like, yeah, when it's hot and, um, and then you get over it because you're like, cool, I get to go drive around this track.

But for a second there, it's like wedging yourself in a bucket seat is crazy. It's just so much different than any other car.

And you've got a 6-point harness that basically they have to strap you so tight because you have to feel the car. Yeah. If you can't be one with the car, you can't catch all the Tiny little tenths of a second, and so once you stack up all these things, and then he does it for two and a half hours straight.

I honestly, that's just one shift.

Yeah, I I really I struggle to wrap my head around that ability. Like in that extreme physical conditions, psychological conditions, you have to be there's zero lapse of focus. You're fighting for tenths of a second.

And do you think do you think it's possible to stay focused for that long because your life is? is on the line because there's so much risk that you're sort of forced into flow state where you recognize like you're going, you know, 150 miles an hour. And if you do make a mistake, like, you know, current, all the different safety mechanisms and the way the cars are built, like, they do protect you to a pretty crazy degree. Like, you'll see crashes that the car rolls over 8 times and catches on fire and the person gets out.

Yeah.

And is walking. But it's still like no one wants to go through that kind of trauma, and not just personally, and the cost, and losing the race, and all these things. But it feels like you're forcing yourself into this fight or flight dynamic that allows you to have a new level of focus.

So when you look at his data— so Toby's awesome because he captures all the data.

So his DHH is just like, I'm getting in the car. I'm putting my foot down. I'm not lifting up until I see the checkered flag or God or whatever.

They have very different styles.

Yeah.

So if you look at his nervous system, it goes to zero chill. Like Toby, he gets in, it is instantaneous, 100% adrenaline, like just full on. Even 30 minutes before the race, his body starts racing. So his heart rate went up to 127 BPM from his baseline of like around 50s or something. So he's able to anticipate And I was in the trailer with him, and, like, I could see him go from, like, it's morning day prep, to, like, I'm locked in.

Mm-hmm.

And so I think on your list, Tobi, he's so softly spoken, right? He's one of the most epic founders of our generation, right? He's an incredible builder, not only just in what he's built as a company, but he leads in how he builds with AI. He's just an incredible human, and— but he's so soft-spoken. No one even knows he drives, let alone what he goes through when he drives. So I wanted to tell his story. Like, this is a serious sport. He's an elite athlete. And so I think people would— I think he's emulation-worthy as a founder. People should be like, I want to do stuff like that, too.

Yeah, you're getting me fired up to go racing.

That was the goal. I mean, I think Toby's a good one.

What about the— like, how do you measure it from a— like, what are the— What are the health impacts of— like, because you can be an elite athlete and be destroying your body.

Yeah.

Right? Like, you've seen this in, you know, people that do like really hardcore running where it's like—

Yeah.

They'll be 30 and they look like they're 40 because of the daily massive amounts of like cortisol and stress. And so, my hope, because I enjoy driving cars fast on a track, is that because the actual periods of like ridiculous intensity and doing the 2.5-hour stint is so rare that it's not going to age you, accelerate aging. But it is so stressful that it can't be all good for you.

I agree, yeah. Like all things, there's probably a trade-off. One of the data points I was looking at is drivers are able to do with one brain region what a novice requires 15. So it does something with the brain which is almost like a neural enhancement. And so I don't know, on balance, it may be the case that you're winning. You may take some this and that. The noise levels are extremely high. Oh, yeah, yeah, yeah.

I had hearing protection. Yeah, last time you had the hearing aids on. And yeah, I've been super conscious of sound since then. And the issue is when you're in the car, your monitors in, but— and I'm sure maybe some of them have noise-canceling elements, but you're still like getting— you're just right there with the engine.

When he takes off, I mean, um, I think it was something like 130 decibels. So even though I have—

He's sitting right by the— in the pit.

Yeah, when I'm in the pit, when he takes off through his run, I had hearing protection in, which does 30% dB protection. Um, over 100. Exactly. So like I needed 20 more. So I needed to do this plus that. The cups. Yeah, yeah, exactly. So You need like 50, 60 of— but yeah, so there's for sure some trade-offs. I mean, I think if you look at the joy that Toby gets from this thing, it lights him up, I think, probably more than anything in the entire world. Like, when he exits that car, it's like— I don't know, maybe that's what Buddha looked like when he achieved nirvana. He's so— like, he's just omnipresent to reality. It's actually just an incredible sight.

Yeah.

I was excited to see it. No, I've had— My experience on the track is you'll be in— I've gone from this intense flow state, which is incredible, and then I think about my email for 1 second, and then I spin out.

Yeah.

Or I'm off the track, right? It's really binary where you're in that because I haven't done actual racing, right? And so it's really just been for fun and not at the kind of speeds that they're doing in an IMSA. Race like this last weekend. And so it is possible to be like, oh yeah, I should get back to that person.

Yeah.

And then again, but the second you do that, you're going at such high speed.

Yeah.

And the other thing that is so remarkable to me is just like the level of consistency across that 2 and a half hours where they could be trying as hard as they possibly can to go faster and gain like a A 10th of a second, right? And then, yeah, it's sim racing too. Did you have a post about that, or did that tie into your—

Yeah, I was looking at—

Yeah, it was tying into that, like, sort of studying these different brain regions. That made me think, do you think there's been brain games in the App Store that force you to activate different— brain functions for a while now. I don't feel like any of them have like caught on fire. Like for me, I'm very addicted to sim racing. So, and I, and so I sent that, that post to my lovely wife about it.

I've seen a lot of people do that.

And she was like, I'm happy for you. You probably should still focus on— the issue for me is it comes out of my sleep time because I've like, the kids go to bed and I'm like, okay, I can go racing now. And then, and then I'm sleeping less. But, but do you think there's more Do you think there's more games to be built that force people to do these types of things? Is it something that you're spending any energy on?

Yeah, I mean, I think the world is absolutely sleeping on how fast the don't die revolution will happen. Like, I think that once AI makes the first couple of breakthroughs, we have our first anti-aging drugs, and people feel like there's a hope, I think the society will snap from wanting to put all of our efforts into status seeking, wealth accumulation, to don't die. So, I think that— I do. I think it is going to be a snap change in society. We'll just revalue— we'll change everything we do and value.

Yeah, were we talking about that about—

what is going on?

That's my pal.

That's my pal, yeah. Awesome.

Yeah. Do you want to pop on?

Yeah.

Bring him on, or does he not get—

What health benefits? Joy?

Yeah, well, so— Guitara.

Here we go. Hello.

Oh, there we go. Hey baby.

Oh, I've been away from him for 10 days. How you doing, baby?

No way. Hey. Oh, so cute.

Oh yeah, amazing. Whoa, what a smile. Happy boy. That's our show. We can close it out.

Um, thank you for tuning in to TBPN. Uh, we will see you tomorrow at 11 AM Pacific. Leave us 5 stars on Apple Podcasts and Spotify. Sign up for our newsletter.

This might scare the dog. Hopefully it's well trained.

Goodbye!