More or Less Podcast

Why Identity Verification Matters For The Internet | Vox Americanus, GLM 5.2, Claude Fable 5 Is Back

3416 summary words 15 min summary Watch video

Start with the signal

15 min read

Summary

At-a-Glance

  • Verdict: Skim
  • Core thesis: Sam Altman launched Vox Americanus, an identity-verified social platform for Americans, while the group debates AI model commoditization, cultural fit over technical skill, and whether the loneliness epidemic will worsen as vibe coding becomes mainstream.
  • Why it matters: This episode contains concrete operator insights on: verified identity as distribution infrastructure, GLM 5.2 threatening Anthropic/OpenAI margin, cultural fit mattering more as coding skill floors rise, and CFO-driven model routing decisions reshaping inference economics.
  • Best use: Listen for Sam's identity verification strategy, skip the Padel/Taylor Swift segments, focus on the GLM 5.2 commoditization discussion (minutes 30-40) and the political infrastructure idea (minutes 10-20).

Executive Summary

Sam Altman soft-launched Vox Americanus, a U.S.-identity-verified social platform using Stripe's ID service, arguing the internet needs an identity layer to separate real Americans from bots and state actors. He built it via vibe coding in days, positioning it less as a standalone network and more as OAuth-style verification infrastructure other platforms could adopt. The core use case: enabling substantive dialogue with verified Americans rather than arguing with bot armies. Distribution remains unsolved—he acknowledges people with existing audiences (like the hosts) don't need it, while unknown voices who would benefit lack the follower base to bootstrap network effects.

The conversation pivoted to political infrastructure: Sam argued recruiting good candidates requires solving the job itself (making elected office not suck via turnkey staff, travel, briefings) rather than just campaign strategy. He resists running for office himself, viewing current incentives as broken, though others pushed him toward a 'czar' or appointed role. The group debated whether nationalism (American-only internet spaces) is healthy tribalism or dangerous fragmentation, with Jess arguing some vitriol is domestic, not just foreign bots.

On AI: Dave reported GLM 5.2 (a trillion-parameter open-source Chinese model from Z.AI) now matches Claude Opus 3.5 in production use—friends running incubators and agents swapped it in unnoticed. This commoditization moment means CFOs are reasserting control, buying bare metal (B200 stacks ~$500k), and routing to cheaper models. Anthropic's revenue still rips due to Q1 marketing momentum and delayed enterprise adoption, but the margin compression race is here. Nvidia spot rental is down 22% week-over-week as companies shift to owned infrastructure.

Cultural fit now matters more than raw engineering skill, since vibe coding and agentic tools lower the skill floor. Britt emphasized teaching a founding engineer to be 'human'—high integrity, street smarts, team vibe—over pure technical chops. The group worries about loneliness: Jess warned people are alone with agents at all hours, eroding real social conversation. Sam countered that tinkering empowers builders (he ran a 24-hour 'bot school' where influencers shipped real products), but Jess stood firm: most people won't need to vibe code; AI will become invisible infrastructure, not a daily practice.

Key Takeaways

  • Claim: Vox Americanus is an American-ID-verified social platform Sam built in days via vibe coding, using Stripe's ID service to prove users have U.S. driver's licenses. | Evidence: Sam: 'Stripe's ID service makes it shockingly easy—you put in a license, twist your head a few times, and it's like, yeah, that's a real person with American ID.' He mentioned the sign-up NUX is better than Truth Social at launch, with video recording and a solid interface. | Caveat: Distribution is unsolved. Sam acknowledged people with existing audiences don't need a U.S.-only platform, while unknown voices who'd benefit lack the followers to bootstrap network effects. He's unsure what unique content this generates beyond 'I'm definitely talking to Americans.' | Implication: For Ken: This is infrastructure thinking, not a social network play. The OAuth/verification layer angle (indexing verified Americans on Twitter, requiring verification on dating apps/Amazon sellers) is the real product insight. If identity verification becomes table stakes for quality discourse, expect platforms to partner or build in-house rather than adopt a third-party service unless it becomes a standard (like OAuth). | Timestamp: ~8:00-15:00
  • Claim: GLM 5.2, a trillion-parameter open-source Chinese model from Z.AI, now matches Claude Opus 3.5 in production—developers are swapping it in without teams noticing. | Evidence: Dave: 'A friend running a big SF incubator swapped GLM 5.2 on all their agents and nobody on their team noticed. Everyone across OpenClaw is switching from Opus.' Another friend at a NYC hedge fund confirmed Nvidia spot rental is down 22% week-over-week as companies buy bare metal (B200 stacks ~$500k) to run GLM 5.2 at the cost of power, not tokens. | Caveat: GLM 5.2 is three weeks old; market share lags adoption. Anthropic's revenue still rips due to Q1 marketing momentum and delayed enterprise buying cycles. Political/nationalist concerns ('I don't want my numbers multiplied in China') may slow adoption among certain customers. | Implication: For Ken: The commoditization moment is here. CFOs are routing token usage to optimize cost, not loyalty. If OpenAI's inference advantage (50% cost reduction per Jessica's reporting) lasts only a month before others catch up, margin is fleeting. For agent systems: build routing logic now; don't lock into a single frontier provider. For investing: scrutinize margin durability in foundation model companies—open-source parity is a structural threat. | Timestamp: ~35:00-42:00
  • Claim: Cultural fit now matters more than raw engineering skill because vibe coding and agentic tools have lowered the skill floor to build production-quality products. | Evidence: Britt: 'A founder with 20 years at Apple/Spotify interfacing with engineers asked, "Isn't it too late?" I told him he's more capable than he thinks.' Sam: 'At our 24-hour bot school, social media influencers with minimal coding skills shipped products to actual users—wildly different from hackathons that produced half-working demos in four days.' Sam added he now emails his infrastructure to spin up everything, no longer editing file-by-file. | Caveat: Jess pushed back: 'I don't believe we all need to be vibe coders. AI will become invisible infrastructure.' She's skeptical most people will devote two hours/day to coding. The loneliness epidemic risk is real—people alone with agents, not in community. | Implication: For Ken: When hiring for agent/AI startups, prioritize street smarts, integrity, and vibe over Stanford CS degrees. The skill gap is narrowing; the cultural/execution gap is widening. For content: the 'you're not too late' narrative is a chapter in Sam's upcoming book—early mover advantage in AI app layer is overstated if tooling keeps improving. For workflow: ruthlessly audit where human judgment vs. agent execution matters. | Timestamp: ~43:00-50:00
  • Claim: Recruiting good political candidates requires solving the job (making it not suck via turnkey infrastructure) rather than just campaign strategy or media training. | Evidence: Sam: 'If you had a nonprofit that said, "We have a PC-12 network, briefings done for you, office set, bill writing—you drop in for key moments," you'd get better candidates. The job pays shittily, you're away from family, in Sacramento/D.C. constantly. People focus on campaigning; they should focus on making the job appealing.' | Caveat: Appointed roles (Sriram, David Sachs as 'czars') are not elected jobs—they're 'vice chairman at best,' lacking real power to enact change. Sam believes you need better fundamental politicians, not just influencers with proximity to power. | Implication: For Ken: If building infrastructure for civic engagement or political tech, the insight is job design, not voter outreach. The 'first gentleman' joke aside, Sam's point is structural: reduce friction (travel, admin, family separation) to attract talent. For GTM in regulated industries, similar logic applies—make compliance/bureaucracy invisible so operators focus on impact. | Timestamp: ~15:00-22:00
  • Claim: The Fable 5 release/halt/re-release was 'clearly trust destroying' but strong marketing—Sam is unsure if it's brilliant strategy or intuitive chaos. | Evidence: Anthropic released Fable 5 alongside Mythos, Amazon raised jailbreaking concerns, Anthropic disputed, government halted it, then struck a deal allowing 50% of weekly usage limits until July 7th (after that, pay-per-token). Susie Wiles, Trump's chief of staff, publicly endorsed the re-release. Sam: 'Is this the most brilliant strategy or no strategy? I don't know which category it is.' | Caveat: Jess questioned whether a real safety issue was resolved or if it's political theater ('they strike a deal when it's worse not to'). The process seems performative, not substantive. | Implication: For Ken: Model releases are increasingly politicized. The 'so dangerous we must check' narrative benefits fundraising ('even more dangerous models') and U.S. competitiveness optics. For investing: trust in model safety processes is eroding; expect more regulatory theater. For operators: don't assume model availability is stable—plan fallbacks. | Timestamp: ~50:00-55:00

Detailed Brief

Vox Americanus: Identity Verification as Infrastructure

  • Claims: Sam built Vox Americanus in days using Stripe's ID service to verify U.S. driver's licenses, proving users are Americans.; The core insight: the internet lacks an identity layer, making it impossible to distinguish Americans from Russian bots or state actors.; Distribution is unsolved—people with audiences don't need it; unknown voices who'd benefit can't bootstrap network effects.; The better strategy may be OAuth-style: verify identity once, use it across platforms (Twitter index, dating apps, Amazon sellers).
  • Evidence: Sam: 'Stripe's ID service is shockingly easy—license, twist your head, done. The NUX is better than Truth Social at launch.'; Sam: 'I don't understand why Twitter doesn't just launch "validate my ID to get American who lives in California." It's not expensive.'; Jess questioned distribution: 'Would Truth Social have scaled without MAGA?' Sam agreed the question is what unique content this generates.; Sam: 'I was thinking about how services let you prove your Twitter account—you post a link, they crawl it. I could create an index of actual Americans on Twitter.'
  • Caveats: Sam acknowledged he's 'unsure how to spark more than intellectual interest' beyond 'yes, I want comments from real Americans.'; The use case for high-follower people is weak; Sam himself doesn't need it. The value is for 'user 87342' who looks like a bot but has smart opinions.; Nationalist framing risks fragmentation. Jess pushed back: 'I understand wanting real people, but why Americans specifically?' Sam argued it's about avoiding state actor influence, not just bots.
  • Implications: For agent systems: identity verification is infrastructure, not a product. Build it as a service layer, not a walled garden.; For content/business: if you're building community, verified identity raises discourse quality but requires a bootstrap mechanism (celebrity endorsement, niche use case).; For investing: identity verification is table stakes for any platform dealing with trust/safety. Expect platforms to build in-house or adopt open standards, not proprietary services.

GLM 5.2 and the Commoditization of Frontier Models

  • Claims: GLM 5.2, a trillion-parameter open-source Chinese model, matches Claude Opus 3.5 in production use—teams swap it in without noticing.; CFOs are reasserting control, buying bare metal (B200 stacks ~$500k), and routing to cheaper models to cap token spend.; Nvidia spot rental is down 22% week-over-week as companies shift to owned infrastructure.; Anthropic's revenue still rips due to Q1 marketing momentum, but commoditization will compress margin in 6-12 months.
  • Evidence: Dave: 'A friend running a big SF incubator swapped GLM 5.2 on all their agents—nobody noticed. I'm hearing this across OpenClaw.'; Dave: 'A hedge fund friend said Nvidia spot rental is down 22% week-over-week. All capital is shifting to buying raw metal.'; Sam: 'I talked to two public company CTOs today. The idea of using all models and switching between them isn't avant-garde—it's obvious. No one thinks being a Claude shopper is ridiculous.'; Jess: 'Anthropic started getting more hype in Q1. Businesses that were experimenting then are getting the marketing hit now.'
  • Caveats: GLM 5.2 is three weeks old; market share lags adoption. Political concerns ('I don't want my numbers multiplied in China') may slow uptake.; OpenAI's reported 50% inference cost advantage (per Jessica's team) is a race—if others catch up in a month, the advantage is fleeting.; Sam: 'The race is efficiency. If OpenAI can do inference at half the cost and no one else can, they'll be huge. But if they can do it for a month and everyone catches them, it's worthless.'
  • Implications: For operators: build routing infrastructure now. Don't lock into a single frontier provider. CFO scrutiny is here; experimental budgets are over.; For investing: scrutinize margin durability in foundation model companies. Open-source parity is a structural threat. The 'who's got the cheapest multiplication' race is the only race.; For agent systems: assume model interchangeability. Design for swappable backends. Optimize for cost per task, not model brand.

Cultural Fit Over Technical Skill in the Vibe Coding Era

  • Claims: Vibe coding and agentic tools have lowered the engineering skill floor, making cultural fit more important than raw technical ability.; Sam ran a 24-hour 'bot school' where social media influencers with minimal coding skills shipped production-quality products to real users.; The skill set shift: you no longer need to speak a language a compiler understands—you speak to models, which speak to compilers.; Jess pushed back: most people won't need to vibe code; AI will become invisible infrastructure, not a daily practice.
  • Evidence: Britt: 'A founder with 20 years at Apple/Spotify asked, "Isn't it too late?" I told him he's more capable than he thinks. People don't realize how powerful the tools are.'; Sam: 'Last summer I was in Cursor being like, "This finally works." Now I email my infrastructure and spin up everything—I don't edit file-by-file.'; Sam: 'At bot school, influencers produced in 24 hours what used to take hackathons four days to half-finish. They shipped to actual users.'; Jess: 'I like my agent. It's useful for work. But I don't believe we're all headed for a future where we all need to be vibe coders. AI will become invisible.'
  • Caveats: Jess warned: 'People are alone with their agents at all hours, not having real social conversation. I'm really worried about the loneliness epidemic.'; Sam: 'If you can really binge into it, go for it. But Jess is right—please stay human.'; Cultural fit is subjective. 'High integrity, street smarts, vibing with the team' are harder to evaluate than technical skill.
  • Implications: For hiring: prioritize communication, judgment, and cultural alignment over CS pedigree. The skill floor is rising; the cultural ceiling is the differentiator.; For GTM: if you're selling to non-technical users, emphasize ease of use and community. Vibe coding is a wedge, but most users want invisible AI.; For workflow: audit where human judgment vs. agent execution matters. Don't assume everyone wants to tinker—most want outcomes.

Political Infrastructure: Solving the Job, Not the Campaign

  • Claims: Sam argued recruiting good political candidates requires solving the job (making it not suck) rather than campaign strategy or media training.; The job sucks: low pay, away from family, in Sacramento/D.C. constantly, no infrastructure for travel/briefings/bill writing.; Appointed roles (czars, advisors) lack real power—you need better fundamental politicians, not just influencers with proximity.; The 'first gentleman' joke aside, the insight is structural: reduce friction to attract talent.
  • Evidence: Sam: 'If you had a nonprofit that said, "We have a PC-12 network, briefings done, office set, bill writing—you drop in for key moments," you'd get better candidates.'; Sam: 'In 2026, you can't take a nice guy and build a media strategy for him. You need him to already have a media footprint. But the biggest thing is make the job not suck.'; Jess suggested Sam run for president and use Vox Americanus as the platform. Sam resisted, citing the job's appeal problem.; Sam: 'David Sachs and Sriram figured out how to be in government without being in government. But those aren't elected jobs—they're vice chairman at best.'
  • Caveats: Appointed roles lack legislative power. Sam: 'You need better fundamental politicians, not czars.'; The 'PC-12 network' idea is expensive and elite—only viable for well-funded nonprofits or parties.; Sam's focus on infrastructure may miss the campaign/voter engagement layer entirely.
  • Implications: For civic tech: the insight is job design, not voter outreach. Make elected office appealing (logistics, admin, family support) to attract talent.; For GTM in regulated industries: similar logic applies—reduce bureaucracy/compliance friction so operators focus on impact.; For investing: political infrastructure (bill writing, briefing services, travel logistics) is an overlooked category if framed as enabling talent.

Notable Concepts & Terms

  • Vox Americanus: Sam's U.S.-ID-verified social platform built via vibe coding using Stripe's ID service, positioning identity verification as infrastructure (OAuth-style) rather than a standalone network.
  • GLM 5.2: Trillion-parameter open-source Chinese model from Z.AI that matches Claude Opus 3.5 in production use, signaling commoditization of frontier models and margin compression.
  • Vibe coding: Natural language interaction with AI tools (Cursor, agents) to build software without traditional file-by-file editing—lowers skill floor, making cultural fit more important than technical skill.
  • Bot school: Sam's 24-hour workshop where social media influencers with minimal coding skills built production-quality products using agentic tools—demonstrates accessibility of vibe coding.
  • Router (model routing): Infrastructure that routes token usage to the optimal model (cost/performance) across providers—CFOs are demanding this to cap spend; open-source routers are the strategic play.
  • CFOs are back in control: Shift from experimental AI budgets (no rules) to core spend (strict cost optimization), driving demand for bare metal (B200 stacks), open-source models, and routing infrastructure.
  • Fable 5 / K-Fabe: Anthropic's model halted by government then re-released after a deal—seen as either brilliant marketing ('this model is so dangerous') or trust-destroying theater. K-Fabe = staged drama to entertain.
  • You're Not Too Late: Sam's investment thesis (and book chapter): opportunities persist longer than expected; early mover advantage is overstated if tooling/distribution keep improving.
  • Alpha School: AI-driven education program Britt's kids tried—laptops evaluate skill level, adapt to 80% success rate via gamified learning, plus workshops (suturing, apps, greenhouses). Britt is cautiously optimistic but concerned about screen time and tiger parenting.

Operator Notes / Why Ken Should Care

  • For agent systems: GLM 5.2 commoditization means routing infrastructure is critical. Build swappable backends; optimize cost per task, not model brand. Identity verification (Vox Americanus-style) is infrastructure, not a product—OAuth approach wins.
  • For investing: Scrutinize margin durability in foundation model companies. Open-source parity (GLM 5.2 = Opus 3.5) is a structural threat. CFO-driven cost optimization will compress inference revenue in 6-12 months. Political/nationalist concerns may slow Chinese model adoption but won't stop commoditization.
  • For hiring/GTM: Cultural fit > technical skill in the vibe coding era. Prioritize communication, judgment, integrity over CS pedigree. If selling to non-technical users, emphasize invisible AI (Jess's point), not daily tinkering. Loneliness epidemic is real—design for community, not isolation.
  • For content/media: Vox Americanus distribution problem is instructive—verified identity raises quality but needs a bootstrap (celebrity, niche use case). The 'you're not too late' narrative (Sam's book chapter) is a wedge for late movers in AI app layer. Fable 5 release/halt cycle is trust-destroying but strong marketing—expect more regulatory theater.
  • For workflow: Audit where human judgment vs. agent execution matters. CFOs are demanding cost transparency; experimental budgets are over. If you're in a regulated industry, Sam's political infrastructure insight applies—reduce bureaucracy friction to attract talent.

Watch Map

  • 0:00-8:00: Padel/pickleball banter, e-foiling as a 'churny medium,' setting up the Information's Padel story. Skip unless interested in VC social trends.
  • 8:00-15:00: Vox Americanus launch: Sam explains Stripe ID service, OAuth strategy, distribution problem. Core insight on identity verification as infrastructure.
  • 15:00-22:00: Political infrastructure discussion: Sam on making elected office not suck (PC-12 network, turnkey staff). Debate on running for office vs. appointed roles.
  • 22:00-30:00: Vox Americanus deep dive: nationalism vs. bot filtering, use cases (dating apps, Amazon sellers), whether it's a standalone network or OAuth layer.
  • 30:00-35:00: Transition to AI news: Fable 5 back after government deal, K-Fabe/marketing speculation, trust destruction.
  • 35:00-42:00: GLM 5.2 commoditization: Dave reports teams swapping Opus 3.5 for GLM unnoticed, Nvidia spot rental down 22%, CFOs buying bare metal. Critical for margin compression thesis.
  • 42:00-50:00: Cultural fit over technical skill: Britt on 20-year Apple/Spotify founder asking 'isn't it too late?', Sam on bot school, vibe coding lowering skill floor.
  • 50:00-55:00: Fable 5 trust destruction, Alex Karp 'everyone hates Anthropic/OpenAI' comment on CNBC, speculation on whether it's brilliant strategy or chaos.
  • 55:00-60:00: Loneliness epidemic warning (Jess), 'you're not too late' thesis (Sam's book chapter), invisible AI vs. daily vibe coding debate.
  • 60:00-end: Taylor Swift wedding speculation (banned topic), Alpha School review (Britt), America's 250th birthday, lighthearted wrap.

Source/Metadata

  • Title: Why Identity Verification Matters For The Internet | Vox Americanus, GLM 5.2, Claude Fable 5 Is Back
  • Transcript words: 19376
  • Duration seconds: 3253
  • Timestamp note: Approximate timestamps provided based on topic flow; transcript does not include explicit chapter markers. Key segments: Vox Americanus (8:00-22:00), GLM 5.2 (35:00-42:00), cultural fit (42:00-50:00).
Search this video Resume transcript at 0%
Full transcript 10592 words · 86 min read
0:00

SPEAKER_03

Cultural fit, I think, matters way more now than ever before, when the ability to be a great engineer is much easier than ever before. [SPEAKER_04] I am going to play the normie on this. I like my agent. It's useful for work. I don't believe we are all headed for a future where we all need to be vibe coders. Anthropic started getting more and more hype in Q1. I think things got viral. They've done a good job of marketing. A lot of these businesses that were in the spot we were at in Q4, Q1 are just getting that marketing hit to them now.

0:08

SPEAKER_04

[SPEAKER_05] People who have smart, real opinions, but the internet doesn't give a shit about them. It's like user 87342, and you kind of look like a bot, and you have interesting stuff to say, but they're indistinguishable from a Russian bot. I get no attention, no voice, whatever. [SPEAKER_00] Dave and Britt plus Sam and Jess, put it all right to the test, More or Less. Why hello, friends. Welcome to More or Less. And we've got the full quad here, pre-4th of July. [SPEAKER_05] What's up? [SPEAKER_05] Hey, guys.

0:17

SPEAKER_03

Sam has decor in his background. Sam has decor. I like it. [SPEAKER_05] Yeah, I do have decor. Did you decorate your office yourself, Sam? [SPEAKER_05] Of course. I'm definitely in my office. That's called a macrame hanging, by the way. [SPEAKER_05] Didn't know that. Good to know.

0:30

SPEAKER_05

[SPEAKER_03] Yeah, now you do. [SPEAKER_04] What's up, Morans? Give us the lowdown, because it has not been the biggest of tech news week. We've got tech news. Don't worry. Plenty to talk about, but still, give us the update. [SPEAKER_01] Does that mean the summer's starting? [SPEAKER_01] No. [SPEAKER_04] Maybe. Wait, what is it? Isn't VC summer May to October? [SPEAKER_04] What's VC summer? [SPEAKER_03] I don't think in AI there's VC summer. I don't have a VC summer right now.

0:59

SPEAKER_00

[SPEAKER_03] There's no VC summer? [SPEAKER_01] No. [SPEAKER_03] What will all the e-foils do? Well, the LPs, I think, are all off, because they're all denominator effect, blah, blah, blah, waiting for the IPOs and liquidity. But VCs are still working hard.

1:09

SPEAKER_04

[SPEAKER_01] E-foiling is dead, I think. Why is e-foiling dead? No, I know. It was a bad reference. It's paddle that is... [SPEAKER_05] No, I was actually looking at the next e-foil to buy, because we destroyed our old e-foil.

1:15

SPEAKER_05

[SPEAKER_04] Oh, no. Really?

1:16

SPEAKER_03

[SPEAKER_05] Yeah. I think the kids need to get good at it. [SPEAKER_01] Do you think people got too hurt on them? What happened? [SPEAKER_05] No, I just think it's very simple, which is the e-foil is a great gateway, Doug. But if you get into foiling, they're stupid. Yeah. And so anyone serious gets off them. There's foil drive, which is cool. There's natural foil. There's a lot of foiling things that are awesome. But e-foils are worth about three days of someone's attention. And so I think either everyone tries it and spins out or tries it and upgrades. But there's no consistent... There's no way to stay on an e-foil, right? It's a very churny medium.

1:20

SPEAKER_05

[SPEAKER_04] You outgrow it pretty fast.

1:21

SPEAKER_03

[SPEAKER_05] The reason to buy it is for children. But then again, it's a $15,000 toy for children, which is...

1:24

SPEAKER_05

No. [SPEAKER_03] And it's too heavy. If they made it lighter, then kids and women could carry it. It's pretty dangerous. Pretty dangerous.

1:26

SPEAKER_03

[SPEAKER_05] They have. I've looked at this. There's the new one. There are two new Lyft models. One is much lighter, looks great, but it has an unshielded prop, which I'm fine with, but defeats the kid purpose. And then the shielded prop one is still big. So it's a great thing to own as a gateway drug, but there's no consistent usage past them is the problem.

1:28

SPEAKER_05

[SPEAKER_04] Okay. But do you guys know what... I think our story will have come out now, but the Information will have a blockbuster story around this time about the rise of Padel among the venture capital community. And I promise you, this is a story you want to read because we broke down the coolest clubs. [SPEAKER_03] Not again. Ugh. They're all in New York. San Francisco has not gotten into Padel yet.

1:30

SPEAKER_03

[SPEAKER_01] I'm sorry, Jess, but Padel and Pickleball are the snowboarding of tennis.

1:30

SPEAKER_04

No, Pickle is. Pickle is. Padel has a slightly more warmer relationship with tennis. [SPEAKER_05] East Coast vibe? [SPEAKER_05] Well, it's European. [SPEAKER_03] They called it Padel just to sound like it's classier. It's like I went shopping at Target today. It doesn't actually make it classy.

1:36

SPEAKER_04

No, you can call it Padel or Paddle. It doesn't really matter. Anyway, we have breaking news about Philippe Lafad's latest placement in the Padel tournament. And the entire New York to Miami tech scene, Dara Karshawi, did I say his last name? Of Uber. You've got, what is it? Dan Sundheim, the founder of D1 Capital Party. These people are obsessed. So this is all the rage. I hear the hottest. [SPEAKER_01] It sounds very East Coast to me.

1:39

SPEAKER_01

Same. [SPEAKER_04] It is.

1:41

SPEAKER_04

No, you can call it Padel or Paddle. It doesn't really matter. Anyway, we have breaking news about Philippe Lafad's latest placement in the Padel tournament. And the entire New York to Miami tech scene, Dara Karshawi, did I say his last name? Of Uber. You've got, what is it? Dan Sundheim, the founder of D1 Capital Party. These people are obsessed. So this is all the rage. I hear the hottest. [SPEAKER_01] It sounds very East Coast to me. Same. It is. But there are new clubs opening in San Francisco, which you'll have to learn about. Neil Meadow of Green Oaks is pushing this. Is this like the new golf? [SPEAKER_01] Isn't it just outdoor racquetball?

1:46

SPEAKER_03

[SPEAKER_04] It's a little bit different. And the ball isn't as dangerous as a racquetball. And indoor. You can play indoor. [SPEAKER_05] They're not racquetball. Not dangerous. I think it's like, look, here's the upside. People... Sam, please clarify this for me. [SPEAKER_04] A year from now, we will all be members of clubs in the Bay Area. Mark my words.

1:52

SPEAKER_01

[SPEAKER_05] No. It's very simple. The media needs a new racquet narrative. And so there has to be a new racquet sport to feed that. Obviously.

1:52

SPEAKER_03

[SPEAKER_04] I could just milk tennis for all it's worth. [SPEAKER_05] That's what I'm saying. What's wrong with tennis? No, I'll explain. Tennis is actually a very difficult sport. And it's great. It's the best. But it's very hard, right? And it requires a lot of space and dah, dah, dah, dah. So pickle splits off as effectively ping pong for people who want to pretend they're playing tennis. But if you're a fancy person, you don't want to associate with pickleball players. You know what I mean? It's disgusting. Yeah. It's snowboarding. [SPEAKER_04] I want a $30 kit smoothie. That's what I want.

1:58

SPEAKER_03

[SPEAKER_05] To be clear, I actually came in second in a pickleball tournament last year. So I actually got a great prize for it. [SPEAKER_04] Who won one? Who won one? [SPEAKER_05] Oh, that was a friends one. This was a non-friends one. But Jessica was very good too. But people who are, the founder of D1 Capital doesn't want to associate with the plebs that play pickleball. That's disgusting. And so you have to come up with a new version. It has the benefits of pickle, but can be more elitist. It's the most East Coast banter of all time.

2:05

SPEAKER_01

[SPEAKER_04] Guys, it's great. It's coming to the Bay Area is my point.

2:07

SPEAKER_04

[SPEAKER_05] And so now we have elite non-pros who will play Padel. You have committed good athletes who will play tennis. Yep. And then you have everyone else who's playing ping pong on steroids, which again, is quite fun. I am not actually anti-pickle, but I also love McDonald's. [SPEAKER_03] I'm going to stick in the poker league. I'm going to be part of the tech community that's in there. Have you gone full Mahjong yet, Britt? I can now play. [SPEAKER_03] I know the rules. I know that's the other one. All the girls are doing Mahjong. It's actually, there is a poker analogy here, which is poker is a great game.

2:11

SPEAKER_05

It's hard. It's hard to play poker at an elite level, right? Yes, it is. And it's the tennis of card games. And so I don't know, what is the pickle of card games? Is it spit? There's got to be like any- [SPEAKER_03] Speed.

2:16

SPEAKER_04

[SPEAKER_05] What is it? Spoons.

2:17

SPEAKER_05

Spoons. And then there's the elite, which is getting back into hearts, which is fun. And then there's an elite level, which is getting back into bridge, which is really hard, right? And so you just have this differentiation set. Everyone needs a community. Everyone needs a cult. And they need to create differences amongst each other. And then the media loves it because they can write stories about it and make money. [SPEAKER_04] This is going to be a blockbuster for us, but it's also a really important trend because for all of you VCs who need to know how to network, you have to pick up the sport, I promise you. Or you, founders who want to network with VCs.

2:18

SPEAKER_05

If you want to hang out with really rich people, this is a great sport to learn.

2:20

SPEAKER_01

You don't need to take it up to hang out with me. I'll hang out with you at a coffee shop in San Francisco. [SPEAKER_04] And you have to go to Mill Valley to hang out with you. Dave, you have to go to Mill Valley. You're going to make people go to the lodge in Mill Valley.

2:23

SPEAKER_05

But Dave, you're not that rich. If you want to hang out with really rich people. [SPEAKER_04] What an asshole. No, I'm just like, there's a spectrum. I'm not either. I'm not. I didn't found D1 Capital. There's a spectrum. You and I are just on the... [SPEAKER_01] I don't even know what D1 Capital is. [SPEAKER_04] Guys, they're such a big tech. They're hugely important investors. The basic point is this. It's like, there's always these shortcut codes. It's like, if you happen to play polo, right? It turns out that you're going to end up hanging out with a bunch of really rich people because they want someone to play on their polo team, right? There's just a thing, right?

2:34

SPEAKER_05

[SPEAKER_04] Okay. There's a spectrum. You and I are just on the... [SPEAKER_01] I don't even know what D1 Capital is. [SPEAKER_04] Guys, they're such a big tech. [SPEAKER_04] They're hugely important investors. The basic point is this.

2:51

SPEAKER_04

[SPEAKER_05] There's always these shortcut codes.

2:52

SPEAKER_05

If you happen to play polo, right? It turns out that you're going to end up hanging out with a bunch of really rich people because they want someone to play on their polo team, right? There's just a thing, right?

2:59

SPEAKER_03

[SPEAKER_04] Okay. [SPEAKER_04] This is relevant because the networks and tribes and power centers are relevant. [SPEAKER_05] Look, I don't even have enough people who can go heli skiing with me. [SPEAKER_05] It's the same thing.

3:06

SPEAKER_05

It's if you can heli... [SPEAKER_01] Or just... [SPEAKER_01] Here's another idea. [SPEAKER_01] Just make an amazing product and everyone will come to you. [SPEAKER_01] Yeah, but that's hard. [SPEAKER_01] It's much easier to learn Padel. [SPEAKER_04] If you can get 60 billion in ARR in year three. [SPEAKER_04] Okay. [SPEAKER_04] Speaking of amazing products, there has been a product launch among this group this week.

3:29

SPEAKER_04

And it has not been me. Although our app is out there. If anyone wants to try it. Jess's app is good. Sam, would you like to talk about your new product?

3:48

SPEAKER_03

[SPEAKER_01] I like your app, Jess. Thanks guys. I think almost all of us are launching a new product every week, but for what it's worth... [SPEAKER_01] It's just that the barrier is so low.

3:53

SPEAKER_01

Are we talking about Sam's America thing?

3:58

SPEAKER_04

We're talking about Vox Americanas. Vox Americanas.

4:06

SPEAKER_05

Vox Americanas. Vox Americanas.

4:09

SPEAKER_03

[SPEAKER_05] So look, here's the upshot. [SPEAKER_01] Which I feel I should get a little tiny bit of credit here. [SPEAKER_05] We can all take tiny bits of credit for all sorts of things.

4:16

SPEAKER_04

[SPEAKER_05] Here's the deal. [SPEAKER_05] I have never understood why Twitter, for instance, doesn't just launch the validate my ID to get a special, an American who lives in California. [SPEAKER_05] Or at least I have a driver's license in California ID. [SPEAKER_05] It makes no sense to me. [SPEAKER_05] This is not something that used to be really hard to do. [SPEAKER_05] There's now so many services that make this easy. [SPEAKER_05] I don't really understand why the big platforms don't do it. [SPEAKER_05] It's not that expensive.

4:37

SPEAKER_04

[SPEAKER_05] When you think about the quality of online discourse, if you don't want to talk to Russian bots or whatever, it's a really easy way to raise the bar on fake accounts and stupid people and whatever a little bit. [SPEAKER_05] Right. [SPEAKER_05] So I think that's probably what should happen at some point.

4:40

SPEAKER_01

[SPEAKER_05] But in the spirit of it's 2026, I was like, look, I'll just make my own. [SPEAKER_05] Right.

4:43

SPEAKER_04

[SPEAKER_05] And so this is a classic vibe coded plus a little secret Sam sauce. [SPEAKER_05] But let's be honest, mostly vibe coded app where I was thinking, look, Stripe is an ID service. [SPEAKER_05] It's not expensive. [SPEAKER_05] Why don't I just make it so that you basically have to sign up and you have to have a license on file that will tell what state you are, that you have an American ID.

4:51

SPEAKER_01

[SPEAKER_05] And what your real name is.

4:53

SPEAKER_04

[SPEAKER_05] And then let's just have a dialogue with real Americans where we can actually talk about things we care about with people who are actually in America and not quite citizens, but directionally citizens versus random bots and getting mad at them on the internet. [SPEAKER_05] Now, the interesting thing, which I'm figuring out, and I'm going to play with this more. [SPEAKER_05] There's a few things. Actually the first place I started was doing full anonymity. [SPEAKER_05] So you prove you're in America and you prove you're in California, but no, and there's some benefits to that, but it gets confusing and complicated.

4:57

SPEAKER_04

[SPEAKER_05] So the version I'm working with for now is just...

4:58

SPEAKER_05

[SPEAKER_04] You're Mr. OG on real identity. I am an OG. I really care about this. I think this is a fundamental problem of the internet. The reason the internet worked was because there was no identity, which made it super easy to join.

5:05

SPEAKER_04

[SPEAKER_05] This goes all the way back. [SPEAKER_05] It was born without any sort of identity layer.

5:09

SPEAKER_05

Fine. We have now spent the last 30 years trying to bootstrap an identity layer on top of it to fix that, because when it actually becomes important, it really matters. If the person flaming on the internet is real or not, it really matters if they're American or not. It's a very different vibe. And so, look, I was very hopeful.

5:16

[SPEAKER_05] There was an era where Facebook was super focused on this. [SPEAKER_05] And I think Facebook, original Facebook, Harvard Facebook, did a really good job. The thing that made it work was bootstrapping. I think it was bootstrapping real identity in an unsafe space. That was it. It was internet scary, Sandra Bullock the net. Oh, shit. We can bootstrap off of fast.harvard.edu and then different schools and have trust in real names and real people. And that made it a safe, friendly, inviting place where you could have real conversations with real people. We're now 16 generations past that. And with bots, this is a really big problem.

5:28

SPEAKER_05

And so I do think candidly for this to be solved at scale, it's a platform problem. It's not me building Vox Americanas. But I also think the only way to play and advocate and figure it out is to build it. And so it's cool. It works. I got to give credit to Stripe's ID service. It's great. [SPEAKER_04] Very easy sign up.

5:44

SPEAKER_04

[SPEAKER_05] It's so easy to sign up. [SPEAKER_05] You put in a license, you twist your head a few times, and it's like, yeah, that's a real person with American ID.

5:47

SPEAKER_05

[SPEAKER_03] So Sam, what's in it for me, though? [SPEAKER_03] Why do I need to rush to sign up?

5:53

SPEAKER_04

[SPEAKER_05] Well, so this is the interesting thing, right? [SPEAKER_05] And I think this is one of the things I'm going to struggle with as I evolve this and play with it.

5:55

SPEAKER_05

The people for whom this is probably most relevant, right, are people who have smart, real opinions, but the internet doesn't give a shit about them, right? [SPEAKER_04] Very easy sign up. It's so easy to sign up. You put in a license, you twist your head a few times, and it's yeah, that's a real person with American ID. [SPEAKER_03] It's wait, so Sam, what's in it for me, though? [SPEAKER_03] Why do I need to rush to sign up? Well, so this is the interesting thing, right? And I think this is one of the things I'm going to struggle with in terms of as I evolve this and play with it.

6:10

SPEAKER_05

The people for whom this is probably most relevant, right, are people who have smart, real opinions, but the internet doesn't give a shit about them, right?

6:13

SPEAKER_04

[SPEAKER_05] They're user 87342, and you kind of look like a bot, and you have interesting stuff to say, but they're indistinguishable from a Russian bot. [SPEAKER_05] And so you're I get no attention, no voice, whatever.

6:16

SPEAKER_05

Someone like you or me, we have enough of an online identity and a following that it isn't that valuable for us to post to just Americans. We might as well just post to everyone. And so there is that is the fundamental question is in theory, I would love to have a place I put things out, and I actually care about the feedback and the engagement because they are actual fellow Americans, right? That is the thing I want, and I don't really want to put things out and get flamed by a bunch of bots, which I also do. But there's that subtlety. The first principle, if this makes sense, is great.

6:34

SPEAKER_03

[SPEAKER_05] The budget, I want to be taken more seriously on the internet, and so I validate my idea on Twitter, is obvious and it's annoying to me. [SPEAKER_05] It doesn't happen.

6:38

SPEAKER_04

[SPEAKER_05] But then the, what, if you really try to bootstrap a new service, the exact use case of why I care enough is the interesting question, right? [SPEAKER_01] I think we've talked about it several times on the pod.

6:41

SPEAKER_03

[SPEAKER_01] One of my favorite hobby horses is the question of why doesn't America do this on two different levels? [SPEAKER_01] Why doesn't America provide and then protect American digital identity on a global basis? [SPEAKER_01] Yeah. [SPEAKER_01] But then secondly, why don't we do a better job of protecting the American internet from rogue external actors, but also, because it's free speech inside of America doesn't mean you should be able to be a citizen from the internet.

6:48

SPEAKER_05

[SPEAKER_01] And so you should be able to get people from outside of America and come inside of our internet and mess around inside of our internet, right? [SPEAKER_01] And so the platform question you mentioned is one that I've never really understood why, say, Facebook doesn't care more about this. [SPEAKER_01] It might actually improve the quality of the discourse on, say, threads or original Facebook if they provided a space like this. [SPEAKER_01] But I get it. They don't have the incentive, right? [SPEAKER_01] It's complicated. And look, we've talked about this intellectually a thousand times, but it's 2026. You just build stuff.

6:57

SPEAKER_05

[SPEAKER_01] And the reality is I tinker and it is fascinating. The big delta from today versus even a few years ago is it's shockingly easy to validate IDs, right? The APIs are now really robust for it. There's Persona, there's Stripe. It's so easy to do that rather than argue about this intellectually, the better thing is just to build stuff and see what you learn. Because everything you're saying is right. And it is a hard set of product problems. But I think we're just at a moment where you do stuff. Sam, maybe you should run for office and use this as... Oh, James, do it. Do it. What are you talking about?

7:03

SPEAKER_05

Maybe you should extend your Harvard run from last year and all of your learnings to a higher level run for office. And Vox Americanis should be the platform that you carry with you. [SPEAKER_05] So I've had a few people very flatteringly reach out and tell me I should run for office in various forms. Here's the interesting thing that I've found. I'm on the record. I think you should. A lot of people go out there and they say, especially on the Democrats side, because they're so lost. And they say, hey, we need to find young-ish, credible-ish people with reasonable views.

7:12

SPEAKER_05

And they're like, we'll help you with your media strategy and campaigning so you can run for office. What I've said to these people is you guys have this backwards, right? In two ways. First of all, in 2026, you can't take someone who's the nice guy liked in the community and build a media strategy for them. You need them to already have a media footprint themselves. [SPEAKER_01] That's why you're perfect. But then here's the second part of my talk track is one, you can't just find good people and give them that. You're not good enough at this. [SPEAKER_04] And there are lots of people. Find celebrities and well-known people that already have that.

7:18

SPEAKER_05

[SPEAKER_04] And then here's the biggest thing I think I keep saying to people is the thing you have to solve if you want people to run for office is not the campaigning. [SPEAKER_04] It's make the job not suck. If you had a nonprofit infrastructure, which it looks, Sam, here's the deal. You should run for whatever.

7:23

SPEAKER_04

[SPEAKER_01] And you have the media voice. We'll have you develop, on the margin, a few connections, whatever. [SPEAKER_01] But the real reason that you should do this is you know the job sucks, but we will provide all the infrastructure, the bill writing, the relate. [SPEAKER_05] We'll do. We'll basically create a platform where you can just be the candidate to make the final decision and we'll be the huge infrastructure behind it. [SPEAKER_05] That works so much better. I thought that's what every institute in D.C. is. No, I think that they're still oriented on the side.

7:35

SPEAKER_05

I don't the job sucks too much. I think if you started from the perspective, I want the job to suck zero.

7:38

SPEAKER_01

[SPEAKER_05] How do I make politics an appealing job? And then you backfill and get the candidates versus being, well, we'll help you campaign to get the job. [SPEAKER_05] And then you have to do the job. I think the job only sucks for people who care what people say about them.

7:42

SPEAKER_04

[SPEAKER_05] If you don't care about all of the drama that surrounds being in the public eye, then the job might not suck. [SPEAKER_05] But I'll give you a really basic, silly example in California. I know some people who are running for office in California. [SPEAKER_05] And what a shitty life trail. You have to be in Sacramento all the time. It's the worst.

7:49

SPEAKER_05

I'm not talking about California. I'm talking about America. Yeah, but let's just expand this. D.C. is the same problem. You're like, look, here's the deal. Yeah, you have to be in Sacramento all the time. But we have a PC-12 network and we will just make this an afternoon for you rather than trying to get there on your own. Right. That's the type of infrastructure. I want briefings done for me. I want my office already set. I want everything turnkey so I can drop in for key moments effectively. And I don't want to act. I think that's the thing is the job. You pay shittily. I'm not talking about California, man. I'm talking about America.

7:58

SPEAKER_05

Yeah, but let's just expand this. D.C. is the same problem. You're like, look, here's the deal. Yeah, you have to be in Sacramento all the time. But we have a PC-12 network and we will make this an afternoon for you rather than trying to get there on your own. Right. That's the type of infrastructure. I want briefings done for me. I want my office already set.

8:01

SPEAKER_01

[SPEAKER_05] I want everything turnkey so I can drop in for key moments effectively. And I don't want to act. I think that's the thing—the job pays shittily.

8:03

SPEAKER_04

[SPEAKER_05] It's a shitty life. It's away from your family. It's all these things you'd have to solve for. [SPEAKER_05] You just look at the reason that we can fix politics is we'll solve all the bad parts for you of the actual job. Then I think you get better candidates.

8:07

SPEAKER_05

That's a fair point. You do have to be on the road. That's probably it—it's like being a musician. The life is extremely difficult. You're just on the road all the time. It's awful. And even another friend who ran for a big office recently didn't see their kids forever. And you're like, why would you do that for yourself? For what? For a low paying job? What about what David Sachs did or what Sriram did? I mean. Those aren't real jobs. That is a version of it. But you can have impact. Right? You could have impact. I mean, Sam, let's say a libertarian Democrat won next election, which is highly unlikely.

8:21

SPEAKER_05

But you could go do that kind of role and have a serious impact on this exact idea.

8:21

SPEAKER_04

[SPEAKER_05] I get it. I just think there's a deeper thing to solve here. Right? Which is the reason why don't good people run for office? [SPEAKER_05] Whether or not I'm a good person is a separate question. But I think everyone focuses on that—well, campaigning sucks and they don't have the media strategy and voice.

8:26

SPEAKER_05

[SPEAKER_01] I actually think you guys just acknowledge the fact the actual job sucks. Let's solve for that. And then you can backfill and do the other things and find the right candidates. The thing I would want to be marketed is this job actually sounds awful. But if you will provide all the infrastructure so that it's a great job. If that makes sense. Then you figure out the rest.

8:30

SPEAKER_01

[SPEAKER_05] I think you should have a writer like celebrities do when they do concerts. I'll do your job, but here's my writer. [SPEAKER_05] And I do think there's a difference between the electives. Exactly what Americans want from their tech politicians. [SPEAKER_05] Hey, if they're gonna get a 10x increase in intelligence, I think America would do it. And by the way, I also think you should be a czar. [SPEAKER_05] I think David Sachs was onto something with the czar title. But the thing about that is I do think to Jess's point there are people who have figured this out how to be in the government without being in the government.

8:37

SPEAKER_01

And that is what they've done. The problem is those are not elected jobs. They're nowhere near—you get to say some stuff and be in there with some meetings.

8:38

SPEAKER_04

[SPEAKER_01] But it's a vice chairman role at best. And those are actually not what you need. You need better fundamental politicians is my view. Right. [SPEAKER_05] Britt, do you have any political aspirations? Britt, you could do it. [SPEAKER_05] People have also told me that I should run for office. I've always thought Dave would make a great first gentleman. But. [SPEAKER_05] Dave would make a great first gentleman. I'm in. It's not in my next decade. I might reconsider in my 50s. We'll see. [SPEAKER_05] Man candy number one. I'm definitely never doing it, but I'm happy to be the first gentleman.

8:50

SPEAKER_04

[SPEAKER_05] I'm gonna put Dave and his agents on my opposition research team. [SPEAKER_01] Oh, crazy. And then he's gonna figure out all the stuff to talk about. [SPEAKER_01] Oh god. About the other people. This is why we're never doing this.

8:55

SPEAKER_01

[SPEAKER_05] You just want Dave's agents. I mean, I think the reason this conversation was interesting is it's about distribution, Sam.

8:57

SPEAKER_03

[SPEAKER_05] Your idea is amazing. I actually want to participate in Vox Americana at scale without question. [SPEAKER_05] I think it would be deeply healthy for the American populace and for society.

9:02

SPEAKER_01

[SPEAKER_05] But the question of distribution. That's why I threw that idea out. [SPEAKER_05] Would Trump Social or Truth Social have ever gotten to scale if it weren't for the cultural movement of MAGA, etc.

9:07

SPEAKER_04

[SPEAKER_05] Well, but I'll actually go a step further today. Truth Social got there because they had. [SPEAKER_05] There's a type of content I want. I can't get on the mainstream platforms. Right.

9:10

SPEAKER_05

And the question is very simply for any platform: you have to provide something where I'm coming here because there's something I want that I fundamentally can't get anywhere else. And I don't. There's value to the ID registration and there are a lot of things. But I have to play with how to figure if there's anything new in steps there.

9:12

SPEAKER_01

[SPEAKER_05] It's easy for OnlyFans. Easy. Right. I know exactly the content I'm getting nowhere else. MAGA world. Easy.

9:17

SPEAKER_05

I know what content I'm getting. I think this carries that. I think this carries the idea that I'm definitely talking to Americans. I think that's really interesting. I understand. I think that's true. I just. There's a subtlety to what content does that generate? The medium is the message. [SPEAKER_01] What is the thing I'm coming for? Other than yes, I do want comments from real Americans. [SPEAKER_01] I'm actually quite interested in what real Americans say. I think it fits into an important political narrative right now. Right.

9:37

SPEAKER_05

[SPEAKER_01] Which is like, who is American on the internet. Right. It fits into the fact that most of us interacting with algorithms on TikTok are not controlled by Americans. People that aren't Americans. This dovetails into the immigration conversation. I mean. I obviously deeply agree. I think the problem is though, and I think this is the benefit of the problem. I think every thinking person recognizes that if you truly had the American internet or the French internet, this can be global in various ways. It is so much more moderate than what you get on Twitter. Right. It would be so much more moderate because it turns out. I don't know about that.

9:41

SPEAKER_05

I mean, the human internet versus the bot and human internet is one thing. But I found it interesting that this goes back to TikTok. Both David and Sam, you guys feel this nationalist. I want to talk to Americans. I obviously deeply agree. I think the problem is though. And I think this is the benefit of the problem. I think every thinking person recognizes that if you truly had the American internet or the French internet, this can be global in various ways. It is so much more moderate than what you get on Twitter. It would be so much more moderate. Because it turns out that I don't know about that.

9:47

SPEAKER_05

The human internet versus the bot and human internet is one thing. But I found it interesting that this goes back to TikTok. Both David and Sam. You guys feel this sort of nationalist. I want to talk to Americans.

9:57

SPEAKER_05

I understand wanting to talk to real smart people. But Jess, it's twofold. It's not just talking to Americans. It's not being influenced by nation state activities of external nations inside of these networks. They're different things. You post, I post something, or more extreme, like Joe Lonsdale, or I just talked to Joe today. I love Joe. Sean McGuire. Post something. The amount of vitriol and hate that you get back on it is wild. And you kind of look at it. You're like, look, this is 95% bots. But some of that vitriol and hate is from Americans. But I'm not sure how much. And I'm actually not sure. I think it's actually way smaller than you realize. If that makes sense. So you're using it as a proxy for a filter. Well, no one knows. To double click on this, I had not confronted the bot army recently until as we talked about on the pod a few weeks ago.

9:58

SPEAKER_05

I spoke at this Dell World Conference in Las Vegas and they posted a video on their account and tagged me. And the unbelievable backlash against AI that filled my Twitter feed for days on end. And every time I would click on one of them it was a short name. They were verified accounts. But they all had these long set of numbers.

10:01

SPEAKER_05

[SPEAKER_04] On the end of it. And it was very clearly bot mediated. I don't know if it was computers or whether it's people in other countries. But there's very clearly an agenda happening here. Of course. And so the problem is the thing. I actually, I'm actually quite. Data center is a great example. I'm actually really curious how actual Americans think about this. You know what I mean. And I'd actually be willing to engage with actual Americans on it. But the dumbest thing in the world is fighting bot state actors on the internet and having flame. What are they doing?

10:05

SPEAKER_05

[SPEAKER_04] They look incredibly insane. When you see people responding to bots with cogent arguments. You're like, that bot is literally just wasting your time. And so for me, that's the thing. I was like, I really think there is a place for this type of conversation. But the thing I'm quite unsure about is how do you spark more than intellectual interest in it. You know. It's a good point. I think you have to broker a deal with one of the social networks that claims you'll get more distribution if you sign up for this thing and verify. That is.

10:05

SPEAKER_05

100% what it should be. There's no question that that's the way the world should work. And I only understand why it doesn't work that way because it's not that hard to do anymore.

10:10

SPEAKER_05

[SPEAKER_01] Or maybe partner with people who are running for president. Like whoever's running for president, be the partner. So the other thing I've been playing with around Vox Americanas and what to do with it is to just be like, look, it actually shouldn't have its own. I mean, it can have a feed, whatever, but it really is more OAuthy. And you think of it as a way to prove Americanness, right? And reality and off a bunch of accounts and who you actually are in scenarios that are social. There are ways you could think about that. I love that. That's what I thought it was going. Yeah. It's like you could do it. I was thinking, Dave, you might like this. I was thinking about how certain services, if you want to prove this is your Twitter account, you post the link and you crawl it and you basically create an index of actual Americans on Twitter. Is that the better strategy? There's a thing to play with. This is my 250th or Tinder. I bet all the dating apps would want this, right? Yeah. You just create a proxy layer. Can I tell you where I want this the most? Yeah. I want this on Amazon. I want to filter to only American sellers on Amazon. Only American sellers. Yeah. And look, I think for what it's worth, this scales globally. I don't know what the French auth services are, but they have the same thing. We're in this era of nationalism. For what it's worth, I'll take French sellers, Italian sellers. I'll take all Europeans and Japanese. All the highest quality places where craftsmanship matters. And the reality is this is completely against the original, one world open market thesis, that utopian end of history. Right. I don't think it is, Sam. I think it's totally fair to want verification of identity and to understand. I remember growing up in Montana, there was a sticker that people would put on goods that they made in Montana that said made in Montana and Montana actually invested a lot in supporting those businesses and helping them grow. It's really no different. The internet is huge, global, and amazing. I mean, the speed at which open claw is growing right now is blowing my mind. The scale of the internet's enormous, but I don't think there's anything wrong with wanting to understand who you're talking to or where a good is coming from. A hundred percent. I just think for me, the moment, especially because vibe coding and a lot of this stuff is just, I mean, I actually, if you go to Vox Americanis and use it, it is actually pretty good. It's great. You can do video live recording. It's got a great NUX. It's pretty better than Truth Social was at launch. It actually is pretty good. And the fact that I could build that in a few days, but honestly on boats, texting my bot for changes is pretty sweet. And it's like, you, I just think we're in this moment of just stop arguing about this stuff and build things and tinker. And I think that's the most uplifting part of all this AI is not the tinkering. For what it's worth to shift to a different topic, I think this is a really important thing that Britt and I were on the phone yesterday with one of our founders. And this is a founder who spent years at Apple and Spotify in product roles, interfacing with engineers.

10:15

SPEAKER_05

but honestly on boats, texting my or emailing my bot for changes is pretty sweet. And it's like, I just think we're in this moment of just stop arguing about this stuff and build things and tinker. And I think that's the most uplifting part of all this AI is not, is the tinkering. For what it's worth to shift to a different topic. I think this is a really important thing that Britt and I were on the phone yesterday with one of our founders. And this is a [SPEAKER_03] founder who spent years at Apple and Spotify in product roles, interfacing with engineers. He's definitely has 10,000 hours working with engineers, right? 20 years of experience, very talented founder.

10:29

SPEAKER_05

[SPEAKER_03] And it's interesting how much people who were not engineers believe they don't have the skillset to work in the world of agentic engineering now. And I found myself telling him you are more capable than you think today to do this. And that I think that people really don't realize how powerful the

10:41

SPEAKER_05

[SPEAKER_03] tools are and how much more powerful they've even become even in the last three months. And that they should just try it. Yeah. It's wild. So it's funny. I distinctly remember last summer being at the same place we're at now and being in cursor and just being like, oh my God, this stuff finally works. Right. And I compare that to now and it's actually insane, right? Like the level of abstraction you can work on, it's mind blowingly different. Before I was like, MVC frameworks and editing file by file. And now I'm like, I don't need to do any of that. I literally sit here and email my infrastructure and spin up everything. Right. And it's incredible.

10:45

SPEAKER_05

[SPEAKER_01] And so when you do that, you're just like, this is an age that is perfect for tinkerers. And this has been like my most uplifting experience was running our bot school, right? Where we basically took all of these social media influencers, people with big cults and followings. We sat them down, we gave them the tool chain, how to use digital ocean that are really, and it was amazing. What they produced in 24 hours, I remember hackathons where you'd come out and something really half working in four days. Now you can come out to 24 hours, products you can ship to your actual users, and they're good. It's wildly different. And

10:50

SPEAKER_05

[SPEAKER_04] I think that is the most uplifting part. Like the mega wars, like who's got the cheapest way to multiply numbers is all boring. But the empowerment of people with minimal skills to build entire worlds is wild. The thing I was going to also say is in that conversation yesterday with this founder talking about what it came down to at the end, he's trying to recruit his founding engineer, everyone's great, a lot of people are great engineers, of course, some are better than others, fine. But cultural fit, I think matters way more now than ever before, when the ability to be a great engineer is much easier than ever before. And so it's like teaching these people how to be human, maybe a little throwback to etiquette school, but also like, having street smarts being a cool person, high integrity, vibing with the team, I think is going to matter more than ever over the next few years for teams.

10:59

SPEAKER_05

Yeah, I completely agree.

11:01

SPEAKER_04

[SPEAKER_05] It's a wild moment. It's a great moment. This stuff is the fun stuff.

11:04

SPEAKER_05

I think it's important to note that it's just not too late. Like that's what talking to this founder, I was shocked by how he's like, isn't it too late? I'm like, No. The skill set of this new style of engineering is a brand new skill set that, it's like, as if it used to be you had to speak a language that a compiler could understand, you no longer have to speak a language that a compiler has to understand. Now you just have to be able to speak a language that the models can understand. The models speak compiler. And the compiler speaks to the assembler, the assembler speaks machine language, and everything's, it's amazing. And it's a totally new skill set, and you're not too late. And though we talk about this a lot on the pod, I also hear from people every week that listen to the pod that are entrepreneurs thinking about starting something. And I just think it's really important for people to hear that you're totally capable, and it's definitely not too late. And take an hour a day and just really try to learn this stuff, because everyone's learning it right now.

11:05

SPEAKER_05

Although I don't think so I agree with everything you said, except for one thing, I don't think you take an hour a day, I think you binge it for as much time as you can. Well, Sam, I'm saying if you've got another job, if you're doing something else, right, try to find, it's like learning guitar or something, you got to find time to practice every day in order to make progress. Sure, if you can really binge into it, go for it. But

11:08

SPEAKER_05

Except for please stay human, because I really think we're entering the way I know we're already in the loneliness epidemic because of social media. But this is entering this new, scary, dark place, I think where people are alone with their agents up at all hours, not having community and real social conversation. And I'm really actually worried about it. I am going to play the normie on this. I like my agent. It's useful for work. I don't believe we are all headed for a future where we all need to be vibe coders. I'm somewhere in the middle. [SPEAKER_01] We know it's your opinion. We had that argument two weeks ago.

11:16

SPEAKER_05

[SPEAKER_01] Sam wasn't there, though. Sam didn't get to weigh in on that. [SPEAKER_01] I would just want the listeners to know that I think that, we're in this exciting early adopter moment. There's a lot. That doesn't mean I think AI won't touch everyone. It doesn't mean that I don't think AI is transformative. It's begun to become invisible is the thing that I said before, which I still stand by, which is, you won't have to be on your computer vibe coding all night. But AI and agents will be doing stuff for you no matter what. Cool. But does it mean that I have to devote two hours of my day to vibe coding or I'm going to be left behind?

11:31

SPEAKER_05

I think it's somewhere in the middle. It's pretty powerful stuff. We'll see. It's going to be an interesting era. I think for now it's super fun if you have business problems you're trying to solve. I mean, Dave's line on you're not too late, I think is a really important one. I actually hilariously started because it's the summer. I'm going to write a short Sam book that I've But AI in agents will be doing stuff for you no matter what. Cool. But that's fine. But does it mean that I have to devote two hours of my day to vibe coding or I'm going to be left behind?

11:37

SPEAKER_05

I think it's somewhere in the middle. It's pretty powerful stuff. We'll see. It's going to be an interesting era. I think for now it's super fun if you have business problems you're trying to solve. Dave's line on you're not too late, I think is a really important one. I actually hilariously started because it's the summer. I'm going to write a short Sam book that I've outlined on what I've learned from the investments I've made. And one of the chapters is called You're Not Too Late. Because I think it's a mistake even on investing I've continuously made.

11:43

SPEAKER_05

[SPEAKER_01] And I've gotten a few things right where there's this class of investments where you're like, well, I've seen this for two years. I can't believe I missed it. I loved it, but I'm too late. And it just turns out you're not. Right. And that's one of my chapters in the book I want to write this summer.

11:46

SPEAKER_05

[SPEAKER_01] Well, that's because we're seed investors. And if we broke ourselves of that mentality, I think we would learn that faster. Speaking of bots, I did log into mine today to see Fable 5 is back. So for those of good marketing from Anthropic, they have reached a deal with the United States government. What incredible marketing. Well done. It's great marketing. Until July 7th, you can use up to 50% of your plan's weekly usage limit on Fable 5. Which is interesting, actually. They're basically like, after that, you got to pay by token. Which is interesting. What if I don't, the stuff I'm doing, I definitely don't need Fable 5 for.

11:50

SPEAKER_05

You never know. That's actually the whole business model. The beauty of the whole thing is you're almost certainly right, Jessica. But what if? The entire margin of the industry is based on people asking the question. Look, I'm asking you to fold laundry is a funny one. I'm asking you to make a dinner reservation. An IQ 100 person can do it. But do I need to hire an IQ 170 person to make a dinner reservation on the off chance it matters? That is the margin of the system. And we'll see how that plays out. I definitely don't need Fable 5. But it's back.

11:51

SPEAKER_05

This back and forth with the government is wild, I think. And for those who forgot, this is Fable 5 was released along with Mythos. In this case, Amazon raised some doubts about some ability for it to be jailbroken. Anthropic essentially said it's no more vulnerable than other things. Amazon isn't as smart as us that it got in this whole back and forth. But it's out now. There's some security thing Anthropic added. But we're just going to see this for every single model release. Yes. It's weird. Isn't it weird? Do we? I don't know. How do we feel about this? I don't feel great about it. About what? That we need the government to sign off on models?

12:02

SPEAKER_05

About what? I feel great about what specific, what specifically?

12:05

SPEAKER_05

Yeah, it's a good question. I mean, on one hand, I like that there's someone checking out for the safety of these models. But it also, the whole process plays out in a way where it seems like they strike a deal when it's politically worse to not have a deal. I don't actually believe that if there was a problem, there isn't still a problem, or it just seems like it's all been very politicized. You have Susie Wiles, the chief of staff to Trump, being like, we're so glad Fable 5 is back. What is Sarah's thing to see? The chief of staff of the president weighing in on a model release? It's weird. That's all I'm going to say.

12:07

SPEAKER_05

[SPEAKER_04] The question is whether it's all just K-Fabe, you know, it's a fight to create drama to entertain the people, and that this was the plan all along, you know? It's very good for America to be like, this model is so dangerous. We must check and make sure best model is in America. [SPEAKER_04] Yeah, I get it. We need to have the most powerful models. Everyone fear, right? This is a very good political message, and it enables us to continue raising capital for even more dangerous models, right? And etc. So that's the other way to look at it. [SPEAKER_01] I mean, the thing, I think it's clearly trust destroying. That's obvious.

12:17

SPEAKER_05

[SPEAKER_01] Trust with what though, Sam? Just trust in institutions. It's just everyone's like, what the fuck? I don't even want to pay attention anymore. Next time someone says a model is so great, I'm like, whatever. I don't care. Totally.

12:18

SPEAKER_05

[SPEAKER_01] So it's clearly institutional trust destroying. There's good marketing value to it, because it's spicy, right? In various ways, people can spin that. The thing that I don't know, and I'm curious what your guys' take on, is how intuitive versus intentional it is, right? Yeah. And this is the thing I don't know about a lot of things these days. There's the other stuff that's happening in the marketing and the narratives construction, whatever. And part of me is like, this is the most brilliant strategy I've ever seen. And part of me is like, this is no strategy. This is just how it works, you know? I just don't know with a lot of these things, which of the two categories it really is, you know?

12:19

SPEAKER_05

Is it brilliant or is it just intuitive? I mean, that's the question we may never know. I have a related question. It seems like the knives are out for Anthropic and OpenAI as it relates to other tech companies and customers. So my latest example is Alex Karp, the CEO of Palantir, kind of went apeshit against them on CNBC this morning. Really? And he was a little unhinged. I haven't actually watched the full interview.

12:22

SPEAKER_04

[SPEAKER_05] But that's normal for him, right? No, but more so, so much so that it's worth watching. I didn't watch the thing yet. But his upshot is everyone hates these guys.

12:23

SPEAKER_05

But this is why we need Vox Americanas because we have no idea. We don't actually know if these companies are hated or it's all bots. No, but Alex Karp is saying that. So again, this isn't a case of Alex Karp, the CEO of Palantir, partner to both of these companies in a deep way, although I think Anthropic more, is saying on CNBC that everybody hates them.

12:26

SPEAKER_05

And he was a little bit unhinged. I haven't actually watched the full interview. But that's normal for him, right? No, but more so, so much so that it's worth watching. I didn't watch the thing yet. But his upshot is everyone hates these guys. But this is why we need Vox Americanas because we have no idea. We don't actually know if these companies are hated or it's all bots. No, but Alex Karp is saying that. So again, this isn't a case of Alex Karp, the CEO of Palantir, partner to both of these companies in a deep way, although I think Anthropic more, is saying on CNBC that everybody hates them. But I understand. But there's a key to this. He's not saying he hates them. He's saying everybody hates them. And you're thinking, well, what everybody? Is it the Twitter everybody? Because that's true. But what is that? Who are those people? Yeah, I don't actually know all the answers to these questions, but it's also Twitter. I mean, if it's Twitter, it's mostly men and bots, right? I don't think it's Twitter. I think it's also their partners and other tech companies and developers who are using cheaper models. Dave, you're a fan. You were... Was it a fireworks model you were talking about? No, no, no. GLM 5.2.

12:33

SPEAKER_03

[SPEAKER_04] Okay. Tell the people more about it. I know we've talked about it, but... [SPEAKER_04] Yeah. I mean, we talked about it a bit last week, but GLM 5.2 is really good. I think we saw this in our back channel. Who makes it?

12:37

SPEAKER_05

[SPEAKER_04] Yeah. Explain what GLM is.

12:39

SPEAKER_05

[SPEAKER_04] GLM 5.2 is a model made by a Chinese company called Z.AI. And they're one of the foundation model companies in China. And this model is open source. It's a trillion parameter model. And everyone that I've been interacting with and many people across the OpenClock community are switching it out for Opus 4.8 and they don't notice a difference. And so we're actually, I think we, as of last week or the week before, this is, I'm speaking to you from the first week of July. We're at a point in this evolutionary history of AI models where the open source models and the frontier models are neck and neck in terms of your ability to use them for the same purposes, same tasks and not tell the difference, right? We had somebody in one of our back channel conversations over the weekend, who's a friend of ours that runs a big incubator and venture firm in San Francisco who said they swapped out GLM 5.2 on all of their agents and nobody on their team noticed. And I'm hearing this across the board. So I think it's an important moment to pause and say, what does that mean? So why don't these models have more market share?

12:42

SPEAKER_05

[SPEAKER_04] I mean, it's three weeks old.

12:52

SPEAKER_05

[SPEAKER_04] Yeah, fair. Well, I just think this is the moment that we've been talking about forever and why we're skeptical on companies trying to charge premium amounts of money to multiplication. And I think there'll be a fight back. I think a lot of it will be political. People will say, I don't want my numbers multiplied in China. Or with Chinese, this is a negative, a never-ending story. But the question of where the fundamental margin is in these things is the however trillion dollar company, right? Is there margin in them? Because if the most basic, if it's some people got really smart at multiplying numbers, but everyone's going to figure it out. And it's just a metal and power thing. That's a very hard global race. Nothing new here just happening.

13:01

SPEAKER_05

[SPEAKER_01] No, yeah, I mean, it's nothing new. We've been talking about it on the pod, you can go back, literally hundreds of episodes talking about this. But I think the moment is here that you can now make the choice. If you're a business owner and you're trying to make a decision on your capex, in terms of should we put capex into buying bare metal servers, a stack of B 200s costs around half a million dollars, and then run GLM 5.2 for an infinite number of tokens at the cost of power, that trade-off is now real between am I going to put some sophisticated router in place that routes all of my employees token usage to the correct model to optimize the token usage on the frontier. I think that choice is happening. I also talked to another one of our friends that runs a hedge fund in New York. I think if you look at Nvidia, rent spot rental on Nvidia servers is down 22% week over week the last couple of weeks. But all of the capital is shifting to buying actual raw metal, right. And so I think you're just seeing this thing happen, which is the CFOs and all these businesses are saying, I want to understand the exact cost of this part of our business. And I don't want it flying out of control, like it did in Q1 of 2026. And so go spend the money on the servers and get me GLM 5.2. And we're switching over to that. I did two things on this day. I mean, I directionally agree. One is I think that the tweet is the CFOs are back in control, right. And this has shifted from the early days. In the advertising world, there was always the experimental budgets where there were no rules. And so yes, you start whatever you want. And really small companies can subsist on experimental budgets. The second it becomes core spend is a very different pocket and process. And we're moving from this was all experimental. It went way out of hand into core spend. The second is, I actually happened to talk to two very large companies today, public company CTOs. And this idea that obviously, we're going to use all the models and switch between them. It's not even avant-garde. You know what I mean? There isn't, ooh, well, maybe I'll be a Claude shopper. No one thinks that's a ridiculous thing to assert right now, right? I was thinking, yeah, of course, we're using everything. And of course, we're figuring out how to minimize our costs and switch. That would be stupid to not do, right, which then just means this is a race for efficiency. And I think Jessica's team reported something, you know, OpenAI can do inference at half the cost. That's the race. And if they can do that and no one else can, God bless, they're going to be a huge company. But if they can do that for a month,

13:05

SPEAKER_05

[SPEAKER_01] Claude shopper. No one likes that. That's a ridiculous thing to assert right now, right? I was thinking, yeah, of course, we're using everything. And of course, we're figuring out how to minimize our costs and switch it. That would be stupid to not do, right? Which then just means this is a race for efficiency. And I think Jessica's team reported something opening, I can do inference at half the cost. That's the race. And by the way, if they can do that and no one else can, God bless, they're gonna be a huge company. But if they can do that for a month and everyone catches them, then it's a worthless thing. You know what I mean? And that's going to be the real question.

13:12

SPEAKER_05

So why is Anthropic's revenue still ripping? Because there's so many people in the world. And Anthropic started getting more hype in Q1. I think things got viral. They've done a good job of marketing. A lot of these businesses that were in the spot we were at in Q4, Q1 are just getting that marketing hit to them now. But to Sam's point, I think the smarter these businesses get, and after one or two quarters of experimentation, people are gonna be like, wait, wait, wait, we need a router. We need to figure all this out. Maybe it's open source, maybe it's fine metal. And I think you'll start to see the revenue shifting quite a bit over the next six to 12 months.

13:25

SPEAKER_05

For what it's worth, I think the open source router world would be quite interesting because people aren't stupid, right? Being like 101 is clearly we need a router. Someone's gonna raise a closed source router and you're just playing this game of Russian dolls. Well, that's dumb. I'm not gonna use a closed source router. Everyone's gonna want control. And so if there was ever a place to build a really powerful company, it would be open source routing, I think.

13:31

SPEAKER_05

Okay, there we go. If you have an open source router company that's good and call me, not closed source. We are banned from our final topic of the Taylor Swift wedding because some people on the call know too much. That's all I'm gonna say. Well, I think we should at least acknowledge it's happening. There's a lot of speculation. By the time the pod rolls out, it will be the day, the Taylor Swift wedding day. And some people in this pod are gonna then lick the cookie. Some of us will lick the cookie if we were right or not.

13:36

SPEAKER_05

I'm just gonna say, guys, I don't feel like this needs royal wedding treatment. I really don't. As a fan, and I understand the whole thing, I think this is just too much. I think it's too much. Why? Why do we need an American royal wedding?

13:53

SPEAKER_01

She's probably bigger than Princess Diana was back then, isn't she? She's definitely bigger. Sure. Yeah. And William and Kate. For Americans. But why? She doesn't need to do this. She's choosing to do this. [SPEAKER_05] I don't think she's choosing to do this. I think the media has taken on a life of its own around this. I don't think she's trying to put any of this out. But I think that's the acknowledgement. I think that's why it is what it is, is that she has no choice.

14:29

SPEAKER_01

Guys, she gets to control every— But again, and no one whatsoever has this argument, so we will end it. But the whole thing and the mastery of her is that she stays in control. She's always released things on her terms and dropped the hints. And one of the things you love about her is she is this mastermind of storytelling and the narratives around her whole thing. So I just don't buy that all that goes out of the window. Woes me, she has no choice to give the people what they want and get married at Madison Square Garden. She's always in control. I can't think of a public figure that is more in control. And so she's choosing this. And I think it's weird.

14:37

SPEAKER_01

I can't wait to debrief this next week, maybe on our own separate special episode of pop culture. [SPEAKER_05] Pop culture, more or less. We have a whole other thread on this. Thank God. Oh, guys, our kids did Alpha School last week. The AI school, you know?

15:03

SPEAKER_05

Really? Tell us more, Brett. Yeah. Well, okay, I will say I've been very excited about Alpha School. The premise for those who don't know it. Wait, this is like a summer school? You've moved the kids?

15:20

SPEAKER_05

Well, they have Alpha Camp in summer, but it's effectively just a trial week at their school program. And the program is such the first two hours of the day, each kid gets a laptop. The laptop evaluates exactly how smart they are at each subject and then figures out exactly how to get them to an 80 percent success rate every single day. And they play games. So instead of math, they're doing a math game, but they're getting 20 percent wrong and they're slowly growing their skills and teaching them in all these different courses and parts of academics. Then the rest of the day, six hours is workshops and fun stuff and entrepreneurship, financial literacy. Our 10 year old learned how to suture a wound with stitches. They built apps. They did smart greenhouses. They played with micro bits. They built catapults.

15:25

SPEAKER_05

Okay, so you're pro or con? Actually, slightly con. Okay, give it to us.

15:48

SPEAKER_05

So I think this is the future framework for education. I think there's still rough around the edges. I've heard every Alpha School is a little bit different. However, what I really did like is that on the parent side, you get a dashboard every day of their performance and stuff, which could be a blessing or a curse. Tiger parents were definitely gonna love this. And of course, I fed mine right into my agent and had framework for how each of my kids is growing with their education. So I think it could be bad too, because sometimes you just gotta let kids be kids and learn things a little bit more on the fly. I will say for a more ADD kid, he actually preferred the AI screen-based learning because it wasn't a voice where a teacher was teaching to the whole class where he felt a little bit more anxious and less focused. So anyway, it was a good experiment. I think there's something there. I think it's a little early.

15:50

SPEAKER_05

were definitely going to love this. And of course, I fed mine right into my agent and had it about because I feed every report card and I have a whole framework for how each of my kids is growing with their education. So I think it could be bad, too, because sometimes you just got to let kids be kids and learn things a little bit more on the fly. I will say for a more ADD kid, he actually preferred the AI screen-based learning because it wasn't a voice where that was teaching to the whole class where he felt a little bit more anxious and less focused. So anyway, it was a good experiment. I think there's something there. I think it's a little early. And if I were to see an investor, I would invest. But I think I'm going to participate more at the series day to give it a good analogy. Well said. That's a good one. On that note of actually something useful, we will end this week's More or Less pod wishing everyone a happy 250th. I mean, that is the thing. America's 250 years old. That's awesome. Sam likes to say this may be our last consequential birthday. Would you like to elaborate on that, Sam? Well, the world's moving so quickly. You don't think we're going to make it to multi thousands of years? I don't know. It wasn't America 300. We'll be in our 90s. Yeah, but we're going to live to be infinity. Oh, you mean, even our, as in us, us humans. Clearly, you don't subscribe to the longevity thesis, Sam. Well, no, even if I do, it's just math. It was a little bit of physics. We'll be in our 90s. That doesn't mean what we look like in our 90s is a separate question, but we're going to be in our 90s. And I don't know. Could you imagine, I don't know, the United Federation of Western States or some shit by then? I hope it's not where we go. We're going to get another Lincoln out of this. What's going to happen when someone tries to liberate the robots? I use 50 years a long time. By the way, have you seen the interesting, have you seen the clanker videos on Instagram? No, no. Speaking of the Civil War, they're hilarious. Is it like iRobot? No, it's basically like, imagine a guy dressed as Colonel Sanders with all of these clankers, which are his robots doing his things and his son coming in and being like, you can't treat the robots this way. They're really funny. I highly recommend looking

16:02

SPEAKER_05

[SPEAKER_03] up clanker on Instagram. It's the, it's good. We'll do that. We'll do that for our Fourth of July weekend. Okay. Well, very happy Fourth. Have fun. Stay safe. And we'll see you back here next week for another episode of More or Less. Goodbye. Bye guys. Bye everyone. If you enjoyed this show, please leave us a virtual high five by rating it and reviewing it on Apple Podcast, Spotify, YouTube, or wherever you get your podcast. Find more information about each episode in the show notes and follow us on social media by searching for @More or Less at Dave Morin at Lesson at Jay Lesson. And as for me, I'm at Brit. See you guys next time.

16:11

SPEAKER_05

First of all, in 2026, you can't take someone who's like the nice guy liked in the community and build a media strategy for them. You need them to already have like a media footprint themselves.

16:22

SPEAKER_01

That's why you're perfect. Fine. But then here's the second part of my talk track is one, you can't just like find good people and like give them that you're not good enough at this.

16:30

SPEAKER_04

And like there are lots of people like find celebrities and well-known people that already have that. And then here's the biggest thing I think I keep saying to people is like the thing you have to solve if you want people to run for office is not the campaigning. It's make the job not suck. Like if you had a nonprofit infrastructure, which it looks, Sam, here's the deal. You should run for some whatever.

16:52

SPEAKER_01

And you have the media voice. We'll have you develop like on the margin, a few connections, whatever. But the real reason that you should do this is you know the job sucks, but we will provide like all the infrastructure, the bill writing, the relate.

17:05

SPEAKER_05

Like we'll do we'll basically create a platform where you can just be the candidate to make the final decision and we'll be like the huge infrastructure behind it. That works so much better. I thought that's what every institute in D.C. is. No, I think that they're like still oriented on the site. Like I don't like the job sucks too much. Like I think if you started from the perspective, like I want the job to suck zero. How do I make politics an appealing job? And then you backfill and get the candidates versus being like, well, we'll help you campaign to get the job.

17:33

SPEAKER_05

And then you have to do the job. I think the job only sucks for people who care what people say about them. If you don't care like about all of the drama that surrounds being in the public eye, then the job might not suck. But I'll give you a really basic, silly example in California. I know some people who are running for office in California. And like what a shitty life trail. You have to be in Sacramento all the time. It's the worst. I'm not talking about California, man. I'm talking about America. Yeah, but let's just expand this. D.C. is the same problem. You're like, look, here's the deal.

18:03

SPEAKER_05

Yeah, you have to be in Sacramento all the time. But like we have a PC-12 network and we will just like make this an afternoon for you rather than trying to get there on your own. Right. Like that's the type of infrastructure. Like I want briefings done for me. I want my office already set. I like want everything turnkey so I can drop in for a for key moments effectively. And I don't want to act. I think that's the thing is like the job is you pay shittily. It's a shitty life. It's away from your family. It's like all these things you'd have to solve for.

18:34

SPEAKER_05

You just look at the reason that we can fix politics is like we'll solve all the bad parts for you of the actual job. Then I think you get better candidates. That's a fair point. You do have to like being on the road. That's probably it's sort of like being a musician. Like the life is extremely difficult. You're just on the road all the time. It's awful. And it's like it's like even like you know another friend who ran for a big office recently and like they just didn't see their kids forever. And you're like why would you do that for yourself? For what? For like a low paying job? What about what David Sachs did or what Sriram did? I mean.

19:07

SPEAKER_05

I mean those aren't real jobs. Like that is kind of a version of it. But you can have impact. Right? Like you could have impact. I mean Sam let's say a libertarian Democrat won next election which is highly unlikely. But you know you could go do that kind of a role and have a serious impact on this on this exact idea. I get it. I just think there's like a deeper thing to solve here. Right? Which is like there's a the reason they're like why don't good people run for office? Whether or not I'm a good person is a separate question. But like and I think everyone focuses on that well campaigning sucks and like they don't have the media strategy and voice.

19:45

SPEAKER_01

I'm like I actually think you guys just acknowledge the fact the actual job sucks. Let's solve for that.

19:51

SPEAKER_05

And then you can backfill and do the other things and find the right candidates. Like you know. Like the thing I would want to be marketed is this job actually sounds awful. But if you will provide all the infrastructure so that it's like a great job. If that makes sense. Then you figure out the rest. You know. I think you should have a writer like celebrities do when they do concerts. It's like I'll do your job. But I here's my writer. And I do think there's a difference between the electives. Exactly what Americans want from their tech politicians.

20:19

SPEAKER_05

Hey if they're gonna get like a 10x increase in intelligence I think America would do it. And by the way I also think you should be a czar. I think David Sachs was on to something with the czar title. No but the thing about that is like I do think to Jess's point there are people who have like figured this out how to be in the government without being in the government.

20:35

SPEAKER_01

And like that is what they've done. The problem is like those are not elected jobs. They're nowhere near like you get to like say some stuff and be in there with some meetings. But like it's a vice chairman role at best. And like those are actually not what you need. You need better fundamental politicians is my view. Right.

20:50

SPEAKER_05

Britt do you have any political aspirations? Britt you could do it. People have also told me that I should run for office. I've always thought Dave would make for a great first gentleman. But. Dave would make a great first gentleman. I'm in. It's not in my next decade. I might reconsider in my 50s. We'll see. Man candy number one. Man candy number one. I'm definitely never doing it but I'm happy to be the first gentleman. I'm gonna put Dave and his agents on my opposition research team.

21:15

SPEAKER_01

Oh crazy. And then he's gonna figure out all the stuff to talk about. Oh god. About the other people. This is why we're never doing this.

21:21

SPEAKER_05

You just want Dave's agents. I mean look I think the reason this conversation was interesting is it's about distribution Sam. Like your idea is amazing. I actually want to participate in Vox Americana at scale without question. I think it would be deeply healthy for the American populace and for society. But the question of distribution. That's why I sort of threw that idea out. Like would Trump Social or Truth Social have ever gotten to scale. Were it. If it weren't for the cultural movement of mega and etc. Well but I'll actually go a step further today. Which is Truth Social got. Because they had.

21:57

SPEAKER_05

There's a type of content I want. I can't get on the mainstream platforms. Right. And the question is very simply for any platform. You have to provide something where like I'm coming here. Because there's something I want that I fundamentally can't get anywhere else. And I don't like. There's value to like the ID registration and like there's a lot of things. But like I don't. I have to play with how to. If there's anything new in steps there. Like it's easy for like. Like OnlyFans. Easy. Right. You know what I mean. Like I know exactly. I know exactly the content I'm getting. Nowhere else. You know. Maga world. Easy.

22:26

SPEAKER_05

Like I know what content I'm getting. I think this carries that. I think this carries the idea that like. I'm definitely talking to Americans. Like I think that's really interesting. No. I understand. I think that's true. I just. There's a subtlety to like. What content does that generate? The medium is the message.

22:40

SPEAKER_01

Like what is the thing I'm coming for? Other than yes. I do want comments from real Americans. I'm actually quite interested in what real Americans say. I mean. I think it fits into a. Important political narrative right now. Right. Which is like who is American on the internet. Right. Like it fits into like the fact that like most of us interacting with. Algorithms on TikTok that are not controlled by Americans. People that aren't Americans. Like. You know. This dovetails into the immigration conversation. I mean.

23:06

SPEAKER_05

I obviously deeply agree. I think the problem is though. And I think this is like the benefit of the problem is like. I think every thinking person recognizes. That if you truly had the American internet. Or like the French internet. This can be global in various ways. It is so much more moderate. Than what you get on Twitter. Right. Like it would like. It would be so much more moderate. Because it turns out that. I don't know about that. I mean. The human internet versus the bot in human internet is one thing. But I found it interesting that. This goes back to TikTok. Both David and Sam. You guys feel this sort of nationalist. I want to talk to Americans.

23:40

SPEAKER_05

Like I understand wanting to talk to real smart people. But. Well. Jess. It's twofold. It's not just talking to Americans. It's not being influenced by. The. You know. Nation state activities. Of external nations. Right. Inside of these networks. They're different things. You post. I post something. Or like more extreme. Like Joe Lonsdale. Or like. I just talked to Joe today. I love Joe. Sean McGuire. Post something. The amount of vitriol. And hate. That you get back on it. Is wild. Right. And like. You kind of look.

24:10

SPEAKER_05

at it. You're like. Look. This is like. 95% bots. But. Some of that vitriol. And hate. Is. From Americans. But. I'm not sure. How much. And I'm actually. Not sure. I think. It's actually. Way. Smaller. Than you realize. If that makes sense. So. You're using. It. As a proxy. For. Like. A. Filter. Well. No one. Knows. To double click. On this. I had. Not. Confronted. The bot army. Recently. Until. As we talked. About. I think. On the pod. A few weeks ago.

24:35

SPEAKER_05

I spoke. At this. Dell World. Conference. In Las Vegas. And they posted. A video. On their. Their account. And tagged me. And. The. Unbelievable. Backlash. Against AI. That like. Filled. My Twitter. Feed. For. Days. On end. And every time. I would click. On one of them. It was like. A short. Name. They were verified. Accounts. But they all. Had these. Like. Long. Set of numbers.

25:00

SPEAKER_04

On the end of it. And. It was very. Clearly. Bot mediated. I don't know. If it was. Computers. Or whether. It's people. In other countries. But there's. Very clearly. An agenda. Happening here. Of course. And so. The problem. Is the thing. I actually. I'm actually. Quite. Data center is a great example. I'm actually. Really curious. How actual. Americans. Think about this. You know what I mean. Like. And I'd actually. Be willing to engage. With actual. Americans on it. But you know. The dumbest thing. In the world. Is fighting. Bot. State actors. On the internet. And having flame. What are they doing?

25:30

SPEAKER_04

They look incredibly insane. Like. When you see people. Like. Responding to bots. With like. Cogent arguments. You're like. That bot. Is literally. Just like. Wasting your time. Like. And so. Like. For me. Like. That's the thing. I was like. I really. Think there is a place. For this type. Of conversation. But the thing. I'm quite. Unsure about. Is like. How do you spark. More than intellectual. Interest in it. You know. It's a good point. I think. You have to broker. A deal. With one of the social networks. That claims. You'll get more distribution. If you sign up. For this thing. And verify. That is.

26:00

SPEAKER_05

100% what it should be. Like there's no question that that's the way the world should work. And I only sort of understand why it doesn't work that way because it's not that hard to do anymore.

26:08

SPEAKER_01

Or maybe partner with people who are running for president. Like whoever's running for president, like be the partner. So the other thing I've been kind of playing with around Vox Americanas and what to do with it is to just be like, look, it actually shouldn't have its own. I mean, it can have a feed, whatever, but it really is more OAuthy. And you think of it as like a way to prove Americanness, right? And reality and like off a bunch of accounts and like who you actually are

26:30

SPEAKER_05

in scenarios that are social. There are ways you could think about that. I love that. Oh, that's what I thought it was going. Yeah. It's like you could do it. Like I was thinking, Dave, you might like this. I was thinking about like, okay, you know how like certain services, if you like want to prove this is your Twitter account, you like post like the link and you like crawl it and you basically create an index of actual Americans on Twitter. Like, is that the better strategy? Like there's a thing to play with. This is my 250th or Tinder. I bet all the dating apps would want this, right? Yeah. You just create, you basically create a proxy layer. Wait, can I

27:02

SPEAKER_05

tell you where I want this the most? Yeah. I want this on Amazon. I want to filter to only American sellers on Amazon. Only American sellers. Yeah. Yeah. And look, I think for what it's worth, this scales globally. Like I don't know what the French auth services, but they have the same thing. Like we're in this era of like nationalism. Yeah. For what it's worth, I'll take French sellers, Italian sellers. Yeah. I'll take all Europeans and Japanese. All the highest quality places where craftsmanship matters. And the reality is like, this is completely against the original, like one world open market, blah, blah, blah, thesis,

27:35

SPEAKER_05

that utopian end of history shit. Right. I don't think it is, Sam. I think it's totally fair to want

27:41

SPEAKER_03

verification of identity and to understand, you know, like I remember growing up in Montana, there was a sticker that would, people would put onto the goods that they made in Montana that said, made in Montana and Montana actually invested a lot in, you know, supporting those businesses and

27:59

SPEAKER_05

helping them grow. Like, it's really no different. Like the internet is huge, global, and amazing. I mean,

28:04

SPEAKER_04

the speed at which open claw is growing right now is blowing my mind, right? Like the scale of the

28:10

SPEAKER_03

internet's enormous, but I don't think there's anything wrong with wanting to understand who you're talking to or where a good is coming from. A hundred percent. I just think like, for me, like the, the moment, especially cause vibe coding and a lot of this stuff is just, I mean, I actually,

28:23

SPEAKER_05

if you go to Vox Americanis and use it, it is actually pretty good. Like, it's great. You can do video live recording. It's got a great NUX. It's like a pretty better than truth. Social was at launch. It actually is pretty freaking good. And like the fact that I could like build that in a few days, but like, honestly on boats, like texting my, my, or emailing my bot for changes is like pretty sweet. And it's like, you, I just think we're in this moment of like, just stop arguing about this stuff and build things and like tinker. And like, I think that's the most uplifting part of all this AI

28:56

SPEAKER_05

is not, is the tinkering. For what it's worth to shift to a different topic. I think this is a really important thing that Britt and I were on the phone yesterday with one of our founders. And this is a

29:07

SPEAKER_03

founder who spent years at Apple and Spotify in product roles, interfacing with engineers. He's

29:16

SPEAKER_05

definitely has 10,000 hours working with engineers, right? 20 years of experience, very talented founder.

29:24

SPEAKER_03

And it's interesting how much people who were not engineers believe they don't have the skillset to work in the world of agentic engineering now. And I found myself telling him you are more capable than you think today to do this. And that I think that people really don't realize how powerful the

29:45

SPEAKER_05

tools are and how much more powerful they've even become even in the last three months. And that they should just try it. Yeah. It's wild. So it's funny. I, I distinctly remember last summer being at the same

29:56

SPEAKER_03

place we're at now and like being in cursor and just like being like, oh my God, this stuff finally works. Right. And like, I compare that to now and it's actually insane, right? Like the level of abstraction you can work on, it's mind blowingly different. Like, you know, before I was like, you know, like an MVC frameworks and editing file by file. And now I'm like, I don't need to do any of that. I literally sit here and email my infrastructure and spin up everything. Right. And it's incredible.

30:23

SPEAKER_01

And so when you do that, you're just like, this is an age that is like perfect for tinkerers. And this

30:28

SPEAKER_03

has been like my most uplifting experience was running our bot school, right? Where we basically

30:33

SPEAKER_01

took all of these social media influencers, people with big cults and followings. We sat them down,

30:38

SPEAKER_03

we gave them the tool chain, you know, how to use digital ocean that are really, and like, it was amazing. What they produced in 24 hours, you know, I remember hackathons where you'd come out

30:49

SPEAKER_01

and something really half working in like, you know, four days. Now you can come out to 24 hours, like products you can ship to your actual users, and they're good. It's like wildly different. And

30:58

SPEAKER_04

I think that is the most uplifting part. Like the mega wars, like who's got the cheapest way to

31:04

SPEAKER_01

multiply numbers is all kind of boring. But like, the empowerment of people with like minimal shins to build entire world is like wild. The thing I was going to also say is in that conversation yesterday with this founder days talking about what it came down to at the end to you know, he's trying to recruit his like founding engineer, everyone's a great a lot of people are great engineers, you know, of course, some are better than others, fine. But cultural fit, I think matters way more now than ever before, when like, the, the ability to be a great engineer is like much easier than ever

31:37

SPEAKER_01

before. And so it's like teaching these people how to be human, maybe a little throwback to etiquette school, but also like, having street smarts being a cool person, high integrity, like vibing with the team, I think is is going to matter more than ever over the next few years for teams.

31:55

SPEAKER_05

Yeah, I completely agree. It's a wild moment. It's a great moment. It's really this stuff is the fun stuff. I think it's important to note that it's just not too late. Like that's what talking to this founder, I was shocked by how he's like, isn't it too late? I'm like, No, like, the skill set of this new style of engineering is a brand new skill set that, you know, it's like, as if it used to be you had to speak a language that a compiler could understand, you no longer have to speak a language that a compiler has to understand. Now you just have to be able to speak a language that the models can

32:25

SPEAKER_05

understand the models go speak compiler. And the compiler speaks to the assembler, the assembler speaks machine language, and like everything's, it's like amazing. And it's a totally new skill set, and you're not too late. And though we talk about this a lot on the pod, we also, you know, I hear from people every week that listen to the pod that are entrepreneurs thinking about starting something. And I just think it's really important for people to hear that you're, you're totally capable, and it's definitely not too late. And take an hour a day and just really try to learn this stuff, because everyone's learning it right now.

32:58

SPEAKER_05

Although I don't think so I agree with everything you said, except for one thing, I don't think you take an hour a day, I think you binge it for like, as much time as you can. Well, Sam, I'm saying if you've got another job, if you're doing something else, right, like try to find, it's like learning guitar or something, you got to find time to practice every day in order to make progress. Sure, if you can take your if you can really binge into it, go for it. But you know, Except for please stay human, because I really think we're entering the way I know we're already in the loneliness epidemic because of social media. But this is entering this new, scary, dark place,

33:34

SPEAKER_05

I think where people are alone with their agents up at all hours, not having community and real social conversation. And I'm like, really actually worried about it. I am going to play the normie on this. I like my agent. It's useful for work. I don't believe we are all headed for a future where we all need to be vibe coders. I'm somewhere in the middle.

33:55

SPEAKER_01

We know it's your opinion. We had that argument two weeks ago. Sam wasn't there, though. Sam didn't get to weigh in on that. I would I just want the listeners to know that I think that, you know, we're in this exciting early adopter moment. There's a lot. That doesn't mean I think AI won't touch everyone. It doesn't mean

34:14

SPEAKER_05

that I don't think AI is like, transformative. It's begun to become invisible is the thing that I said before, which I still stand by, which is like, you won't have to be on your computer vibe coding all night. But like, AI in agents will be doing stuff for you no matter what. Cool. But that's fine. But does it mean that I have to devote two hours of my day to vibe coding or I'm going to be left behind? I think it's somewhere in the middle. It's pretty powerful stuff. We'll see. It's going to be an interesting era. I think for now it's super fun if you have business problems you're trying to solve.

34:44

SPEAKER_05

I mean, Dave's line on you're not too late, I think is a really, really important one. I actually hilariously started because it's the summer. I'm going to write like a short Sam book that I've kind of outlined on like what I've learned from the investments I've made. And one of the chapters is called You're Not Too Late. Because I think it's a mistake even on investing I've continuously made.

35:04

SPEAKER_01

And like I've gotten a few things right where like there's this class of investments that where you're like, well, I've seen this for two years. I can't believe I missed it. I loved it, but I'm too late. And like it just turns out you're not. Right. And like I think that's like one of my that's one of my chapters in the book I want to write this summer. Well, that's because we're seed investors. And if we like broke ourselves of that mentality, I think we would learn that faster. Speaking of bots, I did log into mine today to see Fable 5 is back. So for those of good marketing from Anthropic, they have reached a deal with the

35:38

SPEAKER_01

United States government. What incredible marketing. Well done. It's great marketing. Until July 7th, you can use up to 50% of your plan's weekly usage limit on Fable 5. Which is interesting, actually. They're basically like, you know, after that, you got to pay by token,

35:55

SPEAKER_05

which is interesting. What if I don't, I, the kind of stuff I'm doing, I definitely don't need Fable 5 for. You never know. That's actually the whole business model. The beauty of the whole thing is you're almost certainly right, Jessica. But what if? The entire margin of the industry is based on people asking the question, like, look, I'm asking you to fold my, and then fold laundry is a funny one. I'm asking you to like make a dinner reservation. I mean, an IQ 100 person can do it. But do I need to hire an IQ 170 person to make a dinner reservation on the off chance it matters? Like that is the margin of the system.

36:29

SPEAKER_05

And like, we'll see how that plays out. I definitely don't need Fable 5. But it's back. This back and forth with the government is wild, I think. And you know, for those who forgot, this is Fable 5 was released along with Mythos. In this case, Amazon raised some doubts about some ability for it to be jailbroken. Anthropic essentially said it's no more vulnerable than other things. Amazon isn't as smart as us that it got in this whole back and forth. But it's out now. There's some security thing Anthropic added. But like, we're just going to see this for every single model release. Yes.

37:06

SPEAKER_05

It's weird. Isn't it weird? Do we? I don't know. How do we feel about this? I don't feel great about it. About what? That we need the government to sign off on models? About what? I feel great about what specific, what specifically? Yeah, it's a good question. I mean, on one hand, I like that there's someone checking out for the safety of these models. But it also, the whole process plays out in a way where it seems like

37:30

SPEAKER_04

they strike a deal when it's like politically worse to not have a deal. I don't actually believe that, like, if there was a problem, there isn't still a problem or, you know, like, it just seems like it's all been very politicized. You have Susie Wiles, the chief of staff to Trump, being like, we're so glad Fable 5 is back. Like, what is Sarah's thing to see? Like, the chief of staff of the president weighing in on a model release? It's weird. That's all I'm going to say. The question is whether it's all just K-Fabe, you know, like, it's like a fight to create drama, to entertain the people, and that this was the plan all along, you know? It's like very good for America

38:15

SPEAKER_04

to be like, this model is so dangerous. We must check and make sure best model is in America. Yeah, I get it. We need to have the most powerful models. Everyone fear,

38:27

SPEAKER_01

right? Like, this is a very good political message, and it enables us to continue raising capital for even more dangerous model, right? And etc. So that's like the other way to look at it. I mean, the thing, like, I think it's clearly trust destroying. That's obvious. Trust with what though, Sam? Just like trust in institutions. Like, it's just like everyone's like, what the fuck? I don't even want to pay attention anymore. Like, next time someone says a model is so great, it's whatever. I'm like, whatever. I don't care. Totally. So like, it's clearly institutional trust destroying. There's good marketing value to it,

38:56

SPEAKER_01

because it's like spicy, right? In various ways, people can spin that. The thing that I don't know, and I'm curious what your guys' take on, is like, how intuitive versus intentional it is, right? Like, yeah. And this is the thing I don't know about a lot of things these days. Like, you know, there's the other stuff that's happening in the marketing and the narratives, construction, whatever. And like, part of me is like, this is the most brilliant strategy I've ever seen. And part of me is like, this is no strategy. This is just like how it, you know what I mean? Like, I just don't

39:25

SPEAKER_05

know with a lot of these things, like which of the two categories it really is, you know? Is it brilliant or is it just intuitive? I mean, that's the question we may never know. I have a related question. It seems like the knives are out for Anthropic and OpenAI as it relates to like, other tech companies and customers. So my latest example is Alex Karp, the CEO of Palantir, kind of went apeshit against them on CNBC this morning. Really? And he was a little bit unhinged. I haven't actually watched the full interview. But that's normal for him, right? No, but more so, so much so that it's worth watching. I didn't watch the

40:06

SPEAKER_05

thing yet. But his upshot is like, everyone hates these guys. But this is why we need Vox Americanas because we have no idea. Like, we don't actually know if these companies are hated or it's all bots. No, but Alex Karp is saying that. So again, this isn't a case of like Alex Karp, the CEO of Palantir, partner to both of these companies in a deep way, although I think Anthropic more, is saying on CNBC that everybody hates them. But I understand. But there's a key to this. He's not saying he hates them. He's saying everybody hates them. And you're like, well, what everybody? Is it the Twitter

40:38

SPEAKER_05

everybody? Because like, that's true. But like, what is that? Who are those people? Yeah, I don't actually know all the answers to these questions, but it's also... Twitter. I mean, if it's Twitter, it's mostly men and bots, right? I don't think it's Twitter. I think it's also their partners and other tech companies and developers who are using cheaper models. Dave, you're a fan. I mean, you were... Was it a fireworks model you were talking about? No, no, no. GLM 5.2.

41:05

SPEAKER_04

Okay. Tell the people more about it. I know we've talked about it, but... Yeah. I mean, we talked about it a bit last week, but, you know, GLM 5.2 is really good. You know, I think we saw this in our back channel. Who makes it? Yeah. Explain what GLM is. GLM 5.2 is a model made by a Chinese company called Z.AI. And they're one of the

41:25

SPEAKER_05

foundation model companies in China. And this model is open source. It's a trillion parameter model. And everyone that I've been interacting with and many people across the OpenClock community are switching it out for Opus 4.8 and they don't notice a difference. And so we're actually, I think we, as of last week or say the week before, this is, you know, I'm speaking to you from the first week of July. We're at a point in this evolutionary history of AI models where the open source models and the frontier models are neck and neck in terms of your ability to use them for the same purposes,

42:08

SPEAKER_05

same tasks and not tell the difference, right? We had somebody in one of our back channel conversations over the weekend, who's a friend of ours that runs a big incubator and venture firm in San Francisco who said they, you know, they swapped out GLM 5.2 on all of their agents and nobody on their

42:25

SPEAKER_04

team noticed. And I'm hearing this across the board. So I think it's an important moment, you know, to sort of pause and say, what does that mean? So why don't these models have more market share? I mean, it's three weeks old. Yeah. Fair. Well, I just think this is the moment that we've been talking about forever and like why we're kind of skeptical on companies trying to charge premium amounts of money to multiplication. And like, look, I think there'll be a fight back. I think a lot of it will be political. People will say, I don't want my numbers multiplied in China. Or with Chinese, like, this is like, this is a negative,

42:57

SPEAKER_04

a never ending story. But the question of like, where the fundamental margin is in these things

43:01

SPEAKER_01

is like the however trillion dollar company, right? Is there margin in them? Because if the most basic, if it's like, yeah, some people got really smart at multiplying numbers, but everyone's going to figure it out. And it's just a metal and, you know, and like power thing. That's a very hard global race, you know, nothing new here just happening. No, yeah, I mean, it's nothing new. We've been, we've been talking about it on the pod, you can go back, you know, literally hundreds of episodes talking about this. But I think it's the moment is here that you can now make the choice. Like,

43:33

SPEAKER_01

if you're a business owner, and you're trying to make a decision on your capex, in terms of should we put capex into buying bare metal servers, you know, a stack of B 200s costs around half a million dollars, and then run GLM 5.2 for infinity number of tokens at the cost of power, you know, that trade off is now real between, you know, am I going to put some sophisticated router in place that routes the all of my employees token usage to the correct model to optimize the, you know, the token usage on the frontier, you know, I think that choice is happening. I also talked to another one of our

44:13

SPEAKER_01

friends that runs a hedge fund in New York, you know, I think if you look at Nvidia, you know, rent, spot rental, and on Nvidia servers is down like 22% week over week the last couple of weeks. But all of the capital is shifting to buying actual raw metal, right. And so I think you're just seeing this thing happen, which is the CFOs and all these businesses are saying, I want to understand the exact cost of this of this part of our business. And I don't want it flying out of control, like it did in Q1 of 2026. And so like, go spend the money on the servers and get me GLM 5.2. And we're switching over to that.

44:53

SPEAKER_01

I did two things on this day. I mean, I directionally agree. One is I think that the tweet is like the CFOs are back in control, right. And like, this has shifted from the early days, like in the advertising world, there was always the experimental budgets where there were no rules. And so it was like, yeah, you start whatever you want. And like, really small companies can subsist on experimental budgets. The second it becomes core spend is a very different pocket and process. And like, we're moving from like, this was all experimental bullshit. It went way out of hand into core spend. The second is,

45:20

SPEAKER_01

I actually happened shockingly to talk to two very large companies today, public company CTOs. And this idea that like, obviously, we're going to use all the models and switch between them. It's not even avant garde. You know what I mean? Like there isn't like, ooh, well, maybe I'll be a Claude shopper. Like no one like that's like a ridiculous thing to assert right now, right? I was like, yeah, of course, we're using everything. And of course, we're figuring out how to minimize our costs and switch it like that would be stupid to not do, right, which then just means like, this is a race for

45:49

SPEAKER_01

efficiency. And you know, I think Jessica's team reported something like, you know, opening, I can do inference at half the cost, like, that's the race. And like, by the way, if they can do that, and no one else can, God bless, they're gonna be a huge company. But if they can do that for a month, and everyone catches them, then it's a worthless thing. You know what I mean? And like, that's going to be the real question. So why is Anthropix revenue still ripping? Because there's so many people in the world. And Anthropix like started getting more and more hype in Q1. I think things got viral. They've done

46:18

SPEAKER_01

a good job of marketing. A lot of these businesses that like were in the spot we were at in like Q4, Q1 are just getting that marketing hit to them now. But to Sam's point, I think the smarter these businesses get, and after one or two quarters of experimentation, people are gonna be like, wait, wait, wait, wait, wait, we need a router. We need, you know, to figure all this out. Maybe it's open source, maybe it's like fine metal. And, and I think you'll start to see the revenue shifting quite a bit over the next six to 12 months. For what it's worth, I think the open source router world would be quite interesting,

46:51

SPEAKER_01

because people aren't stupid, right? And so being like, the stupid, everyone, 101 is like, well, clearly, we need a router. You're like, well, someone's gonna raise a closed source router, and you're just playing this game of Russian dolls. We're like, well, that's dumb. I'm not gonna use a closed source router. Like everyone's gonna want control. And so if there was ever a place to build a really powerful company, it would be open source routing, I think. Okay, there we go. If you have an open source router company, that's good and call me, not closed source. We are banned from our final topic of the Taylor Swift wedding because some people

47:27

SPEAKER_01

are all on the call know too much. That's all I'm gonna say. Well, I think we should at least acknowledge. I think we need to acknowledge it's happening. There's a lot of speculation. By the time the pod rolls out, it will be the day, the Taylor Swift wedding day. And some people in this pod are gonna then lick the cookie. Some of us will lick the cookie if we were right or not. I'm just gonna say, guys, I don't feel like this needs royal wedding treatment. I really don't. As a fan, and I understand the whole thing, I think this is just too much. I think it's too much. Why? Why? Why do we need an American royal wedding? Why?

48:09

SPEAKER_01

She's probably bigger than Princess Diana was back then, isn't she? She's definitely bigger. Sure. Yeah. And William and Kate. For Americans. But why? Like, she doesn't need to do this. She's choosing to do this.

48:23

SPEAKER_05

I don't think she's choosing to do this. I think the media has taken on the life of its own around this. I don't think she's trying to put any of this out. But I think that's the acknowledgement. I think that's why it is what it is probably, is that like she has no choice. Guys, she gets to control every, but I, again, and no one whatsoever has this argument, so we will end it. But the whole thing and the mastery of her is that she stays in control. And she's always released things on her terms and dropped the hints. And one of the things you love about her is she is this mastermind of storytelling and the narratives around her whole thing.

49:01

SPEAKER_05

So I just don't buy that all that goes out of the window. And oh, woes me. She has no choice to give the people what they want and get married at Madison Square Garden. She's always in control. I can't think of a public figure that is more in control. And so she's choosing this. And I think it's weird.

49:18

SPEAKER_01

I can't wait to debrief this next week, maybe on our own separate special episode of the pop culture.

49:24

SPEAKER_05

Pop culture, more or less. We have a whole other thread on this. Thank God. Oh, guys, our kids did alpha school last week. The AI school, you know? Really? Tell us more, Brett. Yeah. Well, OK, I will say I've been very excited about alpha school. The premise for those who don't know it. Wait, this is like a summer school like you've moved the kids? Well, they have alpha camp in summer, but it's effectively just a trial week at their school program. And the program is such the first two hours of the day, each kid gets a laptop. The laptop evaluates exactly how smart they are at each subject and then figures out exactly how to get them to like

50:01

SPEAKER_05

an 80 percent success rate every single day. And they play games. So it's like instead of math, they're doing a math game, but they're getting like 20 percent wrong and they're slowly growing their skills and teaching them and all these different courses and parts of academics. Then the rest of the day, six hours is like workshops and fun stuff and entrepreneurship, financial literacy. Like our 10 year old learned how to suture a wound like with stitches. They built apps. They did smart greenhouses. They played with micro bits. They built like catapults. OK, so you're pro or con? Pro or con? Actually, slightly con. OK, give it to us.

50:42

SPEAKER_05

So I think this is the future framework for education. I think there's still like rough around the edges. I've heard every office school is a little bit different. However, what I really did like is that on the parent side, you get a dashboard every day of like their performance and stuff, which could be a blessing or a curse. It's like tiger parents were definitely going to love this. And like, of course, I fed mine right into my agent and had it about like because I feed every report card and I have like a whole like framework for how each of my kids is like growing with their education. So I think it could be bad, too, because sometimes you

51:16

SPEAKER_05

just got to let kids be kids and like learn things a little bit more on the fly. I will say for a like more ADD kid, he actually preferred the AI like screen based learning because it wasn't a voice where that was teaching to the whole class where he felt a little bit more anxious and less focused. So anyway, it was a good experiment. I think there's something there. I think it's a little early. And if I were to see an investor, I would invest. But I think I'm going to participate more at the series day to give it a good analogy. Well said. That's a good one. On that note of actually something useful, we will

51:53

SPEAKER_05

end this week's more or less pod wishing everyone a happy 250th. I mean, that is the thing. America's 250 years old. That's awesome. Sam likes to say this may be our last consequential birthday. Would you like to elaborate on that, Sam? Well, you know, like the world's moving so quickly. You don't think we're going to make it to multi thousands of years? I don't know. It wasn't America 300. We'll be in our 90s. Yeah, but we're going to live to be infinity. Oh, you mean, even our like as in us, us humans. Clearly, you don't subscribe to the longevity thesis, Sam. Well, no, even if I do, it's just math. It was a little bit of physics. We'll be in our 90s. That

52:33

SPEAKER_05

doesn't mean what we look like in our 90s is a separate question, but like we're going to be in our 90s. And like, I don't know. Could you imagine like, I don't know, the United Federation of Western States or some shit by then? Like, I don't, I mean, I hope it's not where we go. Like, we're going to get another Lincoln out of this. Like, what's going to happen when like someone tries to liberate the robots? I use 50 years a long time. By the way, have you seen the interesting, like, have you seen the the clanker videos on Instagram? No, no. Speaking of the Civil War, they're hilarious. Is it like iRobot? No, it's basically like, imagine a guy dressed as Colonel

53:15

SPEAKER_05

Sanders with all of these clankers, which are his robots doing his things and like his son coming in and being like, you can't treat the robots this way. They're really funny. I highly recommend looking

53:26

SPEAKER_03

up clanker on Instagram. It's like the, it's good. We'll do that. We'll do that for our fourth of July weekend. Okay. Well, very happy fourth. Have fun. Stay safe. And we'll see you back here next week for another episode of more or less. Goodbye. Bye guys. Bye everyone. If you enjoyed this show, please leave us a virtual high five by rating it and reviewing it on Apple podcast, Spotify, YouTube, or wherever you get your podcast. Find more information about each episode in the show notes and follow us on social media by searching for at more or less at Dave Morin at lesson at Jay lesson. And as for me, I'm at Brit. See you guys next time.

Reading tools

Type to find a passage

Appearance
Ask this transcript

Notebook

Export Markdown

Add a note