Open Source AI Wins. Now Labs Are Running to Washington |Thinking Machines, Johnny Ive, Realtime API

Transcript

If the plan was, we America, we need to take our military and bomb every matrix multiplication facility globally, and we're going to have all of them, and then we're going to control AI, that would at least be logical. One thing that is different today, Wednesday, July 15th, is that America now has at least one open weights model that just went out the door. I'm convinced women should be microdosing testosterone. More or less. 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. Deep summer edition. We're deep summer. Is this the deepest of summer? Or is that late July? I don't know. When is the deep summer? It also depends where you are. When do most people vacation? It is right around now. This is pretty much the middle, right? Sure. I don't know. Not if you're European. Yeah, not if you're East Coast. Yeah, East Coast is August. I'm saying, if you go back to school in the middle of August, then July 15th is pretty much the middle. But East Coasters go back in the middle of September, so that's the difference. Oh, they do? But we're West Coasters, so happy deep summer. What is wrong with West Coast? Well, we get out earlier. We're here. There's news. A little bit. Let's not pretend there's a lot of news, guys. There's not a lot of news. TITV turned one. That was big news this week. Oh, congrats. Congrats. Woohoo. Thank you. Thank you. We made it. Keep on going. We've got some AI headlines. We should talk hardware, new models. Potential acquisitions happening. Potential acquisitions. I feel like the acquisition, it's always hard because the bankers just want you to think the acquisitions are already around the corner. But I think we're going to see a lot. People are going to throw in the towel and take the valuation. But you guys are the actual investors. What do you have cooking? Tell us about all your deals. Never happening. I have a number of companies that are in M&A processes, so I do think there are real acquisitions that are going to happen. There you go. Finally. Remember, we've been doing this pod so long, and the first year or two were so dry out there in the M&A world. And every year I kept predicting, this is the year M&A is going to happen. But here we are. Here you are. So what's catalyzing it, Britt? Why now? Without telling us which. Well, I think there are real IPOs happening. There is real liquidity happening. The AI races have never been hotter. What real IPO are you talking about? Okay, their Narrative IPO is happening. Their AI race is happening. No, Anthropic is a real IPO, guys. Even OpenAI is a real-ish IPO. Real-ish. What? Really? The valuations of these IPOs are not real. These companies are actually going to go public this year, Jess? Anthropic will go public. I would be shocked if they're not public in September. Yeah. Really? I'd be very surprised. Yeah. Agree. Yeah, and OpenAI is going to follow. They might wait until next year. Yeah, I don't know. The SpaceX IPO would pause on that for me if I was either of them. I don't know. It was fine. But Anthropic is a real company. Really? None of them are real companies that warrant the valuations they have. What? Guys, Anthropic, if you look at its revenue, profitability, growth rate, it's a real company. Why do you say it isn't, Dave? I may disagree with some of the decisions they make, but why do you think it's not a real company? Well, we've talked about it ad nauseam. These companies are burning enormous amounts of capital. They're not profitable. But Anthropic is profitable. On which metrics? Actually, on the good GAAP ones, I think. Not just the... On the good metrics, not the bad metrics. Yeah. No, I think they are. But to highlight your point, you think the numbers still don't add up ultimately, and maybe that they can't sustain this revenue growth. That kind of thing. Yeah. Yeah. We'll see. Sam, you're very quiet. What do you think? I'm just looking at OpenAI's hardware. Oh, it's Supply... What is Supply Co. with OpenAI? Is this their thing? The Codex thing? It's just... It's merch. It's not hardware. Oh, this is... It's the easy button. Is this the easy button? Yes. Yes. Yes. For 2026? Yes. Got it. So it's like a Staples play. Yeah, it's just merch. I think everyone's out securing their bags before it all falls apart. Yep. That's what's happening. What signs are you seeing that it's all going to fall apart? Well, there's no news. As Dave said, this is the same thing we've been saying forever. It's just everyone now is saying it in the open and out loud, right? Which is that none of this stuff makes sense, which people are saying out loud. And everyone's going to race for the bag. Sell your company wherever you can. Get the cash. Yeah, I'm seeing people sell a bunch of companies. And I don't think it's wrong. I think they should. If you can get a real company, if you can get a real company, if you can actually do an EBITDA multiple on it, I give them credit for that. Then yes, they're a real company. But I'll go back to the Bob Lesson definition of a real company as my dad. A real company is very simple. A real company is a company you can sue, and you get paid if you win the lawsuit. Interesting. So I think they're real companies. Okay. I like it. I like it. Okay. But everyone's getting super nervous. For all the reasons we've been talking about every week. Yeah. This isn't new. It's not like something changed week over week. Well, Dave, we do do this podcast weekly, so dig deep and find something to say. We have to keep plotting our chart, Dave, to make sure we're all moving in the same direction week over week toward the top of the bubble. Here's something that's changed. Britt's tweets about AI have changed. I've noticed a change in the tenor of them. Ooh. What does that mean? Interesting. [SPEAKER_03] So I think they're real companies. Okay. I like it. I like it. Okay. But everyone's getting super nervous. [SPEAKER_03] For all the reasons we've been talking about every week. Yeah. This isn't new. It's not as if something changed week over week. [SPEAKER_01] Well, Dave, we do do this podcast weekly. So dig deep and find something to say. [SPEAKER_01] We have to keep plotting our chart, Dave, to make sure we're all moving in the same direction week over week toward the top of the bubble. [SPEAKER_03] Here's something that's changed. Britt's tweets about AI have changed. I've noticed a change in the tenor of them. [SPEAKER_03] Ooh. What does that mean? [SPEAKER_04] Here's the other thing, which I'll just say. The OpenAI's 5.6 sole is clearly superior to Fable. [SPEAKER_01] There's nobody serious who doesn't agree. And the open source models are less than two months behind. [SPEAKER_03] I've switched out GLM-5.2 for almost every utilitarian use case that matters in any of my agents. Because why would I pay for Frontier Intelligence to do all utility use cases? It just doesn't make sense when GLM-5.2 is actually Frontier Intelligence, and it's insanely cheap. [SPEAKER_04] So of course I'm going to use that. Why would I pay more? Are you saying you're going with discount multiplying big numbers? [SPEAKER_01] Yes, I am. I like discount multiplying big numbers. [SPEAKER_01] Oh, this is what you tweeted, Britt. You tweeted that the Morin household has now gone from token maxing to token budgeting. [SPEAKER_01] Yeah, it's not negative about AI. No, it's such a sensational headline. [SPEAKER_01] Not sensational. [SPEAKER_04] That was her ex post. It wasn't a headline. Your wife put that on the internet. No, I know. [SPEAKER_03] I put it because I'm using all these models. It was a lot. [SPEAKER_04] How much is your household spending on tokens? [SPEAKER_04] I'm just going to say it was a lot. Dave keeps some secret credit cards from me. [SPEAKER_04] So I don't know. There's the real answer and then what I think is the answer. [SPEAKER_04] But even my own, I was like, oh my gosh, how am I spending all this money? [SPEAKER_04] Because I have to. I'm testing all of them. I'm on every, I'm playing with all this stuff all the time. [SPEAKER_05] And so I have done some consolidation, and I'm really proud of my financial cleanup work. [SPEAKER_05] So who are the losers in the consolidation? [SPEAKER_04] Anthropic. [SPEAKER_04] Yeah, probably Anthropic. No question. Those agent turns were being done by Opus 4-6 or Opus 4-7 or Sonnet 4 or Sonnet 5. And there's no point in paying for those turns anymore when you can get the same quality from GLM 5-2. Well, okay. But just to play devil's advocate to that, I obviously broadly agree with you, right? Broadly. [SPEAKER_01] But I also just don't care about any of this. It doesn't matter. It's not that expensive. [SPEAKER_01] And my view is I just use whatever. And I'm not switching because I just don't care about the money. It's not enough to matter. [SPEAKER_01] But Sam, that might be your opinion, but enterprises do care. And we are speaking to a lot of people running enterprises. [SPEAKER_01] Of course. And that's fine. And I'm happy for you to interrupt me anytime you want. But the... [SPEAKER_05] Obviously, enterprises aren't going to operate that way. And as they go from experimental budgets to whatever, they'll be figuring this stuff out and optimizing it. Sure. I will say on a purely personal basis, though, the reason... [SPEAKER_05] If there's a pro-Anthropic or model-whatever argument, the argument is this. [SPEAKER_05] Paying 10 bucks for something, and let's pretend even 1% of the time it doesn't fuck it up or it comes back and asks... [SPEAKER_05] It doesn't ask me a stupid clarifying question. I'll pay for it, right? [SPEAKER_05] There is this veil of, I don't care. It's not that expensive. Whatever. And even if it's only going to save me one email once a month, whatever. You know what I mean? So I do think there's... And that's going to be the interesting nexus here about trust and accuracy around open source. Anyone at scale, this whole story, obviously there's going to be massive pricing pressure on multiplying big numbers. [SPEAKER_01] And I totally agree with Dave. I've been saying this, I think, as long as anyone. I think we all have. Which is, you're just multiplying big numbers. Open source has caught up. It will be where the volume is. It's a volume game. [SPEAKER_01] These are monopoly... I'm sorry. These are utilities with none of the financial benefits of being utilities. [SPEAKER_01] Which is why they're all in Washington trying to put up trade barriers and get, blah, blah, blah. We get that. But I will say the only counterargument to all that is in scenarios where you're not 100% sure what the answer should look like, right? [SPEAKER_04] Or there's real return on it being slightly more accurate, and you just don't care about the cost. You'd rather save another back and forth to clarify something. And I will say, I moved my infrastructure up to Fable. I only did it just now. I literally haven't upgraded because it hasn't been relevant. But I finally was like, eh, I just told my bot over email, I was like, just start using Fable instead of Opus, whatever. And it did. I'm like, oh, this actually is a little better. That's cool. It's not going to change my life, but it did one or two smart things. That's fine. You know what I mean? That's kind of the vibe I'm at at this point, if that makes sense. Which is, it's not exciting. It's not important. But it's fine. Which I think that sentiment, Sam, kind of matters to... No, I agree. I think that... But that's the point of the curve we're at. The business question we're asking is, does the revenue ramp continue to look like this over the next year? [SPEAKER_05] And I think that the sentiment that you're referring to actually matters, which is that it's kind of all just boring. [SPEAKER_05] They don't move forward that much every time. [SPEAKER_05] Well, I think it's kind of like, imagine you have... [SPEAKER_05] It's just like, I go back to you've got to use human analogies. It's like, look, you all were... [SPEAKER_05] We were all talking to high school students. And then we're talking to college students. [SPEAKER_05] And we're like, oh, these college students. I'd rather talk to them than the high school students. They seem better. [SPEAKER_05] And then we all started talking to PhD students. [SPEAKER_05] And I think that the sentiment that you're referring to actually matters, which is that it's all just boring. [SPEAKER_05] They don't move forward that much every time. [SPEAKER_05] Well, I think it's, imagine you have... [SPEAKER_05] It's just, I go back to you got to use human analogies. It's like, look, you all were... [SPEAKER_05] We were all talking to high school students. And then we're talking to college students. [SPEAKER_05] And we're like, oh, these college students. I'd rather talk to them than the high school students. They seem better. [SPEAKER_05] And then we all started talking to PhD students. [SPEAKER_05] We're like, oh, these PhD students, these grad students seem smarter than the college students. Good. [SPEAKER_05] And now we're at the point where, all right, well, maybe there's a PhD student who's getting an A versus a B in some class. And if I really care about specifically that one topic and that one scenario, God bless that they're now getting an A versus a B. [SPEAKER_04] But I'm actually asking it to fucking fold laundry, and the college kid can do it. Right? [SPEAKER_04] And that's where we're at, and it's fine. [SPEAKER_04] I'm glad that people should keep making slightly smarter PhD students. [SPEAKER_04] But it's not going to change anyone's life. [SPEAKER_04] Okay. Britt, what do you think? This is back to my evolution, not revolution, that I brought up last week. Which is, I agree. It's fine. It's cool. I've been using Fable in the past week to do some interesting things. [SPEAKER_04] I think it's failed less at getting things wrong. [SPEAKER_04] So that's good. [SPEAKER_03] But I don't know. [SPEAKER_03] Demis was posting this whole thing today about AGI yet again and how we need a standards body to review all the models and all these different things. And then I was like, is AGI really years away? Is this really happening? And I know we debate this a lot. [SPEAKER_04] But... [SPEAKER_01] Well, AGI is a meaningless term at this point. [SPEAKER_01] It doesn't mean anything. [SPEAKER_01] It's just computers. [SPEAKER_05] Well, let's... I want to pivot us a little bit to... [SPEAKER_01] Because we'll go nowhere on this topic, which is a good topic. [SPEAKER_01] But, okay. [SPEAKER_01] I am hearing more buzz about real regulatory action around open source because of the issue. [SPEAKER_01] Open source or open weights, Jess? [SPEAKER_05] Open source, right? [SPEAKER_01] So, folks inside the government saying, okay, we have a problem here. [SPEAKER_01] Anthropic and OpenAI are worried about the Chinese models undercutting them. [SPEAKER_04] At the same time, objectively as a country, if you're worried about national security, you're not thrilled with everyone going to the Chinese models. [SPEAKER_04] At the same time, if the Chinese models are also advancing more rapidly because they're distilling the U.S. frontier models, that doesn't seem great for protecting intellectual property in the U.S. So it seems to me whether or not something happens is up to the Washington gods and goddesses. [SPEAKER_01] But there's clearly... Dave, you've now said for weeks, you've talked about GLM and all these things. Do you think the U.S. will let this situation just continue to exist? [SPEAKER_03] I mean, it depends how good OpenAI and Anthropic are at lobbying to create margin for themselves. [SPEAKER_01] Yeah. [SPEAKER_03] That's what it comes down to. [SPEAKER_03] This is what I don't like about our moment in history, is I don't think we're playing capitalism anymore. [SPEAKER_05] I think we're playing some political capture game that is way less interesting than capitalism, right? [SPEAKER_03] And so that sucks, right? [SPEAKER_03] And we should compete in fair and open markets as much as possible, and the best team should win, and whatever. [SPEAKER_03] And now we're just in this game of people... [SPEAKER_03] Which is, again, it's been stories for a while. [SPEAKER_03] AI is scary, blah, blah, blah. [SPEAKER_03] All these types of things. [SPEAKER_03] But now we've moved to a new version of the game, which feels much more desperate, right? [SPEAKER_03] Which is, man, we better get some regulatory capture going because otherwise we're in trouble. [SPEAKER_03] Sam, what's your point of view on what the state of play with... And, Dave, you finish your point. [SPEAKER_03] And I want your point of view on this question, too. But what is in the U.S. interest with regards to these open source Chinese models, though? [SPEAKER_01] Yes, it would be regulatory capture if... [SPEAKER_01] I don't think it's Chinese. I think it's just... That's too specific, right? It's just open source, right? [SPEAKER_01] In broadly. Look, this goes back to years ago. I used to hate on GDPR constantly. And specifically this thing called the right to be forgotten, which is one of the dumbest laws ever written. Because you're like, how the hell are you going to enforce that? You can't... It makes no sense, right? [SPEAKER_01] From an intellectual... [SPEAKER_01] It's not how the Internet works. [SPEAKER_01] It's like people used to make T-shirts with encryption algorithms. Encryption, we can go back further. In the early Internet, there was this big argument about whether encryption technology was a weapon. And whether it could be exported from the U.S. legally or not. And then all these hackers would literally just put encryption algorithms on T-shirts and walk around. Because they're so simple, you can't stop this from moving cross-border, right? And so the basic point with all this stuff, again, is it's one thing if you're just okay, the Chinese models are bad. And we're going to put up a trade embargo on copying Chinese... Using Chinese models from China. But we're going to do it in the U.S. None of that's how the Internet works. That's not how this shit works, right? And so to me, weirdly enough, there's even a weird argument about, oh, would you stop American open source models? Because that's technically our domain for me to develop, but not the Chinese one because that'll cause a trade war. That's insane, right? All this shit's just insane and doesn't map to how capitalism or the Internet works, right? And that's what I object to. But we're going to do it in the U.S. None of that's how the Internet works. That's not how this shit works, right? And so, to me, weirdly enough, there's even a weird argument about, oh, would you stop American open source models? Because that's technically our domain for me to develop, but not the Chinese one because that'll cause a trade war. That's insane, right? All this shit's just insane and doesn't map to how capitalism or the Internet works, right? And that's what I object to. Is there a path back or what? I agree with everything you're saying. What do you think solves it or ameliorates it, or is it a one-way door? One thing that is different today, Wednesday, July 15th, is that America now has at least one open weights model that just went out the door. Miramirati's Thinking Machines appears to be a really solid open weights model. It's an almost 975 billion parameter model, million token context window, mixture of experts. It's really good. And this thing's open weights, and I'm free. You can get it on Hugging Face today. So that's the first American really heavy-hitting open weights model. Right. There's Reflection and some other things. Nemotron. Yeah, NVIDIA would be the biggest competitor, right, in the U.S.? Nemotron's fine. [SPEAKER_05] This is substantially better. This one's competitive with the Chinese models. I just think, at the end of the day, stopping people from moving software around the world, it's the same thing as stopping them from moving books around the world. It's just really weird, right, as a concept to push on. [SPEAKER_05] And I'm just... [SPEAKER_05] So, Sam, does this become... [SPEAKER_05] I think, if I'm putting myself into the ideologies of the big labs, is this more like the Napster fight than it is anything? Where, when Napster happened, that was also effectively songs moving around or software moving around the Internet, right? And so what happened is entire... What was it called? The DMCA kind of... Metallica. It was Metallica. It was Jimmy Iovine. It was the record labels. They were, we are powerful. We are going to get control over this. And today, on modern computers, moving around digital music is still incredibly difficult and very hobbled. It's a pretty gnarly system they built. The problem with it is there's some great irony to that example, although I think it's a good one to discuss, which is: what do the models do immediately? OpenAI, whatever. They went and ripped the entire Internet and broke all copyright and stole everything, right? [SPEAKER_01] It's a, in order to build version one, and then they get all mad about people distilling them, and that's literally just people doing to you what you did to everyone. So I think... I get it. Yes. [SPEAKER_01] But I'm trying to get into what's the policy fight here? [SPEAKER_01] Is the policy fight really similar? [SPEAKER_01] Are we trying to, in Anthropic's case, they believe this is insanely unsafe. [SPEAKER_05] Only they should be able to decide what's safe and what isn't. [SPEAKER_05] And so let's lock it all down because biological weapons and weapons and weapons, weapons, weapons, right? [SPEAKER_05] It's a great... [SPEAKER_05] There are two things that end all arguments, right? [SPEAKER_05] National security and then child porn. [SPEAKER_05] Those are the things you use if you want to win an argument because no one's going to argue against them. [SPEAKER_05] This is, child porn was the best example ever because all you have to say is child porn, and people lose their shit and will sacrifice anything to stop it. [SPEAKER_05] Maybe that's good. [SPEAKER_05] Maybe that's bad. [SPEAKER_05] But it's a tough one, right? [SPEAKER_05] It'd be... [SPEAKER_05] The analogy would be if listening to Metallica would explode your brain, right? That would be the argument. We have to stop Metallica. I mean, it might. [SPEAKER_05] Depends on how loud. [SPEAKER_05] Yeah, exactly. [SPEAKER_05] So, I don't know. [SPEAKER_05] I just... [SPEAKER_03] I don't know. [SPEAKER_03] My thing more is I'm just depressed about the entire thing. [SPEAKER_03] I offer no advice. [SPEAKER_03] I think this has gone... [SPEAKER_03] All of these AI wars are politics and not business or technology at this point. [SPEAKER_03] And that's a bummer. [SPEAKER_03] But do you see that changing? [SPEAKER_03] But, Sam, between last week and this week's pod, you had Meta released MuseSpark at a significantly lower cost, which was a good business move. [SPEAKER_03] And I had access to it early. [SPEAKER_03] And I was playing with it. [SPEAKER_03] And I was, wow, this is really fast and really cheap. [SPEAKER_03] It is good. [SPEAKER_03] It's good. [SPEAKER_03] I might actually use this instead of GLM. [SPEAKER_03] But then... I told my bot to use it for some things. After it released, it was just kind of eh. And so... [SPEAKER_04] Well, it's also been six days. [SPEAKER_04] I understand. [SPEAKER_04] But these were smart business moves. [SPEAKER_04] That's a really smart business move. [SPEAKER_04] Use your money tree. [SPEAKER_01] We've talked about Meta's advertising money tree for years now. [SPEAKER_04] Use that to subsidize the... Lower the cost. Compete on cost. But it kind of was... I guess we'll see. You're right, Jess, in a month. [SPEAKER_04] Whether or not that mattered or not. [SPEAKER_04] Well, look. [SPEAKER_04] We're back at a story where everyone's, well, it's the application layer. [SPEAKER_04] I don't know if that's true. I think what it is is there are companies that can use this shit to be way more profitable. [SPEAKER_04] They will use that shit. [SPEAKER_04] Use that to subsidize the... [SPEAKER_04] Lower the cost. [SPEAKER_04] Compete on cost. [SPEAKER_04] But it was... [SPEAKER_04] I guess we'll see. [SPEAKER_04] You're right, Jess, in a month. [SPEAKER_04] Whether or not that mattered. [SPEAKER_04] Well, look. [SPEAKER_04] We're back at a story where everyone's, well, it's the application layer. [SPEAKER_04] I don't know if that's true. I think what it is is there are companies that can use this shit to be way more profitable. [SPEAKER_04] They will use that shit. Right? God bless. And those companies will be better. And that's the story of this AI stuff. The actual one model to rule them all. We're clearly past that. Right? The fact that who's ahead keeps leapfrogging. And open source is right there. It's like, we're just multiplying big numbers. And then there's going to be an attempt, because the fundamental business structure is not playing out in a way where any of these trillion-dollar companies really win. Right? [SPEAKER_04] They're just in an expensive forever war. [SPEAKER_04] Right? [SPEAKER_04] There's going to be a... [SPEAKER_05] Take a shot at regulatory capture. [SPEAKER_05] Because that actually is the smart business move. [SPEAKER_05] I guess what I was trying to get into, Sam, is what is making... [SPEAKER_05] What has you depressed? [SPEAKER_05] The fact that the AI war thing will not be decided by technology. [SPEAKER_05] It won't just happen naturally by the laws of economics, which gets you the best product at the cheapest price. [SPEAKER_05] People are making plays which are political, not business. [SPEAKER_05] Right? [SPEAKER_05] If that makes sense. [SPEAKER_05] I see. [SPEAKER_05] So if it goes this way. Right now, it is actually playing out pretty purely capitalistic. Right? Yeah, yeah. [SPEAKER_05] If it plays out capitalist, I think it's very cynical for these companies to even try to switch the field of play from business to politics. [SPEAKER_05] Right? [SPEAKER_05] I think that's a very cynical, anti-American thing to do. [SPEAKER_05] To be like, okay, we're not going to play and compete in capitalism in a fair and open system. [SPEAKER_05] Instead, we're going to switch our venue to regulatory capture because it's the only play we have. [SPEAKER_05] And I get why they do it. And to some degree, as businesses, I don't grudge them the play they need to make. [SPEAKER_05] It's like the classic, you can't win the game, so change the game. [SPEAKER_05] But it's also just a bummer. [SPEAKER_05] Right? [SPEAKER_05] Because you're just like, man, politics sucks. [SPEAKER_05] And business and economics are awesome. [SPEAKER_05] And... [SPEAKER_05] We haven't talked about this much, Sam, but do you believe that there is real danger? [SPEAKER_04] Could there actually be an argument for danger on the other side? [SPEAKER_04] For sure, but I don't think anything you're going to do regulatorily is going to change that. [SPEAKER_01] Right? [SPEAKER_01] But isn't that what Demis' post is about this week? [SPEAKER_01] Isn't that what Anthropic mostly says? [SPEAKER_01] That we want to... This stuff is dangerous. But the problem is, it's being talked about as though any one company can control it. Or any one country can control it. [SPEAKER_04] You can acknowledge the fact that giving technical... [SPEAKER_03] Enormous amounts of technical leverage on all sorts of things has... [SPEAKER_03] It could be bad for biological weapons. [SPEAKER_01] Or this is like... [SPEAKER_01] You can do all sorts of bad stuff with technology. [SPEAKER_01] You can do all sorts of good stuff with technology. Do all sorts of bad stuff with technology. That's what technology is. If we had a monopoly on technology in the U.S., that would be different. [SPEAKER_01] We obviously don't. [SPEAKER_01] Right? [SPEAKER_01] And so, in some ways, having a... [SPEAKER_01] It's like all this stuff about global warming and the environment. [SPEAKER_01] Where one country can take action. [SPEAKER_01] But it's completely irrelevant because we live in a connected globe. [SPEAKER_05] So it's a dumb thing to do. Right? [SPEAKER_01] That example. [SPEAKER_01] That's a good example. [SPEAKER_01] I hadn't thought of that. [SPEAKER_01] It's one of those things where it's kind of like you can go and be like, [SPEAKER_01] this is dangerous. [SPEAKER_01] And you're not wrong. Of course it is. Now we can have an army of drones do things. That sounds pretty wild. [SPEAKER_05] Right? [SPEAKER_05] But you taking any action as a single country does nothing. [SPEAKER_05] Right? And all it does is protect certain companies. Right? And their market caps. So, just to recap for people, what Demis, the AI leader at Google, proposed was, it was this idea that there should be a separate government body that reviews frontier models for 30 days prior to their release. [SPEAKER_05] Are we talking the FCC, kind of a thing? [SPEAKER_05] Kind of. [SPEAKER_05] Yeah. [SPEAKER_05] I mean, it is so vague that any way I answered your question would be a guess. [SPEAKER_05] But his point was that it would be comprised of technical people and technologists. [SPEAKER_05] And instantly got the thumbs up from everyone in AI. [SPEAKER_05] I mean, all the AI CEOs, in part because it wasn't really saying anything, in my opinion. [SPEAKER_05] And also everyone has their different style. [SPEAKER_05] But his style is, I worked my entire career on AGI because I believe it [SPEAKER_05] has the potential to do all of these things. [SPEAKER_05] But who knows if it will, right? Trying to pull back the Armageddon is nigh version of the stuff from the top labs. But I think the reality is these proposals are just all over the place. [SPEAKER_05] But to your point, Sam, that this is about politicking, you put something out like that [SPEAKER_05] to show that you're a thoughtful, concerned participant in the ecosystem or something, right? [SPEAKER_05] You don't do it because it's an actual tangible roadmap. Right. has the potential to do all of these things. But who knows if it will, right? [SPEAKER_01] Trying to pull back the Armageddon-is-nigh sort of version of the stuff from the top labs. [SPEAKER_01] But I think the reality is these proposals are just all over the place. But to your point, Sam, that this is about politicking, you put something out like that to show that you're a thoughtful, concerned participant in the ecosystem or something, right? You don't do it because it's an actual tangible roadmap. Right. No one has any roadmap that makes any sense that actually does anything, right? And so, if the plan was we, America, need to take our military and bomb every matrix multiplication facility globally, right? And we're going to have all of them. And then we're going to control AI. [SPEAKER_03] That would at least be logical. [SPEAKER_01] I'm not saying you should do it. [SPEAKER_01] You obviously shouldn't. [SPEAKER_01] But I'm saying at least that is a plausible thing. [SPEAKER_01] We're going to make it illegal to have a computer more than this powerful [SPEAKER_01] outside of the U.S., right? [SPEAKER_01] It's insane. [SPEAKER_01] It's World War III. [SPEAKER_01] But it's logical, right? With the other stuff, people, these, we're going to do a committee on this. It's like, God bless. [SPEAKER_01] Enjoy being on a committee forever. [SPEAKER_04] The only thing you could possibly do, what you could do, is somehow say, if you're [SPEAKER_01] an American taxpaying, law-abiding company, which is a lot of companies, but certainly [SPEAKER_01] not everyone, you have to only use this and you cannot use that. [SPEAKER_01] And in theory, a bunch of people and lawyers would make that happen. And then you'd accomplish nothing except for making those companies very valuable, right? Nothing. You accomplished literally nothing except for that. Okay. I'm going to give you guys a topic and then I'm going to bounce because I'm going to another [SPEAKER_01] Benson Boone concert tonight, and I'm super excited and not missing it. [SPEAKER_01] That is more important. [SPEAKER_01] It's your fourth Benson Boone concert. [SPEAKER_01] Guys, Jessica is getting on an airplane to see Benson Boone. That's the point we've reached. This is a big deal. For not the first time, the eighth time or something. I'm crossing a mini ocean. Are you actually in love with Benson Boone? I think she might be. This is the story of two music. Britt and I are going to Chris Stapleton this weekend. You're going to Benson Boone. [SPEAKER_05] I am not in love with Benson Boone, but there is something about seeing him perform live that [SPEAKER_05] I just find electrifying. He does seem good. [SPEAKER_05] He moves his body well. [SPEAKER_05] Guys, his voice is absurd, and I need to hear him perform his new song. [SPEAKER_05] And it's not as far as it seems, although I do have to take four modes of [SPEAKER_05] transportation to get there. [SPEAKER_05] I'll go in the Bay Area with you sometime. Jessica asked me if I wanted to come, and I was like, I'm not going to get between you and Benson. Go enjoy your Benson evening. I think that was a wise husband. But I'm bringing another couple. I'm bringing two friends. [SPEAKER_01] I'm more of the merrier. [SPEAKER_01] Are they both coming now at the last minute? They're both coming. This is the bandwagon. Well, now I'm glad you got two hotel rooms because that would be a little weird after the Benson. It would have been a little weird when my girl trip was crashed by her boyfriend. [SPEAKER_01] Let me know when you come back. You can give me your stories of your wild bender [SPEAKER_01] with Benson. [SPEAKER_01] I'm so excited. [SPEAKER_01] Guys, if you have not listened to his new song, just listen to it. It's fine. Wanted Man tour. You still have time to get tickets. [SPEAKER_05] Wanted Man tour. Oh, I get it. Oh my God. This is what it's called. Story writes itself. It just does, doesn't it? And as much as I could talk about AI with you guys, I'm not missing his first song. So with that, I suggest. I'm really happy we have this all on a podcast because when I report back next week, [SPEAKER_01] there could be amazing. [SPEAKER_01] We could have amazing documentation of what happened. I want you guys to talk about the report now that OpenAI's first consumer hardware device [SPEAKER_01] is basically Alexa. [SPEAKER_01] It's a smart tool. [SPEAKER_01] It's a developer tool, though. [SPEAKER_01] It's a developer tool. [SPEAKER_01] No, you guys are conflating two things. [SPEAKER_05] Oh, oh, oh. [SPEAKER_05] There have now been more reports, The Information, Bloomberg, others. [SPEAKER_05] Bloomberg came out with. [SPEAKER_05] So we're not talking about the merch we were talking about earlier. [SPEAKER_05] We're not talking about the merch. [SPEAKER_05] We're talking about the fact that sometime this year, they are expected to unveil a smart [SPEAKER_05] speaker. [SPEAKER_05] I swear, it's Alexa. [SPEAKER_05] What this means, will anyone care? [SPEAKER_05] No, it's a portable smart speaker, isn't it? [SPEAKER_05] You guys can discuss. [SPEAKER_05] Sorry, I'm here for someone else. [SPEAKER_05] Bye. [SPEAKER_05] Oh, I got it. [SPEAKER_05] That's a Benson-doing-there for anyone. [SPEAKER_05] And now we don't have to discuss it. [SPEAKER_05] We can keep discussing what we were already discussing. No, I do think we should change topics. Right, which is if I'm going to be married in a week or if my wife is now going to hang out with Benson Boone. [SPEAKER_02] I feel like Samson is in his I-don't-give-a-fuck phase of his 40s. So it's totally fine. 100%. Which is great. Does anyone care about the smart speaker that OpenAI has? I don't care. No. Did you think they were going to launch this or something else? We can keep discussing what we were already discussing. No, I do think we should change topics. Right, which is if I'm going to be married in a week or if my wife is now going to hang out at Benson Boone. [SPEAKER_02] I feel Samson is, I don't give a fuck, face of his 40s. So it's totally fine. 100%. Which is great. Does anyone care about the smart speaker that OpenAI has? I don't care. No. Did you think they were going to launch this or something else? You didn't care either way, it sounds like. [SPEAKER_05] You don't think any hardware OpenAI launches will be successful? [SPEAKER_05] I did think it was going to be something you talk to. [SPEAKER_05] It just seems very clear that that's. [SPEAKER_05] Of course it means something you talk to. [SPEAKER_05] Verbally, though. [SPEAKER_04] That seemed clear to me. [SPEAKER_04] Are we clear that Sam Altman paid Johnny Ive $8 billion to make him an Alexa? [SPEAKER_04] And to get sued by Apple. [SPEAKER_01] That we actually paid them more than $8 billion. [SPEAKER_04] I had people complaining to us that we didn't talk about the Apple lawsuit last week. [SPEAKER_04] And I was like, look, it came out after we recorded. [SPEAKER_04] I do think it's fine. [SPEAKER_04] I actually think this whole thing is pretty interesting. [SPEAKER_04] Because I do think, look, Sam Altman is an incredible storyteller. [SPEAKER_04] An aspiring Elon-level storyteller, I would say. In terms of narrative capitalism. And using that for cheap capital. The Johnny Ive acquisition thing. [SPEAKER_04] Especially if it just turns out to be an Alexa. Is particularly interesting for this. Because there was one step along that line of, oh, I'm buying. I'm going to pay a ton of money for Johnny Ive. [SPEAKER_04] Because it will make raising more capital cheaper, not more expensive. [SPEAKER_04] So it actually, even if it costs a billion dollars on paper. [SPEAKER_04] Is a free-to-negative-cost acquisition. Just on the brand of Johnny Ive. [SPEAKER_05] Which is an interesting thing. [SPEAKER_05] Yeah, totally. [SPEAKER_05] Who are other people who have $5 to $10 billion brands? [SPEAKER_05] Other legacy people. This kind of fits with the... [SPEAKER_05] Now that Twitter shows me my friends again. [SPEAKER_05] I actually saw your post about the Hollywoodification of Silicon Valley. Or, I guess, the world. Which I kind of like. [SPEAKER_05] But that's kind of the same question, right? Explain it. I didn't see the post. [SPEAKER_05] Can you explain Sam's post? [SPEAKER_05] Who else are the Steven Spielbergs? [SPEAKER_05] Is the question. [SPEAKER_05] Yeah, well, I was saying, look. [SPEAKER_05] Everyone's joking about how Hollywood's dead. [SPEAKER_05] Which it obviously is. But I would actually argue in some ways it's... It's kind of like an Obi-Wan Kenobi situation. [SPEAKER_05] Which is, by striking me down, I'll become more powerful than you can possibly imagine. [SPEAKER_05] Because now everything is Hollywood, right? [SPEAKER_05] And what does that mean? [SPEAKER_04] I mean, one... Companies are valued on what they actually do. Profit, revenue, whatever. And then a multiple on top of that, right? [SPEAKER_04] But everything is multiple-dominated. You can make your revenue go up. It doesn't matter. [SPEAKER_03] What matters is people's... [SPEAKER_03] What multiple they're going to assign you. [SPEAKER_03] And how they think about you. Studios, cabals rule everything. [SPEAKER_05] Just like in Hollywood, right? [SPEAKER_01] You need to be able to align with Club Elon or Club OpenAI or Club Meta or Club Google [SPEAKER_01] to be valuable. [SPEAKER_01] If Elon says Space Data Center overnight, anything that has Space Data Center in the title is worth a fortune. [SPEAKER_04] Even if the day earlier it was worth nothing, right? And so there's a whole cabal studio system developing. [SPEAKER_03] And then there's star power that really matters, right? Meaning there are people who can open movies. There are people who can open startups. So who else are those people in Silicon Valley? I mean, I think AI researchers is the easy answer. [SPEAKER_05] Alexander Wang, these types of people? [SPEAKER_05] No, no, no. [SPEAKER_05] Researchers. [SPEAKER_03] Oh, the researchers. [SPEAKER_03] Sorry. [SPEAKER_04] But Alexander Wang also got a billion dollars. [SPEAKER_04] No, I think Alexander Wang was a character who can open movies. [SPEAKER_05] For sure. [SPEAKER_04] He wasn't a Steven Spielberg. [SPEAKER_04] He's a leading man. [SPEAKER_05] Okay, so the researchers are worth billions of dollars, leading men or women. [SPEAKER_05] Well, no, but Sam's not even talking about what they're worth. [SPEAKER_04] It's that can they lead, can they open movies? [SPEAKER_05] Are they headliners, right? [SPEAKER_05] Are they? [SPEAKER_04] Or stars. [SPEAKER_04] Nira Maradi is, obviously, there was a whole, it was basically a comedy [SPEAKER_04] troupe, which was called OpenAI, that spun out several leading characters that were able [SPEAKER_04] to open movies, right? [SPEAKER_04] Is what I would say. [SPEAKER_04] I like this. [SPEAKER_05] OpenAI is SNL. [SPEAKER_05] Yeah, yeah. [SPEAKER_04] OpenAI, or Meta, was SNL for a while. [SPEAKER_04] And Google was clearly an SNL, and PayPal was an SNL. [SPEAKER_04] There's just these little things, it's like there are these platforms where people ping off each other, they get whatever. [SPEAKER_04] And then some of them get to lead movies and some don't. [SPEAKER_04] Some become Tina Fey. [SPEAKER_04] Okay, so Karpathy was another, right, that went to Anthropic. [SPEAKER_05] No. I'm just wondering who else there is. I don't know. [SPEAKER_04] I think you know, you know a star when you see them, Brit. [SPEAKER_04] I don't know. [SPEAKER_04] I think it's a little hard to, a priori, be like, well, this person's just... [SPEAKER_03] It goes through a little to your creator thing, Sam. [SPEAKER_04] And then some of them get to lead movies, and some don't. [SPEAKER_04] Some become Tina Fey. [SPEAKER_04] Okay, so Karpathy was another, right, that went to Anthropic. No. I'm just wondering who else there is. I don't know. [SPEAKER_04] I think, look, you know a star when you see them, Brit. [SPEAKER_04] I don't know. [SPEAKER_04] I think it's a little hard to, A prior to be like, well, this person's just... [SPEAKER_03] It goes through a little to your creator thing, Sam. [SPEAKER_04] I wonder, can you get the upside of these people somehow? [SPEAKER_04] No, no, for sure. [SPEAKER_04] I love that story. [SPEAKER_04] But look, at the end of the day, I said, how do you know someone's a star? [SPEAKER_04] Well, there's one of two ways. Either they opened a movie or a few movies, and they were super successful at those openings. [SPEAKER_04] Jack Dorsey is a leading man, isn't he? [SPEAKER_04] Elon's clearly... [SPEAKER_04] Elon is the... [SPEAKER_04] What's his name? [SPEAKER_04] He's the Tom Cruise. [SPEAKER_01] No, Elon's Spielberg at this point. [SPEAKER_04] It's, you can bring him to the... [SPEAKER_04] You can bank his movie from before the day, just because Elon's attached, right? [SPEAKER_04] And so, there's characters like that. [SPEAKER_04] And then the only way you become a leading star is you... [SPEAKER_01] Someone makes a bet on you, and you become a leading star. [SPEAKER_01] But there's a... [SPEAKER_03] I don't know. It's not a perfect analogy, but there is something going on that I think is relevant here. Yeah. [SPEAKER_03] Okay. [SPEAKER_03] Well, regardless, Johnny Ive was theoretically worth, whatever, six to eight billion. Well, again, only because he actually... [SPEAKER_03] Because the market paid for him. [SPEAKER_03] Right. [SPEAKER_03] Right. [SPEAKER_03] Because the market paid for him. The market paid for him. [SPEAKER_03] Even if all he was making was a mini Alexa. [SPEAKER_03] There has to be a suite of products that he's working on. [SPEAKER_03] I would imagine there's more than the Alexa coming. [SPEAKER_01] Isn't it funny how the Alexa has become a generic brand? [SPEAKER_01] And a big Alexa and a small Alexa? [SPEAKER_01] You've got the Google Home. [SPEAKER_01] You've got the HomePod. You've got all the Alexas in the house. A car Alexa? [SPEAKER_03] All of the above. [SPEAKER_03] Sam, what product would you most want to create if you had to create a hardware AI product? [SPEAKER_01] Something that I could sell to someone before launch and take the cash. [SPEAKER_01] Do you think there's any white space opportunities for that right now? I'm completely uninterested in hardware. I like my iPhone. [SPEAKER_05] I think it's fine. I like my Garmin watch. [SPEAKER_05] I could imagine putting a button on my Garmin watch with a microphone. That would be useful. Would you ever wear something that's turned on all day, listening and tracking everything about you? [SPEAKER_05] I would be the poster child for this in theory. [SPEAKER_05] Right? [SPEAKER_05] Because I have built my own custom shit to upload every photo. And I spend a huge amount. I have millions and millions of data points consolidated on my health. [SPEAKER_05] Literally millions through apps I built. So I would be the poster boy for passive listening. [SPEAKER_01] I tried the narrative clip back in the day. I've played with every generation of this. [SPEAKER_03] At Finn, we built what's now in Vogue, which is screen recorders plus clickstream data, and built huge data systems. So I love the theory of track everything. In practice, most of what people say isn't that valuable. So I don't know that passive listening is actually all it's cracked up to be at the end of the day for usefulness. [SPEAKER_01] It's mostly a lot of ums. [SPEAKER_01] But look, marginally lowering the barrier to recording stuff is never going to be bad. [SPEAKER_01] I just don't know how valuable it is and how differentiated it'll ever be. [SPEAKER_03] Dave, what's your take? [SPEAKER_03] I don't feel like you're a passive recorder type. [SPEAKER_03] Is this because you're speculating that this is what it's going to be? [SPEAKER_01] Yeah. [SPEAKER_01] I feel like the way we were seeing rings and necklaces and Alexas and spin-offs of Alexa. Is this where the future is headed? If you just look at the release of, I don't remember what the API is called, but I think it's called Live. It's clear that streaming real-time interaction with models is one of the major things that OpenAI is researching. And so it seems clear that your question is actually not the question. The question is, would you have a participant in your conversations every day that is a third party that's interacting with you in real time? [SPEAKER_05] I think that's the thing, because a lot of tech-forward people have spent their last year using Granola and then using those transcripts to do things. [SPEAKER_05] That's how they think the future looks. [SPEAKER_05] I don't think that's actually how the future looks. I think the future looks much more real-time and that you're just going to be streaming in and out of these things in real time. And they'll tell you things, and it's going to look much more like that. It's going to look like what this model was this week. Meaning you won't have to prompt something later. [SPEAKER_05] It's just going to be doing things for you. [SPEAKER_05] No, that's what this model this week is that everybody's so impressed with. Which model? It's called Live, I believe. I don't think you've played with it. It's a voice model. This is the one where we can talk and listen at the same time. Yeah. And so it's a purely live, real-time voice model. And it seems clear that that's where this stuff is going. And if I were to guess what this hardware stuff is, it's based on that. Because I think a lot of people are looking at... [SPEAKER_03] Who released Live? OpenAI. [SPEAKER_03] Okay, just checking. I'm clarifying for the listeners. It's called Live, I believe. I don't think you've played with it. It's a voice model. This is the one where we can talk and listen at the same time. Yeah. And so it's a purely live real-time voice model. [SPEAKER_05] And it seems clear that that's where this stuff is going. [SPEAKER_05] And if I were to guess what this hardware stuff is, it's based on that. [SPEAKER_05] Because I think a lot of people are looking at... [SPEAKER_03] Who released Live? OpenAI. [SPEAKER_03] Okay, just checking. I'm clarifying for the listeners. And so that's part of the announcement, that it can talk, it can listen, all at the same time. [SPEAKER_03] So it can be doing things proactively. [SPEAKER_03] Everyone who's used a voice model knows that it's been this pretty janky experience where if you say the wrong thing, it interrupts it. It restarts the inference. [SPEAKER_05] It gets confused. [SPEAKER_05] This is really good. [SPEAKER_03] It doesn't do that stuff anymore. [SPEAKER_05] You can just talk to it. Have you played around with it? [SPEAKER_01] Yeah. [SPEAKER_01] What have you done with it? [SPEAKER_01] You can just do it on ChatGPT and play with it. [SPEAKER_01] All right, so it seems the consensus is we believe voice models that can listen and talk all at once are going to be... [SPEAKER_01] They're part of the present. [SPEAKER_05] They're part of the future. [SPEAKER_05] You both seem less bullish on recording every single thing every single day. There are very funny Instagram memes about people talking to the voice mode, being like, or some guy's like, ChatGPT, build me a billion-dollar business. Make no mistakes. [SPEAKER_01] Be quiet, be quiet, be quiet. [SPEAKER_01] Do this. Do this. [SPEAKER_05] SaaS app. [SPEAKER_05] And some of those will get better, I guess. Maybe that's an impact. Look, I think there's a different question here, which is that agents in general have a long way to go in order to understand you and your daily context. And that's the bigger, longer story here that I think maybe you're getting at, Britt, is that you just have to have a lot of context in order to make these things as smart as everybody wants them to be. And that's going to require totally different ways of thinking about memory, totally different ways of thinking about learning algorithms, totally different ways of capturing context. [SPEAKER_03] That's what the entire robotics industry is focused on, is how do you understand the world? [SPEAKER_03] How do you understand physics around people? [SPEAKER_03] How do you understand people's internal state? There's so far to go in terms of getting machines to understand us other than through the text we type into them or the text we speak into them. That, to me, is very interesting, but kind of a completely different set of questions. What else is captivating you guys lately? Anything exciting on your plate, Sam? [SPEAKER_03] I've started messing with peptides, which are quite interesting. [SPEAKER_03] We'll do another one of those first. Oh, yeah. We were supposed to come back to this. Yes. I'm so glad you brought that up. What does this mean? [SPEAKER_05] Oh, did you talk to Mazio? [SPEAKER_05] Is he going to come on the pod? [SPEAKER_05] Oh, I'm sure he'd love to. [SPEAKER_05] I'll message him now. [SPEAKER_05] Yeah, I'm just taking some, and then I'm running a lot. [SPEAKER_01] It seems good. [SPEAKER_05] What are the before-and-after impacts on your health and body based on your peptide consumption? [SPEAKER_05] I've only been doing it for a week and a half. [SPEAKER_05] And so I don't think anyone thinks there's real impact yet. [SPEAKER_05] But I will just say I've been upping my training volume a lot in terms of the amount of running and working out. [SPEAKER_05] And I feel much better than I would normally expect to. [SPEAKER_05] Way better. [SPEAKER_05] Mazio says he'd love to come on the pod. [SPEAKER_05] So maybe we can do that next week. [SPEAKER_05] He can explain peptides and show us his bicep. Because that man loves a bicep. [SPEAKER_03] I mean, we've all known this guy, Matt Mazio, for 20 years now. [SPEAKER_03] I can say the before and after. [SPEAKER_05] He was my fraternity big brother when I was in college. [SPEAKER_05] Shockingly. [SPEAKER_05] I love that. [SPEAKER_05] Shockingly. [SPEAKER_05] Sam, is this an investable category? [SPEAKER_05] I don't know. [SPEAKER_05] I mean, Matt will tell us. [SPEAKER_03] He certainly thinks so. [SPEAKER_03] I mean, I don't know. [SPEAKER_03] He'll say yes, invest in my company. Yeah, I mean, I think his company's pretty good. [SPEAKER_03] Well, we know he'll say that. [SPEAKER_03] But I want a more sophisticated... [SPEAKER_03] I don't know. [SPEAKER_03] I mean, I think the question is, what do we think of Hims & Hers and Ro and things like that? [SPEAKER_03] Right? All these generics and compounding pharmacies and the whole supply chain there. [SPEAKER_05] I mean, there is brand value. If you can get lower CAC, there is a bunch of spend. I don't think this stuff is going to be FDA... [SPEAKER_05] I mean, forget FDA coverage. [SPEAKER_05] I don't think it's going to be covered by insurance anytime soon. [SPEAKER_05] So that really cuts the market way, way, way down. [SPEAKER_05] If you don't have payers involved, like GLP-1s, it's much harder to make huge businesses out of it. [SPEAKER_05] But I don't know. [SPEAKER_03] Who knows? [SPEAKER_03] This is a good question. [SPEAKER_05] I don't know, man. [SPEAKER_05] I've been hearing that the FDA, I think the FDA is actually meeting next week to talk about them. [SPEAKER_05] FDA has to figure out how they're going to get their shit together to approve so many new things. [SPEAKER_05] There's so many new drug discovery funds and companies... [SPEAKER_05] But not peptides. [SPEAKER_05] I know. [SPEAKER_05] I'm not. [SPEAKER_05] I'm just talking more broadly. [SPEAKER_05] I think what's coming at the FDA in the next 10 years is going to be insane. And they're going to have to figure out how to pace themselves more efficiently and get through all of these decisions. But yeah, I think peptides would be a really interesting topic. So I'm glad we're talking about it next week. Can I tell you that this is more of a feature launch, but I thought it was interesting in the realm of online dating. So Henge now has a feature called Friends Take. But not peptides. I know. I'm not. [SPEAKER_05] I'm just talking more broadly. [SPEAKER_05] I think what's coming at the FDA in the next 10 years is going to be insane. [SPEAKER_05] And they're going to have to figure out how to pace themselves more efficiently and get through all of these decisions. [SPEAKER_05] But yeah, I think peptides would be a really interesting topic. [SPEAKER_05] So I'm glad we're talking about it next week. [SPEAKER_05] Can I tell you that this is more of a feature launch, but I thought it was interesting in the realm of online dating. [SPEAKER_05] Henge now has a feature called Friends Take. [SPEAKER_05] You enlist up to 10 of your close friends who then have to share written feedback, photos, or voice notes about you. [SPEAKER_05] Finally. [SPEAKER_05] I love it. This has been so long coming. [SPEAKER_03] Oh my God. But Dave, I'll tell you a story you don't know, which is, Mike Galpert, who you've been working with on Open Claw. [SPEAKER_01] Yeah, of course. Do you know where he and I started 15 years ago in our product-in-New-York era? No. [SPEAKER_03] A dating app. [SPEAKER_01] Oh no, sorry. [SPEAKER_01] This is funny. [SPEAKER_01] Leave this in for the editing because it's funny. [SPEAKER_01] It wasn't him. [SPEAKER_01] It was Nate Westheimer. [SPEAKER_01] Or maybe it was, I can't remember who it was, Nate Westheimer or MSG. [SPEAKER_01] But we started a service with Chris Danzig, who you also know, called Yenta. [SPEAKER_01] Oh, I remember Yenta. [SPEAKER_01] I remember Yenta. [SPEAKER_01] Yeah, of course. [SPEAKER_01] And this was Yenta. [SPEAKER_01] We were like, you don't want to make your own dating selections. [SPEAKER_01] Other people should make them for you. [SPEAKER_01] And people have tried this before. [SPEAKER_01] But this isn't your friends picking your dating selections. [SPEAKER_03] This is like your references. It's like you have 10 references. [SPEAKER_01] I was on the board. [SPEAKER_01] Oh, I see. [SPEAKER_01] It's a reference. [SPEAKER_01] Yeah, yeah, fine. [SPEAKER_01] I was on the board of Hinge for six years, and we talked about this idea ad nauseam. [SPEAKER_05] And it's still shocking to me that it doesn't exist more broadly in the dating ecosystem. [SPEAKER_01] Well, to be clear, what you're saying, Britt, is this is not actually the help picking. [SPEAKER_01] This is like references. [SPEAKER_01] This is like LinkedIn references for Hinge. [SPEAKER_01] I would imagine, hopefully, this can evolve to picking. [SPEAKER_01] Because if you have networks of your, it's like your top five on MySpace, like we all used [SPEAKER_03] to know. [SPEAKER_03] And then those people know you so well, they provided references. [SPEAKER_03] And then maybe there's some social graph where they can move people into your profile [SPEAKER_03] or something. [SPEAKER_03] I think that would be a cool next feature move. [SPEAKER_03] But I do think, why don't we have this already for actual recruiting of talent? [SPEAKER_01] Shouldn't everyone have photo, video, voice note references from lots of people? [SPEAKER_01] I don't know. [SPEAKER_01] I think the way that we recruit is so outdated. [SPEAKER_01] We're doing so much recruiting right now. [SPEAKER_01] If you go use Vox Americanis, you will notice that the recommendations from other Americans [SPEAKER_01] is a key feature. [SPEAKER_03] When you follow someone, it asks you to recommend them or write about them, et cetera, endorsements. [SPEAKER_01] So I'm all for endorsement. [SPEAKER_03] Great. [SPEAKER_01] Well, we should bring clout back while you do that. [SPEAKER_03] So that people of status have higher weight. [SPEAKER_03] Clout was, no, but clout was so stupid. [SPEAKER_03] Clout always annoyed me. It was such a misunderstanding of how social capital works. [SPEAKER_05] So fundamentally. [SPEAKER_05] But people liked it because people liked scores. [SPEAKER_05] Yeah, it was gamified. [SPEAKER_05] Dave, anything else catching your attention this week? [SPEAKER_05] I just think it's awesome that Jay-Z turned New York City into a music festival last weekend [SPEAKER_05] and is back on the stage. [SPEAKER_05] Wasn't he six hours late on stage? [SPEAKER_05] That's how all the greatest musicians do it. [SPEAKER_01] No, the demand for the show, they originally were only going to do two nights. [SPEAKER_01] And they ended up doing a third because the demand was so insane. [SPEAKER_01] And then a bunch of people got into a pretty dangerous situation, it sounded like. [SPEAKER_01] The only thing I saw from this, because I get all of my information through Instagram, was a video of a guy being like, the only thing that saved my night waiting six hours for Jay-Z to show up was this woman twerking. And it was this woman in front of him in this concert twerking for hours. [SPEAKER_03] That's how I knew something happened. [SPEAKER_03] Here is my takeaway. [SPEAKER_05] I feel like Jay-Z and Beyonce actually do fairly little in the day-to-day of keeping [SPEAKER_01] their music and marketing brands alive. [SPEAKER_03] And then they can just, their superpowers, they can come together, throw the most insane events, still be relevant, and even include their family. [SPEAKER_03] Blue Ivy was on stage. [SPEAKER_03] And so they're becoming this dynastic music empire that doesn't have to work as hard as the other musicians. [SPEAKER_03] It's also very weird to see Jay-Z and Beyonce as the golden oldies in New York. [SPEAKER_03] Oh, you think they're old now and outdated? They are old. Jay-Z's 56. He's been doing this for 30 years. But weren't young people there? [SPEAKER_03] They're still relevant, right? Yeah, but this is a millennial oldies concert. Let's be real. Oh, really? And Rihanna. [SPEAKER_05] Yeah. [SPEAKER_05] Yeah. [SPEAKER_05] All right. This is like the millennials going to the Eagles. Ew. [SPEAKER_03] I hate that we're that old. [SPEAKER_03] Yeah. [SPEAKER_03] All right. [SPEAKER_03] Well. [SPEAKER_03] Jay-Z's first album came out when I was in high school, right? A lot of these songs. [SPEAKER_01] Yeah. [SPEAKER_05] And you're really old. [SPEAKER_05] I'm super old. [SPEAKER_05] You really are old days. [SPEAKER_05] We're all in our 40s.