Elon Doubles Down, 𝕏 Timeline Reactions | Andrew Yang, Ishan Mukherjee, Tony Holdstock, Tomer London, Andrei Serban, Tim Higgins

27:33And there's also certainly investor demand for an American humanoid robot play. Today, Figure Robotics raised a billion dollars at a$39 billion post-money valuation. When was that announced? That was announced. Months ago, right? Or leaked. It was like leaked. Leaked, what, six months ago? It felt like something that was designed to generate interest. Sure. And there were screenshots of emails and stuff. At the time, I was pretty blown away because that round was being, what it was rumored to be getting done at was greater than the market cap of Ford Motor. That's right. Where's Ford today? Ford is at$45 billion today.

28:11Oh. And so I saw the news today. That take. Age like milk, Jordan. Yeah. I said to John, I said, you might not like this, John. This might trigger you. But Brett Adcock has created more enterprise value than Henry Ford. That actually proved to be true. Henry Ford, as of current marks, has generated about a billion dollars more. Okay. Yeah. We don't need to get into revenue or really try to get too into the math. Yeah, let's skip all that. But yeah, the question - I mean, my read on this is just like, figure is nowhere near as in the zeitgeist as Tesla. And there are clearly investors who want to bet on humanoid robots in America.

28:50They can't even buy unitary stock. I don't know how you would. and so if you just believe that robots are going to be a thing like investing in Tesla doesn't seem like that crazy of an idea right? But I mean he's not just focused on that he's also focused on the next Tesla chip design Well one thing too, Unitree is planning their IPO around the$7 billion mark Sure, oh that's low, wow that's low Where are they, in Hong Kong maybe? It's going to be the Hong Kong Stock Exchange I wonder what it'll do. And given that they are worth it. They're shipping a lot of those robots. They're selling a lot of pups.

29:27They're selling a lot of pups. Oh, yeah, they're selling the pups. Robo pups. Anyway, worth a big deal. I think a Tesla robo pup would crush. That would be good. I would. What would you use it for, though? Would you not churn for that? Just to, well, so I pitched you on an office dog, and you said, that's insane. And I thought about it. It's a tragedy of the common situation. But a Tesla pup that Elon would just be able to zoom into at any time and kind of bark at us if he had announcements, I would be pretty interested in that. What about a Tesla puffer fish? A turbo puffer fish? A turbo puffer.

30:09You'd search every byte. Serverless vector and full-text search. Build from first principles and object storage. Fast, 10x cheaper, and extremely scalable. Check out a turbo puffer. he's also prioritizing AI5, the new chip. He's partnering with Samsung on that. Makes sense. You need edge computing in a self-driving context. You can't do all the inference in a data center. Colossus, you can do inference in the data center. You're probably training a big model. Regardless of what XAI is doing with Grok, having a bunch of, being GPU rich seems valuable to Tesla, right? And there's rumors that maybe those companies merge at some point.

30:47but the plan overall like how he's actually spending his time it seems in line with the opportunity in front of tesla it doesn't seem like wildly wildly like he's managed to almost 2x the stock yeah since april he certainly like got got the flow back and i remember i i distinctly remember mid curving it around april yeah thinking just looking at the looking at tesla earnings Look at them losing market share globally. Look at the quarters that they had had just in the core business of selling cars. And it was just looking really, really, really abysmal. Yeah, Cybertruck was rough. The Roadster 2 was delayed.

31:30Model 3's not selling well. None of the cars had been re-eating. And then there was all this political stuff where were conservatives going to pick up and start buying Teslas? Well, no, you think about it. He lost the left, and then he was losing the right by fighting with Trump. So who's the base? Yeah, yeah, yeah. It was like losses on both sides. But I think that the car buyer has a shorter memory for these things and will return to— When you're leasing cars for$2.99, less than an Equinox. And it does perform well on all the benchmarks, essentially. It is great value. It's great value. Compared to any other car in that price range.

32:13Yep, totally. And so the bear case here for Elon bringing it back and returning home is that he has multiple homes. SpaceX, Neuralink, Boring Company, X, XAI. It can't be easy to juggle everything, but at least he's been able to pull back time spent in D.C. Or not Neuralink? No, I said Neuralink. Oh, you said it. Yeah. And it seems like there are new plans in the works to consolidate house Elon down to fewer entities so it's easy to manage. Overall, his current Tesla plan outlined in this post is a lot clearer to me than the Tesla secret plan V4 that felt written by a committee or even worse than LLM.

32:48Even if you're seeing signs of being stretched too thin, it's very hard to root against someone who's already built such a large automotive company in America and is taking a serious run at a self-driving network in a humanoid robotics play. There's lots of froth, but I like Elon being home at Tesla and focused on making science fiction reality. So that's my take on Elon's latest pay package. Anyway, whatever they're building, they got to do it on Graphite. Graphite.dev, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. We didn't really talk about the Sam Altman Tucker Carlson episode that went out.

33:30It was a crazy news day, of course, with the Kirk assassination. But Chrisman went viral, who's a friend of the show. He said, after spending seven minutes very politely accusing his guest of murder, Tucker just resumes a normal conversation to round out the interview. Incredible. Best to ever do it. And so it goes straight from suspicious death to thoughts on Elon Musk? What jobs will be lost to AI? What are the downsides of AI? Is AI a religion? The dangers of deep fakes. I love it. Yeah. Do you think this was not a win for Sam? Kind of went into. I think every PR person thinks it's an own goal or it's like a mistake to do.

34:09I don't know. It was so bold. It was so unexpected. Remember, Tucker had Suchier's parents. No, you know that that's what's going to happen. So you're going into the arena. You're going into the lion's den. You're debating the most critical critic possible. And there's something brave about that. I don't know. You're not afraid. Yeah, yeah. I took it as you're able to overcome your fear as bullish. I think Luke Metro had this post where he was saying that his fear of the going direct movement was that was that CEOs would never do a hard interview ever again because there was always someone who is willing to do an easy interview with you.

34:51There was always, oh, just do your own podcast. OpenAI has their own podcast. Sam Altman's brother has a podcast. He's friends with tons of tons of people. He can come to us and we'll talk to him just about AI and tech and the core business and stuff and have like a business conversation. To go to Tucker is a very, very aggressive, bold move. And I think it's good leadership. I don't know. I think it's like it's hopefully closing the book on that very sad and tragic saga. But who knows? I haven't really dug into like the actual response. Were people in Tucker's community swayed? That's the real metric.

35:29probably because you don't want to come out of that and be like, Tyler, wow, I have more. You're going to middle America. You're going to interview thousands of Tucker Carlson, super fans and ask them, did he do it? Just ask them. Did he do it? No. No, no, no. Explain to them this. Explain to them the full history of open AI, all the connections and see if they're bullish on the stock. That's what I want to know. Do they think 500 billion is undervalued or overvalued in middle America. That's what I care about. So we have our chart of the open AI and all the connections. Tyler, you were late here last night.

36:11You didn't leave last night. You were here all night. Kind of an all-nighter. Yeah. I was basically, the goal was try to, you know, I want to explain kind of in an easy, simple way, like how to think about open AI broadly. Oh yeah. How to think about people who are influencing it. I think you nailed it. in as clear of a way as possible. I think you nailed it. So who do we have here at the center? Yeah, so I think it's obviously best to start here. Yes. This is right here in the middle. OpenAI. Okay, so obviously, Sam Altman, OpenAI, this is the core. Yes. And so every line of string here is a real connection, correct?

36:46Yeah. Okay, take us through some of it. Yeah, so we can see over here, these are some of the main deals. Yes. So we have Larry Ellison, Oracle. Yes. We have Microsoft. $300 billion plan to build data centers. And then Broadcom is obviously doing chips, custom chips for OpenAI now. SoftBank is bankrolling some of it with Stargate. Microsoft has the classic deal. What's the connection between Johnny Ive and Growing Daniel? Oh, yeah. Okay, so this is actually a double. So this is most people on Twitter, they see this picture, they think Growing Daniel. But, I mean, that's. It's actually Steve Jobs.

37:25That's Steve Jobs. It's Steve Jobs. So it's really a double. Oh, OK. So you see, this is pretty good. I want to talk about this. So we have Marc Andreessen. Yes. You know, famous kind of interplay between growing Daniel and Marc Andreessen. OK, yes. But also with Steve Jobs and Johnny Ive. Steve Jobs and Johnny Ive. OK, got it. And then you're saying this goes deeper than.. This goes deeper. It goes really deep here. How deep does it go? Let's find out. OK, so then let's just keep moving. So let's go to some investors. Please. So we can go from OPI. Let's see Josh Kushner. Josh Kushner, of course.

37:58Okay, but who's Josh Kushner's brother? Jared Kushner. Where's Jared? First Trump administration. Okay, yes. Okay, now we're in politics. Let's go down PT. Okay. You can see a lot of strings coming out of here. This is a core player. Core player. So, I mean, there's so many directions we can go here, but let's go to Delian, right? Yes. Delian. Where does Delian work? Founders Fund. Who is? Yes. You know, it's obvious. Okay. Yes, it's obvious. You're putting it together. Okay. Delian's first company. Yes, Varda. No. Oh, first company. His first company, YC Company. Oh, yeah, it was a YC Company.

38:29That's right. What was another YC Company? Alexander Wang. Oh, wait. Scale was a YC Company. Okay. Where does Alex work now? Where does he work? Mark Zuckerberg. Oh, Mark Zuckerberg. Okay. Nat Friedman. They work together. MSL. Okay. It's so clear now. Who did Nat Friedman sponsor? Who? Luke Ferretor. Oh, Ferretor. And where does he work? Doge. Doge. Elon. It's all connected. And who did Elon start a company with? PT. It goes back. There we go. Okay. Okay. So we're piecing it together. So Wilmanitis. How is Wilmanitis connected? This is just a joke. He's not actually involved, right? Well, Wilmanitis was a what kind of fellow?

39:05Oh, a teal fellow. A teal fellow. That's right. This does get pretty deep. So, and then we can go from Wilmanitis. He's obviously, you know, very popular on X. He is. This is another kind of, well, at some point, there were too many strings. I didn't want to cover too much. Yes. But there's a lot of connections here. obviously we can go down to Dwarkech. Okay. So he's kind of another superpower. He's a super connector. He's interviewed most of these folks. So let's go from Dwarkech to Gwern, right? Okay, yes. Gwern, obviously connected, scaling hypothesis, very influential on Ilya. Yes. This obviously goes back to OpenAI.

39:40Yes. It also goes to Carpathie. Yes, one of the co-founders of OpenAI. Yes, also Tesla autopilot. Yes. Back to Elon. Okay. Okay, let's go to this side of the board here. Yes, who we got? So let's... How is Scarlett Johansson involved? Yeah, so there's a couple ways she's involved. So obviously the first one, she was in her, right? Yes, yes. Famously, when OpenAI voice mode was coming out, they wanted to do her voice because that's like the movie, and then she was maybe suing. And was it Sam or the official OpenAI account that just tweeted her? Yes, I believe it was Sam. Yes, and it was unclear if the licensing was proper at that point.

40:24So there was maybe some backlash, maybe some legal letters traded. Who else was in her? Who else was in her? Joaquin Phoenix. Yes, Joaquin Phoenix. And he was also the Joker. This is not related to other things. We're getting Joker-fied. Yeah. It won't happen. So let's go back to ScarJo. In a fast takeoff, we all become Joker-fied. So she was in a movie with another actor. Yes, who? Joseph Gordon-Levitt. What movie? Don John, I believe it's called. That's right. Don John. Yes. So who is Joseph Gordon-Levitt's wife? This goes deeper than I thought. Whoa, this does go deeper than I thought. Joseph Gordon-Levitt's wife was previously a board member of OpenAI during the ousting.

41:01That's true. That took Sam Altman out. Wow. It's all connected. How can you not see this? It's so clear now. Okay. So now we're in kind of this ousting. Yes. Effective altruist. Safety. Yes. This is all connected here. Yes. Helen Toner. It gets really jammed up. Adam D 'Angelo. Yeah. Because we can go from here to EA and then to Eliezer. Eliezer, of course. Right. And here's, you know. Yes. Eliezer's first organization, MIRI, Machine Intelligence Research Institute. Yes. That was, I believe, early on funded by PT. Okay. How is Eliezer connected to Grimes? Is that Grimes on there? Well, okay. Do you not remember this famous picture between Grimes, Eliezer, and Sam?

41:42Oh, I do remember this. Yes. Good sound cue. I completely forgot that image. That is lore. Yeah, okay. So then Eliezer, very influential on this famous book about safety, Superintelligence by Nick Bostrom. Okay, Bostrom, yes. He is kind of this online philosopher, almost part of the blogosphere a little bit. Sure, sure. And connected to Scott Alexander, right, who was on Dwarf Kesh's podcast. Yeah, he was AI 2027, part of that crew, right? Or Lucy, really? No, he was a commentator on it. Oh, I think he was actually an advisor on it. I don't know if he was a co-author. He was involved. He was involved.

42:21He's implicated. It's never been clearer that there's a connection. So let's go back to where Cash from here, he was on the podcast. Yes, yes. How did Will catch a red string? Just from being a poster on X? Oh, he also works at OpenX. He was just through, oh, Will Depew. Oh, Will Depew. Will Depew. Oh, well, he's at XAI. No, no, he's on X. He's at OpenAI, right? Yes, yes. Okay. We're getting things mixed up here. Got it. Okay. But, okay. We're at Dwarkesh. Yes, at Dwarkesh. Back to Dwarkesh. Yes. An early guest of Dwarkesh was Tyler Cowen. Okay. Now, Tyler Cowen runs the Progress Studies Institute.

43:00Yes, that's correct. I think that's what it's called. Okay. With Patrick Collison. Whoa. This does go deep. Now, Patrick Collison, famous for being Irish. Yeah. Oh, okay. And we see here this is linked with a lot of, you know. Yes. Who else is Irish? Irish cabal, some kind of. Interesting. So we see Will O 'Brien. The Hibernian conspiracy. Yeah, Will O 'Brien. We see Will O 'Brien. We see Jordy John. Yeah, we're both Irish. We'll see Joe Lonsdale, I believe. Oh, he's Irish? No way. I believe that's true. I'm going to have to fact check that. I have no idea. I think that's true. But, I mean, the main connection here is Ireland to the Illuminati.

43:36The Illuminati are Irish? Why are they Irish? Well, it's just, you know, general conspiracy. see. And then here we get a little off the rails. You see George Soros and Fauci connected with the Illuminati as well. Okay. So you're saying we're one degree away. Yeah. We're not beating the allegations. But let's stick with Illuminati. Let's go up to Brian Johnson, right? Okay. How is he connected to all this? Well, there's some sense rumors about, you know, blood of children, maybe transfusion or something. But really, Brian Johnson is part of this kind of health movement. Yes. So who else is part of that?

44:09Obviously, Andrew Huberman. Yes, Andrew Huberman, of course. And so he's a health influencer. And now you're kind of in the category of just experts broadly. Yeah. So that takes you over to Joe Rogan. Yes, obviously, because he's a health influencer, but really he's a podcaster at heart. Yes. So now we're in this kind of mannestore. We see Theo Vaughn, who's obviously connected in many ways to Alexander Wang, to Zuck, to Sam. Theo Vaughn's really, I mean, he could potentially be at the center of this whole thing. Yeah, for sure. Potentially. I mean, really, in the past year, he's become kind of a tech power player.

44:42That's the point of all this red string. You have to see who has the most connections. Moss has got two. TBPN has like 12. And I see raw milk up in the left corner. You're kind of dancing around that. Are you afraid to talk about raw milk? How does raw milk? Okay, so we're at Joe Rogan. Let's go back to Andrew Humerun. Okay. Another kind of person in the health area, Justin Mares. Yes. I believe he is a fan of raw milk. He's a fan of raw milk. He certainly is. But then we see the second string here. Let's just follow this down. How does raw milk? Oh, Jordy implicated. Not beating the raw milk allegations.

45:17Okay. But then Cuban was taking his time. Cuban was anti. Yeah. But he's not on the. He didn't make the board. He didn't make the board because he's not implicated. He would have made the board. I think the next iteration is 3D board. What's going on with Larry up there in the center? Yeah. So Larry is. Who is he connected to? Larry's implicated in a lot of ways. So let's just stay here for a second. So we're in this kind of podcasting space. Obviously, David Senra. Yes. James Dyson. James Dyson. Yeah. Latest episode. Look, I mean, this is a very pure, that's the most pure placement on the board, right?

45:48Yes. Hanging out on his own, just connected to Senra. He's connected to, yeah. Oh, yeah. Yeah. James Dyson's just in his lane, just hanging out in Founders Podcast. Exactly. But let's go from Senra to Larry Ellison, right? There's been, I think. Three episodes. I think three episodes. Yeah. On Larry Ellison. Yeah. um lara elson is also clear ties in a lot of ways um to sam because of the you know the deal but also to elon right this famous he wrote a billion dollar check yeah yeah i think i recommend two billion but whatever you think you know yeah yeah um okay help fund the buyout i see rahul sunwalker on there exactly yeah so so rahul is is is kind of an interesting person here so he's connected to Growing Daniel, obviously, because they did this famous stunt.

46:34Yes. His company is in YC. Yes. Which, again, I think we covered YC before, but obviously Sam. Yeah, it all adds up now. Yeah. It's pretty obvious there. Delian, YC. YC is also, I mean, there's not a lot of strings here, but you can, you could, you know, pull a lot of strings just from YC. Now, take me through the chessboard. What's going on over there? Yeah, yeah. So, let's go, let's just start with PT again, right? So, famous chess player. It all comes back. Famous chess player. Yes. So obviously connection there. David Sachs also a chess player. Oh, yeah. They played chess together in that famous image.

47:04Yeah, there's that famous photo where Sachs is beating him. Yes. And then, yeah, we see from chess, we see two other famous chess players. Yes. Botez Sisters and Magnus Carlsen. Clearly related to this. They were, I believe they were speaking at the All-In Summit. Well, they were. I don't know about this most recent one, but in the past they certainly have. Puzzle pieces are starting to fit together. Yeah. So chess is a board game, right? What's another board game? Go. Lee Seedol. Yes. Okay. Whoa. The famous, let me go over here. Also cigarette smoker. Exactly, yeah. Lee Seedol go to smoke. During the famous DeepMind AlphaGo.

47:38That's happened, yeah. I remember. So then obviously Demis. So he's connected to Demis. Yep. DeepMind. Then we see the rest of Google here. We see Eric Schmidt, again, connected to PT. OK. And their famous debate. Yes. Is that Will Brown? Will Brown. How is he connected? So Will Brown is. I think you're letting him off easy here. Will Brown is teal-backed. OK, yes. Prime intellect. Prime intellect, of course. He's also a frequent engager. Reply guy. Reply guy. That'll get you on the board these days. If you're a reply guy into Rune, if you're going back and forth with Rune, you're on the board.

48:11So if you see Rune, there's also a lot of strings coming out of here. He has his hands in a lot of places. Yes. Fingers in a lot of pies. Exactly. Donald Boat. Yeah, how's Donald Boat linked to it? Yeah, so Donald Boat was also one that I could have had a lot more strings coming from. but the main one was i mean donald boat's real recent rise to fame yes was asking um sam for a gpu yes which was sent and he also just offered sam some rice i think he's oh really he looked like he got about 600 pounds of rice on some uh our online arbitrage and he was offering it out symbolism will be his downfall potentially let's figure it out um and then obviously will was also heavily indicated in these kind of Donald Boat sagas.

48:55Yes. I believe he sent him a CPU case or something. I would say those are mostly the broad strokes. I think you're close to figuring it out. I think it's pretty clear now. I think I understand. Yeah. So that's kind of like, if you want to think about opening AI in an easy way, who's kind of making a lot of the decisions, how are they being influenced if they're not making the direct decisions? This is a simple way to understand. I would say this is kind of a good picture. It's a good place to start. Who the players are. Moss? You didn't cover Moss? But he's sitting sort of quietly, confident, not totally in his own lane, but not, you know.

49:28So he's the connection between Sam Altman and Donald Trump on Stargate because he was announcing the Stargate project. Yeah, and obviously Trump is linked to a lot of these people like, you know, Sergey. David Sachs. Just because of the recent dinner. Oh, yes. They had dinner together. That merits some string. Here's the chic of UAE. Okay. I think there is some Stargate development going on. UAE. Yeah, they're working on it. Well, thank you for taking us through it at all. It's pretty crystal clear to me, Jordy. What do you think? I mean, yeah, I mean, it's a surprise it took people this long to kind of crack the code.

50:04Yep, yep. It's all been out there in the open. Yeah. Symbolism will be their downfall. Pretty easy to figure out. Well, thank you for the hard work, Taylor. Well done, well done. All it took was one intern and an all-nighter. Yes. And I guess a gap semester. We got to load up all this into Julius and analyze it, find more connections. What analysis do you want to run? Chat with your data and get expert level insights in seconds. We saw Raul up on the board. They're loved by over 2 million users and trusted by individuals at Princeton, BCG, and Zapier. Let's continue. Rune, also on the board, was on the timeline talking about making the argument that we are currently in the fast takeoff scenario.

50:48Very fun post. He says, right now is the time where the takeoff looks the most rapid to insiders. We don't program anymore. We just yell at codex agents, but may look slow to everyone else as the general chatbot medium saturates. And so Prins, there's a bunch of reactions to this. We'll go through it. Prins says, fast takeoff does not require automation of AI research. The loop starts when AI researchers' productivity starts accelerating due to availability of better and better AI models, including coding agents. Super Dario says that the, this is a little bit of a confusing post, but he says, Dario was wrong about 90 % of coding being done by AI boys in shambles.

51:31So if you were thinking that 90 % of coding would not be done by AI, you are in shambles because 90 % of coding is being done by AI. And we have Alex, who's running Codex at OpenAI, coming on later today. We will press him on. Are we just writing 10 times as much code? Or are we writing the same amount of code, just 90 % more efficiently? What's actually going on? Are we replacing engineers yet? Or are we just now that engineers can do 10 times as much? Is it just more leverage? More of them. Yes. And what abstraction layer are we operating? And then also just how much is just raw coding a gate to better and better super intelligence?

52:17Is that the critical path? Is it just an engineering problem where you need to write a lot of code? Or is there something where we need a vast amount of – I was listening to Mark Zuckerberg on Dwarkesh this morning. and he was saying that he's not a believer in fast takeoff because he thinks at a certain point, it just takes time to marshal the resources to build all the chips and a certain amount of time to build all the infrastructure to generate all the electricity and that that will act as a natural slowing function. But there's some good economic news. There's a variety of different data points that we've been looking at to track how fast of a takeoff are we in?

52:55Are we going exponential or are we on our sigmoid grind set and will we be plateauing at some point? That'd be a good name for somebody's alt account. Sigmoid grind set. I mean, progress historically has been a series of sigmoid functions. And overall, progress looks smooth when you zoom out. But if you go inside of a chip company, inside of Intel, like the creators of Moore's Law, Moore's Law looks very smooth. But if you actually interview the people inside of Intel, They'll tell you that there were tons of little gains that then plateaued, and they added all those up to get incremental gains.

53:33And when you zoom out, you see the density of the transistors and the cost per performance increasing exponentially. But it was a lot of hard-fought wins along the way. And we're probably seeing something similar in AI. Well, if you need a generative media platform, if you're a developer, go to Fall. The world's best generative image, video, and audio models all in one place, develop, and fine-tune models with serverless GPUs and on-demand clusters. Look at those logos, John. Adobe, Shopify, Canva, Cora, Perplexity. Love it. GenSpark, and many more. Dean Ball is replying to Rune. He says, if this mirrors anything like the experience of other Frontier Lab employees, and anecdotally it does, it would suggest that Dario's much-mocked prediction about AI writing 90 % of code was indeed correct, at least for those among whom AI diffusion is happening quickest.

54:25And Mark Zuckerberg's been on the record saying that he thinks that Meta will write 90 % of code with AI as well. And Coinbase CEO Brian Armstrong has also said that the amount of code that's being written by AI is growing exponentially and is probably on the path to being 90 % AI written. Certainly haven't laid off 90 % of the engineers, and so there's some interesting dynamic going on there. But we'll continue to monitor it. on on jane at at uh andreessen's been on tear he's posting yeah he's been posting it up uh he says ai models with consumption consumption usage-based pricing are decimating seat-based pricing in ways most teams don't understand yet sas apocalypse in full swing he had some good supporting charts on that yesterday showing a bunch of you know seat-based sas companies down and to the right.

55:18Yep. Yeah, very tricky. We've seen some of them reposition. I mean, Salesforce, I think like over a year ago, Mark Benioff was saying like, we're going to outcomes-based pricing on AI agents. We're going to charge per resolution, per value. But it feels like some of these models are really sticky and a lot of the clients are like, we like our current setup or we're going to go somewhere else. and it certainly allowed other people to kind of eat off their plate. Outcomes-based pricing when you're providing software to manage a process is difficult because if somebody's selling, for example, like coding agents and the agent can do something and there's a certain economic value to that, somebody's saying, well, if I had an engineer do this, it would have taken them this much time.

56:07It has like a clear value. but if somebody's using salesforce to close a 20 million dollar deal are they going to want to pay salesforce some percentage of that just because they had that when it could have been you know one sales rep just like grinding to get that deal done yeah uh yeah a lot lots of lots of questions for the public company sas ceos um a lot of them will have to set up new narratives for how their pricing works going forward. They should be able to bring their products in line with consumption-based models. But I wonder how difficult it will be for them to actually convince their customers to switch over to something that's more comfortable.

56:50Yeah, which the alternative is, I mean. You know, Palantir specifically has said, you know, we're doing value based pricing, right? We're going to, we're going to, you're going to get this much value from what we're doing for you. Yep. And we're going to charge you this much and you can run the calculus. Prices are high. So prices are high, but the example, I think we were covering it yesterday, said somebody is willing to pay$10 million to do something that's going to make or save them$100 million. That's a trade that, you know, people will take. It does seem, it does seem highly aligned when it can work.

57:24And so, I mean, it, it, it makes sense that so many of these companies that have the new business model are kind of ramping so quickly. And that's certainly the result of those crazy revenue ramps that we've been seeing. Anyway, get your brand mentioned in ChatGPT with Profound. Reach millions of consumers who are using AI to discover new products and brands. You can get a demo. David Senra has some sage advice from David Ogilvy. Investor in Profound. Yes. He rarely invests. Rarely. David Senra says, quoting David Ogilvy, I admire people who work with gusto. If you don't enjoy what you are doing, I beg you to find another job.

58:05Remember the Scottish proverb, be happy while you're living for you're a long time dead. It's a good message. Find something that you love doing and stick with it. Work with gusto. And he's not talking. What's going on here with world labs? Yeah, we can pull this video up. So this is one of the latest persistent generative 3D world. So World Labs is Fei-Fei Li's project. She's hiring generative AI model optimization engineers, biz ops folks, research engineers, lots of different folks. But they have a new world model that we can play the video for and see this. This is generated in real time. I think they upload a single image and then you can move around it.

58:53And all of this is, you like the music?

58:59It's pretty remarkable. Yeah, I mean, it does feel like a video game. It's very pleasant. I have a big question about how this productizes still. I was noodling on it months ago, or maybe it was a few weeks ago, when Google came out with their model. Tyler, do you remember that? Yeah, I mean, so this is, were you talking about, like, the video model that you can control around? Yep, yeah. This is the same thing, right? No, so this is different. So this is, you can actually go on, if you go to marble.worldlabs.ai. Yeah, I'm on here right now. It's pretty, you can play with it? Yeah, so if you move it around, it's basically, it's a Gaussian splat.

59:44Okay. So it's like, looks really good. And then if you keep going, you'll start to notice like the little artifacts and stuff. And then if you go far enough away from like kind of the center spot where the world was originally generated, you'll notice it's like you know completely like random like okay um yes this is a different um like completely different architecture than the than the google one yeah because that was basically just um a video model sure um but it would have the overlay of the keyboard and then when you would um interact with the keyboard it would overlay like the keystrokes onto the video and then it would continually be like rendering new frames.

1:00:21Yeah. So this is just like, this one, it generates the world that Gaussian splat once, and you can just like move around in it. But it's not like a video that's constantly being rendered new. Yeah, yeah, oh, it looks really weird once you leave. This is bizarre. Yeah, I remember Gaussian splats being popular like a year or two ago. And then they sort of fell out of favor. It felt like they were very compute intense just to generate a single one, and we never got to like the one click app that people can just download and then play with. There was no like game mechanic built around it. It was very much like tech preview mode.

1:00:56Yeah, so the big thing that came out with Gaussian Splats like, yeah, probably a year ago was that it became very easy to render your own. So like I made one, I think it's on my website, but it's basically like my room that I made. And if you can shoot on raw, you basically take a ton of pictures of your room from all different, or whatever it is, from a bunch of different angles, then you can like... Load those up. Yeah, you load those up. You basically train a model to figure out where the images were taken. Sure. And then from there, you can train a model from that, which you can do on like...

1:01:26It interpolates everything. Yeah, I trained it locally on my PC. Okay. So it's not that computer... Do you remember like roughly how long the training run took on the... So I have a 4090. It took like 45 minutes. Well, that's not that bad. Yeah, not bad at all. Yeah. So you imagine with some optimization, you could wind up, and if you're running it on server quality hardware, you could imagine an iPhone app where you take some pictures, upload it, and then it's like push notification in five minutes or 10 minutes. Yeah, for sure. Yeah. But this is the first kind of thing, at least I've seen, of where it's like totally like generated.

1:02:03Like, I mean, usually when you see a Gaussian splat, it's of like a real physical location that someone's taken like drone shots of or something. But like, I mean, these are very clearly like AI generated in some way. Like maybe they trained on video games or something like that. But yeah, usually when you see a Gaussian splat, it's of like a physical location. Yeah. Can you think of any like practical application for these? Everyone always says like, oh, you'll use it to train robots or robotic models or something. Like this is like synthetic data for training robotic systems. But I feel like there's a consumer app in here somewhere.

1:02:44whether it's like generating video game levels or something or like using this as some sort of pipeline for creation of some video game. I keep wondering, and this is a question we'll have to talk to the meta team about tomorrow. I keep wondering about the, like we're getting close to a point where someone should be able to make a game as easy as they make a meme and post it on Instagram. but there isn't really a place where that lives right now. We've talked to a couple of YC companies that are like, we will vibe code an app for you that you can put in the, in straight to test flight. But I feel like they're like, we're going to do a deep dive on Roblox at some point, but I feel like there's like something that's going to happen in like AI enabled Roblox world where you, you can, you can kind of like describe and be leveraging a lot of AI tools to describe a world blocks for ai will be roblox maybe they're definitely not asleep at the wheel here we got to get some of their updated numbers i think they were pulling um yeah brandon sent me the the the data we're going to do a whole deep dive on it uh it's we should probably it's in it's in the deck at some point but uh probably set up a ultra dome in roblox that people can watch the show to get the next generation you know hooked on so dextero has a post here that says steal a brain rot game has broken all-time high concurrent player records roblox topped out at 23.4 million while fortnite reached 542 000 are these similar products or the platform wide record i guess was august 23rd roblox hit a record of 47.4 million players simultaneously across the platform.

1:04:39Wow. That is a ton. Roblox by itself has more monthly concurrent users than all of Steam, Xbox, and PlayStation combined. And Jeremiah Johnson says, even the most chronically online people have no idea how staggeringly, unfathomably big Roblox is right now. Is Roblox a public company? Yes. Yeah, right? They got out. Are they ripping? They had a Hindenburg research. Whoa. Oh, Hindenburg got blown out then because Roblox is way up. It's a$94 billion company now. Isn't there, I think there's a VC that David Senra did an episode about, or at least met and kind of did a deep dive on, who invested very early and did very well.

1:05:29Fantastic. Do you have any more context to add? I mean, not about Roblox specifically, but I think broadly like for for gaming I think there's kind of like two ways to think about like how how we can have AI gaming so one is like the google kind of strategy of what they've been doing where it's like the frames are literally like rendered real time um where everything is like all the mechanics all the you know the person the the setting it's all completely like generated on the fly yep and then there's like this kind of strategy which is where like the world is generated but you're still going to have, like, the mechanics of the game are still going to be, like, written normally.

1:06:08So, like, if you play around with this, you'll notice, like, there's no mechanics at all. You can move around, but there's no interaction or anything. I assume they're working on, like... Inventory. Health bar. Yeah, or even just, like, you can't go through the wall. Yeah, yeah, yeah, yeah. So, yeah, I guess we'll see. um but yeah i i wonder it feels like it'll be really slow like i haven't even heard about it gabe does ikea has a roblox store that they pay employees a real wage to work oh yeah that's hilarious so you can go shop in in roblox you can buy stuff in the in the metaverse i wonder what kind of sales that drives her it's just 75 of roblox's revenue or something we did just printing for ikea it's probably roblox has got to be bigger than ikea what's ikea's market cap ikea market cap let's see oh it's wait not it's private interesting okay private company um anyway um i i i don't know their ikea's revenue i think is well beyond roblox way way beyond oh yeah how big is it or 2024 was 45 billion euros okay but you know it's not sas margins here like there is a marginal cost to making a table and shipping it um but yeah i mean it's a fantastic business order magnitude more yeah um i i do wonder um like it feels like even the basic llms like we've passed the turing test we have the ability to just like talk to an npc in a way that's more convincing than the pre-scripted chats that go into most video games and yet i still haven't at least heard like a successful story of just a game that broke through that's really leveraging like gpt4 level technology um so interesting and even and even the generated images like just being able to use a you know older mid-journey model to generate textures on the fly there's this game called no man's sky are you familiar with this probably not uh are you Are you familiar with No Man's Sky?

1:08:13I've never played it, but I know what it is. Yeah, so it's a space exploration game. You move around from planet to planet and all the planets are procedurally generated. So they're generated somewhat randomly, but then they are like baked down into a database. And so there's a whole bunch of different variables that are randomly different sliders for like the size of the creatures on this particular planet, the color of the plants on this planet. How much water? Yeah, exactly. And all of that's generated and then saved into the database. But it creates this kind of unlimited canvas to explore.

1:08:50You could do that in a much more aggressive way with generative AI. But I haven't seen anyone break through. I think it just takes time to actually implement this stuff. Yeah. I mean, that's kind of what Genie 3 is. I remembered what it was called. Yeah, Genie 3. But it's just so compute intensive. I mean, they still are like, it's very much a research preview that they can't run. because it's so expensive. Yeah, you would just think that there would be a, like if you go two versions back, you get a model that can actually be baked down to the point where it can run on an Xbox or run on a PC and can generate.

1:09:27All right, gamers. I got some breaking news for you. Give us the breaking news. TikTok's US business. Details are starting to emerge. One, the deadline for the ban has been pushed back. and details are coming out. The journal has some reporting. What's the Polymarket at? Let's pull that up. Polymarket is... Must be crashing, right? Yeah. So right now there is a 24 % chance that a sale will be announced by September 30th, 60 % chance that it'll be announced by the end of the year. But the journal is reporting that TikTok's US business may be controlled by a consortium including Oracle, Silver Lake, and guess the last one.

1:10:16Larry Ellison, personally. Andreessen Horowitz. Andreessen Horowitz. Under a framework, the U.S. and China are finalizing as talks shift into high gear. Let's go. The arrangement discussed by U.S. and Chinese negotiators in Madrid this week, remember we talked to Bill Bishop yesterday about the meeting in Madrid, would create a new U.S. entity to operate the app, with U.S. investors holding a roughly 80 % stake and Chinese shareholders owning the rest. The people said this new company would also have an American-dominated board with one member designated by the U.S. government. Existing users in...

1:10:48I wonder if Trump ends up with a stake for Uncle Sam. Existing users in the U.S. would be asked to shift to a new app which TikTok has built and is testing. People familiar with the matter said. TikTok engineers will recreate a set of content recommendation algorithms for the app. using technology license from TikTok's parent, ByteDance. Okay. Ify. U.S. software giant Oracle, a longtime TikTok partner, would handle user data at its facilities in Texas. Both sides are still working out the final details of what the proposed deal and terms could change. Negotiations over TikTok came as both Washington and Beijing lay the groundwork for potential meeting between President Trump and Xi, with Beijing pushing for a Trump visit to China.

1:11:34That would be crazy. They said, thank you for... They want that meeting to happen somewhere where Trump can't fly. The Alaska shot. TikTok plan to comply with U.S. law. Tech industry executive argue its algorithms must be created and maintained by an American engineering team insulated from Chinese influence. Beyond the financial terms, deciding how to handle TikTok's algorithm has been a tricky part of the deal because it is seen as arguably the most lucrative part of the company. we've got a deal on TikTok I've reached a deal with China I'm going to speak with President Xi on Friday to confirm Trump said out of the White House Tuesday morning before leaving for a trip to the UK these are very big companies that want to buy it the framework of the arrangement came together anyways starting to just repeat itself here but this is big well if they want to manage their software development process in the new American TikTok they got to get on Linear Linear is a purpose built tool for planning and building products.

1:12:35Meet the system for modern software development, streamline issues, projects, and product roadmaps. I think if Andreessen is buying it, it's got to be all Mark and Ben show. The whole feed, 100 % of TikTok content should be the Mark and Ben show. They saw the Elon, the attention he gets on tax. A16Z, it's a media company that monetized through venture. Put the Andreessen content, put your thumb on the scale. Brent, Eric Torenberg, we want your thumb on the scale. Every other Give TPN a little bit of a snack in the new TikTok, but I want mostly Andreessen Horowitz content with a couple American Dynamism vibe reels, please.

1:13:12That will make America great again. This is the plan. One A60D portfolio company launch video, one Mark and Ben clip, one regular post. Just make the whole feed. Every single TikTok user, they'll love it. There won't be any backlash. I want the thumb on the scale. Truly, that would get me to use TikTok. I enjoy the Mark and Ben show. I enjoy Andreessen's content. It'd be a lot of fun. Anyway, our buddies over at public.com launched a new account on X that you can use to just tag, like, ask Grok, Grok, is this true? You can tag at gen underscore assets and you can tag it in any thread that mentions multiple tickers and this bot will automatically run a back test on their performance against the S &P 500.

1:13:58So I went to a post that was about Tesla and Google, and I asked, how are they doing as a pair? And it immediately told me that I massively missed out because over the last 10 years, those two stocks, the tech titans, as they're called here, had 10-year returns of 7 ,548%. Of course, you'd have to deal with a max drawdown of 56%. But if you held, you did very well with a 70 % annual return. and it gives you the link right there if you want to jump in to that. So if you're interested in investing, head over to public.com, investing for those that take it seriously. Very cool. Ashwin, who is, we've featured him on the show before.

1:14:41Ashwin says he's opening a coffee shop in Southern California. Here's how he's going to win. He's going to open at 5 a.m. He'll have baristas from New York so it doesn't take eight minutes to make your cortado. I don't know what a cortado is. That must be some sort of coffee beverage, But I like the idea of opening early. We were running into this earlier. We left the gym. And there's this cool new cafe on the corner that's making a big push into breakfast. They don't start serving breakfast until 9 a.m. And we were like, they didn't get the memo about the great lock-in. Like, you can't. How can you compete in the great breakfast wars?

1:15:15I've always thought it's funny. Have you tried to get a coffee early in Manhattan? No. It's terrible. It is? I thought Ashwin is saying that it's the best in New York. I don't know. I've had a rough. I mean, if you wake up at 530 in New York, good luck getting a coffee that's not something pre-made at Whole Foods or whatever. Bring a case of the Mataina podcast in a can. This post was funny. The Zoolanderification of Kyle Simani should be studied. He's looking great. He's looking great. He's looking diced. It's called a glow up. It's a glow up. It's fantastic. fantastic i've seen this happen to other people that make billions of dollars it tends to happen but it's not always and i love to see it he locked in he's clearly hitting the gym he's looking fantastic i think this last image is from his guest spot on our show yeah which i love right uh so i'm glad that we're uh we're helping people understand bro sold buccal fat to fund uh the the I think he's just working out.

1:16:17He's just getting in the gym. We'll have to get his routine. We got Alex Karp's workout routine. We'll have to get everyone's workout routine. Geiger Capital here, S &P 500, all-time high. NASDAQ, all-time high. Gold, all-time high. Bitcoin, all-time high. Home prices, all-time high. Time to cut rates. Time to cut rates. Cheers. Geiger Capital was on a tear. Another post. The Atlanta Fed is now projecting the Q3 GDP, will be 3.4%, a massive expansion. The U.S. economy is running hot. I mean, GDP growth is good. I wouldn't really call that running. I mean, it's running a little hot, but you don't necessarily want to pull that back if you're growing.

1:17:00We're in a fast takeoff scenario. All data is either real or fake. That's all I'll say. I know what's driving the higher GDP growth for sure. It's businesses spending less time on sales tax compliance. That's true. They got a numeral, numeralhq.com. They put their sales tax on autopilot. They're spending less than five minutes a month on sales tax compliance. And now they're growing their businesses. They're growing the U.S. economy. It's all crazy. America empire continues. The most extraordinary Fed meeting has yet to be kicked off, or yet has just kicked off. This is from CNN. Federal Reserve officials are convening Tuesday and Wednesday for a pivotal meeting under unprecedented circumstances on Wednesday.

1:17:42At the conclusion of their two-day policy meeting, central bankers are expected to announce their first interest rate cut since December to support America's slowing labor market with hopes that President Donald Trump's expensive tariffs might only have a limited impact on the future. So the Fed decision in September on Polymarket has$163 million of volume. Betting on it. currently pricing that's total volume but they're currently pricing uh 95 95 chance of a 25 uh basis point decrease so so yeah i mean the market's at all-time highs goals at gold's at all-time highs bitcoins at all times hot at all-time highs but uh drop growth during the summer was anemic says cnn business employers added an average of about 29 000 jobs in three months ending in August, according to Labor Department data, slightly higher than July's average, which is the weakest three-month pace since 2010 outside of the pandemic.

1:18:39There are now more unemployed people seeking work than there are job openings. New applications for jobless benefits in the week ending September 6th rose to the highest level in nearly four years. If the Fed cuts, maybe that will stimulate investment in business, new uh uh you know take on extra debt that works its way through the economy you wind up levering up and then something crazy martin scrawley shorting open door oh he is uh right is this happening going into a rate cut is crazy that is bold uh he said i shorted open today at nine dollars 36 cents this is the first trade i've made in the stock i will be doing diligence calls with former employees customers competitors and hopefully management too i will send invites to the calls or anonymous transcripts as appropriate so he's becoming a provocateur yeah i mean he's he's taking up the mantle of uh so many short sellers before him uh not won't be making a lot of friends but uh good luck to him uh uh we already covered tiktok uh you put this post in here in brazil an australian pilot oh you know why i break this down yes why did you put this in here Well, tell us the story and then explain why you thought it was worth talking about.

1:19:58In Brazil, an Australian pilot, Timothy James Clark, died in a plane crash with 180 kilograms of cocaine on board. Drugs packaged with SpaceX branding. Yes, why SpaceX branding? I have no idea. That's why. I just thought it was totally strange. It's a tech story. It's a really – You saw SpaceX and you were like, this is business news. Well, you know I like following the – You love the traffickers. You love going. puts me to sleep and i can't i can't when i'm trying to fall asleep i can't listen to i can't watch tech content yeah my brain's gonna start working sure sure sure i have to just follow the uh so the drugs were packaged with spacex branding they were recovered from the wreckage of a sling aircraft i just need to know the mindset of somebody who's like packaging uh making making drugs look like they're spacex branding thinking okay if the authorities find this yeah they're just going to be like oh this is this is something from spacex i don't need to dig in here at all yeah what were they thinking this is this doesn't look sketchy at all this is blatant trademark infringement you did not license this logo i wag my finger on this i do not like this um and a very odd photo of the culprit the australian pilot timothy james clark uh he has a Very odd photo with some scantily clad women.

1:21:20But dark story. Hope they prevent future drug trafficking. Since we do not endorse anything of the sort on this show. We do endorse any members of our audience trying to give tips on Nicolas Maduro. Yes, yes. And on that reward that we talked about. We've got to stop the narco trafficking. We also got to get on Finn.ai, the number one AI agent for customer service. Number one in performance benchmarks, number one in competitive bake-offs, number one ranking on G2. We also endorse chess, which was on the OpenAI board earlier. Joe Lee has a story here. My friend was a chess grandmaster at 14. By 15, government officials invited him to dinners, and he was giving talks to thousands.

1:22:07Every other type of nerd has gotten their pop culture recognition except researchers. We're lucky to live in an era where the weird nerd is celebrated. Chess GMs post their moves on Instagram, and it's cool. Not weird. Tech founders are household names. YouTubers get invited to the Met Gala. Poker players have TV shows. Gamers have become Twitch millionaires. Researchers get maybe two celebrity moments, winning their state fair at 10 and a Nobel at 70. Unless you're Demis. Demis got a Nobel. What? How old is Demis? He's not 70. Beyond that, the people working on the world's most important problems are invisible to everyone outside their niche.

1:22:44Demis, the founder of DeepMind and the Nobel Prize winner, is 49 years old. Wow. What a run. He's got another 50 years in front of him just researching. Easy. Hopefully he's on Xi's timeline, 150. 150. He could be 20 years out from a midlife crisis. He could be at 75. Does that mean – how old is Xi Jinping? Because if he's going to live to 150 – In his 70s. He's 72. He's 72, so in three years, he'll be at half-life expectancy. We should expect to see him picking up some sports cars, maybe crashing out going through a midlife crisis at 75, if he's going to live at 150. Maybe he decides, I don't want to be in the party.

1:23:31I want to do something else. I want to just be an entertainer or something. Maybe I want to be a musician. Maybe he turns into a DJ. Or he wants to create a social network like Trump. Maybe, maybe. be uh beyond that the people working on the most important problems are invisible to everyone outside their niche researchers are the final boss of nerd culture that we haven't figured out how to celebrate time to change that tyler what's your job what is he talking about what what she uh joe lee oh yes we we are we are celebrating we have we have figured out how we celebrated so hard we had researchers emailing us saying please take me off the medicine list i don't want the attention yes yes which we obliged and the traded cards we need to continue to celebrate the final bosses of nerd culture the researchers i think they are being celebrated and i feel like an sf if you see um like if you see ilia walking around oh yeah you're taking paparazzi photos of him uh people go to the the open ai that that cafe that johnny and sam were at yeah they're like oh i'm at the open ai cafe Give us an update on hype.tbpn.com.

1:24:38How's that going? Well, the video completely flopped. It did. Maybe the vibreel meta is off. Maybe something is tricky. It's very hard to draw attention to a URL in general. But if you're interested in participating in our little questionnaire, where are we in the AI hype cycle, you can go to hype.tbpn.com. And we have the full Gartner hype cycle there. Starts with the innovation trigger. We should screen share this. Goes to the peak of inflated expectations. Then, of course, the trough of disillusionment. And finally, the plateau of productivity. And Tyler has shared some of the results. We're still collecting results.

1:25:27and I believe that the consensus estimate of where we are in the AI hype cycle. People are putting their hands up. They're putting their hands up. I think they're, so the yellow dot is where Tyler is, just past the peak of inflated expectations. You're cooling off, is that correct? Yeah, I mean, I - Justify, what? So I don't know if I'm exactly there. I mean, it's very rough. I would say I'm near the peak. I don't know if I'm on the left or right side of the curve. Okay. I don't know if we're going down yet. Okay, what about you, Jordy? Where are you? You're not in the trough yet? It's hard to be in the trough, given all the good news.

1:26:01The market's at an all-time high. Products are shipping. Like, products are getting better. I'm, like, effectively where the yellow dot is. Yeah, just coming off. But that doesn't mean things aren't hyped. It's just people realize, like, wait, it's just, like, better search, you know? Yeah. We don't have, like. There was, I mean, if you were following the Eliezer-Udikowski crew, it was like in next year there's not going to be any jobs and you're going to be fighting for your life in a terminator scenario right and it's like if i'm not holding a gun mowing down robots i'm in the trough of disillusionment because that's where my expectations were yeah you can see the bulk is is basically where where my dot is yes past the peak um but you can see all the way on the left there is one dot yes did you see this all the way on the left did you see this i I don't know.

1:26:52Can we go full, full screen on this? Yes. So there is one person who says, when it comes to AI hype, we haven't even started yet. Okay. So I made a tweet about this. Yes. And then friend of the show, Jacob Rinamaki. Yes. That's him. That's him. He's all the way on the left. He's all the way on the left. Okay. His rationale was basically, you know. No technology trigger yet. Yes. We're at, you know, one ten thousandth of solar, you know, usage of the sun. Okay. We're not even close. I mean, like, we're not even close to what AI could be. Sure. So he's just mega AGI pilled. Oh, he is. We're not even close to being hype.

1:27:27We're not even close to being hype. Until we have the Dyson Sphere. The hype hasn't even started. I mean, how can we be going down if we don't even have a Dyson Sphere? That's a great take. I love Jacob. He's so much fun. I also like that there's someone on there. There's a few people that said, yeah, the hype cycle's done. I'm in the plateau of productivity. We went through it. They experienced the full. And I would love to know from those people, maybe we should have someone on the show or ask them to write out a little bit more of their rationale for where they are. Maybe that's in the V2. We could have put your point and then write a little free text response on why you said that.

1:28:00Because for that person, what is their hype cycle? Like, obviously, is the innovation trigger GPT-1? Is it GPT-3? Is it chat GPT? Like, the default assumption is probably innovation triggers chat GPT, then peak of an inflated expectation. No, Transformer. Yeah, maybe you go back further. It's neural nets. Sure, sure. Like GPTs are probably a little bit higher on the curve. Yeah, so maybe the trough of disillusionment was like year five of OpenAI as a nonprofit. And you're like, yeah, like they, okay, they did okay against Dota, but they haven't really created anything. I was so excited about this.

1:28:41My expectations were inflated a year or two ago, and now I'm really disappointed. But now ChatGPT comes out, slope of enlightenment, plateau of productivity is, hey, we're building in codex. We got cloud code. We got different products coming together. We're actually delivering value. Then they're in the plateau of productivity. I don't know. Yeah, I think in that scenario, it's like early, less wrong posts. Those were the peak of inflated expectation. That makes sense. And then you see actual people building. Well, we'll have to ask the folks who responded. And if you want to play along, go to hype.tbpn.com and vote for where you are in the AI hype cycle.

1:29:20One thing that's underhyped, customer relationship magic from Adio. Adio is the AI native CRM that builds, scales, and grows your company to the next level. You can get started for free. And we have our first guest of the show coming in to the TBPN Ultradone from the Restream waiting room. Whoa! Hey, look at this. How you doing? Let's go. You shocked. We're hitting the gong for all you guys. We're hitting the gong. What's the news, folks? What's the news? I'll bring the hardcore team out. We launched Rocks to GA. Fantastic. We crossed 25 million Rocks revenue agents running in the global 2000s.

1:30:00It's real. There's real ROI. And we're coming. So we're ready for business and we're ready to go. Amazing. Fantastic. Yeah, give us the update since the last time you were on. We came on February. We come back and launch. We've landed. Obviously, we still have Ram, Cognition, and some of the biggest AI companies. But we've landed the number one for Global 2000 in banking, construction. Also, Saudi Aramco. Whoa. We've built a sales team, which is amazing. They're now kind of going head to head with some of the big incumbents. So how are you positioning the product? Shouldn't you be dogfooding? Why do you have salespeople?

1:30:47Exactly. So we ran up until recently without a sales team. But now we've got to go for market share. So we have a small elite sales team who can go basically work with boards who are rolling out kind of rocks revenue agents after the coding agents. So every customer has rolled out coding agents over the last year. now they're rolling out revenue agents and it's a game of market share and delivering ROI and we're the fastest to it. What's the competitive dynamic like? I don't know a bunch of other companies in your guys' category broadly, meanwhile because of how quickly coding agents have got to market, I know I can name a bunch.

1:31:26The big question is seat-based versus consumption-based this is the debate that a lot of people are having on the timeline these days, the SaaSpocalypse or something How are you thinking about pricing as a durable advantage against incumbents? Absolutely. So our incumbents are obviously Salesforce. I saw the CARP episode. We're team CARP here. We focus on time to value. We deliver ROI in 90 days instead of nine months. We focus on pay for agent actions. So the agents do work, and customers only pay when the agents do work. It's not seed-based pricing. And the last is we do the Palantir style full service forward deployed model, right?

1:32:05So you don't need consultants on professional services firms to implement the software. We do the whole lot. So competitive dynamics is boards want ROI in a quarter or two so that they can change their hiring plans for next year. And we're just trying to basically pedal downhill and deliver ROI fast. And it's only possible because of the team. We can outship the market and build product that works. and now's the time. How big is the team today? I was about to ask. Yeah, so kind of obviously competing with a multi-thousand employee organization. So core applied AI is about 25 and kind of field plus sales is about 10 more.

1:32:47So about 35 people. Wow, that's a crazy balance. I like that. That's great. Yeah, you try engineering as much as possible. Yeah, and then what's the feedback cycle like from the sales folks that you have that are actually using the product. I feel like, are you hiring folks from product management into sales roles or vice versa? Or are there some product management skills that are uniquely useful? It's a great question. And one of our sales leaders is actually a pseudo PM. So we have no PMs, no product managers, no product managers. We have engineers who ship, and then we have salespeople who ship revenue.

1:33:22So the feedback cycle is once we land customers, We are pretty much embedded with these customers. We have field engineers, our forward-deployed engineers. They feed it back to all engineers. And our engineers, everybody here, they're all founders. They're customer-facing. And they take all that signal, and we try to ship in days. So we ship multiple times a day. We have major release every week. And that's kind of been the flywheel. What does the typical forward-deployed engineer process look like for your guys' product? I imagine it's not quite as complex as some of the build outs that Palantir is doing for different companies.

1:34:01Is it down to like you need to land an engineer at a company or a small team for a couple weeks, a week, a few days? What does that actually look like? Absolutely. So we try to do ROI 90 days, but the first three weeks is essentially set up. So we don't do any core engineering out in the field. So that's all done in San Francisco. though. What our four deployed engineers do is deeply understand the domain. So the number one global 2000 in the construction vertical is different than healthcare, is different than kind of the military. So we try to really understand the domain, how do they do customer acquisition and success, encode that into our agents.

1:34:41And then we actually build training content to reskill the whole org, right? So our core mission, if you remember, is to build agents that makes the best better and then re-skill the rest to give them a path to be kind of AI native. So that's kind of what our field is doing. So supercharging and then re-skilling. Well, I feel like the opportunity cost of having everyone on your team stand behind you for this interview is insane. We're going to hit the gong and let you get back to the important work that you do. But thank you so much for coming on TVPN today. Thanks for having us on and we're all huge fans.

1:35:16Thank you. Congrats on all the progress, guys. We'll talk to you soon. Cheers. Bye. Let's go to this post on the timeline from Jerry Howell, quoting the Internet Hippo. This screenshot looks aged. Have you seen this screenshot? It looks like it's been screenshotted multiple times. It has like lore to it. It's vintage. The Internet Hippo says, I stepped away from my computer for a moment. When I returned, it was building something. I did not ask it to do this. We're not ready for what's coming. Tyler, do you have any idea what this image is in reference to? I don't know. You've never seen that.

1:35:56Wow, Jordy, do you know? It's a screensaver, right? Do you know from what machine or what era? Not a Mac. No. I believe it's like a Windows 95 screensaver, maybe Windows. I just remember this from a computer lab. You remember a school would have classes. For sure. Imagine. For sure. That was one of the coolest. It's actually underrated to just think that there used to be a class called, for us, it was Computer Lap. Yeah. No, no, you learn how to type on the computer. You get your words per minute up. Anyway, how did you sleep last night? I had a terrible night. I put up the best night in a long time.

1:36:31You got me beat. I got like a 54. It was a rough night. But if you want to get AIDS sleep. Proof of work. Whoa, 99. Let's go. Play some sound effects for yourself. Go to AIDSleep.com. Get a pod five, five-year warranty, 30-year risk-free trial, free returns, free shipping. You know what I did yesterday? Yes. I reduced my caffeine intake by about 120 milligrams. Okay, that's good. I think I might have done something. Dialing it back. That's good. Well, we have Tony from Ingest joining the stream. He's in the Restream waiting room. We'll bring him in. Welcome to the stream. How you doing? Hey, thank you so much for having me.

1:37:05How's it going? It's going great. Kick us off with just a brief introduction on yourself and then give us the news. Cool, yeah. I'm Tony, one of the founders and CEO of Ingest.com. here just to talk about our Series A with Altimeter. We closed the 21 mil round with them. Congrats. Yeah, they're great people. Jammin's a really, really smart guy. And we also had the follow-on and the continued investment from Andreessen, Notable, and Affle, which is great. So yeah, fun times, you know? Love it. Jordy, you want to hit that gong? Let's talk about... I'll wait till the end. Okay, we'll hit the gong again.

1:37:41Let's talk about what the company does. What do you guys have to do? Yeah, yeah. Yeah, so we're an async execution platform, durable execution. We allow you to build workflows as step functions, make it really, really easy so that you abstract all of the infrastructure, essentially write a couple of different things like step run, step run, step run. We make sure they execute durably. You can write any code in those step functions. And a lot of people use this for AI and AI agents because it turns out that that's a really big thing as you've heard all the time. Yeah. How did you pick the level of abstraction that you're operating at?

1:38:17There's so many other, like there's companies that they just do the inference or they just do fine tuning or they, or they go in even higher level and they sell a complete SaaS solution with an agent wrapped in it. You're in this kind of in-between zone. How'd you land there? Do you see yourself moving up and or down the stack? It feels like such a flexible time right now in the market. yeah super flexible super flexible um maybe i'll talk about the origin stories and then i'll talk about um so i used to run engineering for a healthcare company and healthcare is like dude healthcare is so hard i would not recommend it really nice to help people but the engineering is crazy super workflow driven um and it would take months to build out what should take days you know when a patient does this if they don't do something in a certain amount of time x y and z and so those are those are workflows um so we initially built the sdk to make it really easy for your product engineer that might be full stack to build these workflows without mucking around with infrastructure so you don't want to have to create workers you don't have to create queues you don't want to terraform new new things on kafka you just want to write some code deploy it in your existing code base and have it work basically in a minute and that like that level of abstraction is really key because i think if everyone has access to the same models and if everyone has access to the same sort of compute what really matters now is how fast you can iterate on your product how fast you can actually build the experiences on top of those models and it turns out that like yeah you can build agents for other people but this particular layer that executes the ai that runs the workflows is really tough and you have to have it done right in order to iterate fast and so it's sort of this interesting sort of layer on top of the infrastructure that allows people to build and execute so um that was kind of always the plan yeah i think like one of of the things that we talked about is you know just infrastructureless infrastructure so that you can just write code deploy it and have it work which is um sort of a i think like a modern version of what terraform should be yeah what what categories of agents are you most bullish on today and over the next year we just yeah yeah because because you know again on the consumer agent side yeah felt like deep research was the first breakout killer app consumer agent Before that, everyone was excited about agents, but weren't necessarily daily, even weekly active users of a lot of them.

1:40:40Yeah. It's interesting. And you were just talking about with the person number four, you know, like the whole sales agents stuff. Right, right. Forward deployed stuff. And you see a lot of like support. I think like for the day-to-day things that everyone uses internally, you know, like it's kind of the stuff that you're probably using day-to-day. Claude, codex, that sort of stuff is like killer. You know, I use it for tests. it doesn't do distributed systems well enough for us right now but it's pretty it's pretty good um and then like in the future you know i think it's really interesting to see people craft experiences inside their own products um you know like decagon and whatnot have done that for support for some time which is super interesting and i'm really really curious where we go with context engineering and how people can build their own custom agents which are essentially workflows so i mean like day-to-day for sure the everyone the average ones that people are using but you're like you take nano banana you throw that in a workflow and you have this really sick agent experience on top of that it's like it's pretty good the world is changing pretty fast and it feels like everything will end up becoming this which is craziness uh how does how is soundcloud using this i think that's an interesting case study you have there i'd love to understand more about how a company that just does you know i think of it as like a music player a library of music uh like what like what's their strategy yeah yeah so uh music is tough uh it turns out that every industry has a lot of depth that you wouldn't realize when you go into it you know um everything from like making sure that there's no um i guess piracy making sure that there's no terrorism making sure that if people want to produce their albums and send it out to spotify itunes that entire music distribution is also kind of insane yeah um it requires a lot of workflows too so there's a lot of stuff that you have to do to operationalize music either distribution production content um and essentially they were trying to hack together this for for nine months um their cto at the time found us somehow and he had a lot of stuff up and running within a week and their time to production was like insane from like nine months to one week they ended up pushing us out to production relatively quickly and within like the first few months and uh i think i think they were live in maybe like the first six weeks.

1:42:49So they use this for a ton of their stuff. I think they just shipped the ability to create vinyls for your own music that you upload to Spotify. Wait, like physical vinyls? Like physical vinyls, yeah, yeah, yeah. I'm going to hit the gong for physical vinyls. Yeah, yeah. I want some vinyls. Great hit, Jordy. We've been looking to make vinyl. We want to press the show onto vinyl, of course. Yeah, we've had people reach out and complain that they want to listen to the show every day, but they only listen to audio. They've listened to the whole catalog, but they want to listen at the highest possible fidelity.

1:43:21Do you remember there was that Stripe thing? They made that physical magazine back in the day. I don't know if you actually. Like with paper? Yeah, it was like a physical magazine. It was like a Stripe magazine. This was from Stripe, the payments company. Wow. I know. I somehow missed this. I know Stripe Press. I know the books. They are print maxis. Yeah. I got to get a copy of that. That probably trades at a premium. Probably in the music. It should be in the Business Hall of Fame. the Business Museum. Yeah, that's fantastic. Well, we don't want to keep you too much longer. Thank you so much for hopping on the screen.

1:43:54Congrats on the round. Congrats on the round. We'll see you at the B. Say hello to all the folks in the altimeter. We'll talk to you soon. Cheers. Bye. Bye. Let me tell you about adquick.com. Out-of-home advertising made easy and measurable. Say goodbye to the headaches of out-of-home advertising. Only adquick combines technology, out-of-home expertise, and data to enable efficient, seamless ad buying across the globe. Emmett Shear has a post. The way that ChatGPT and Claude, et cetera, are all trying to be the portal for the entire internet reminds me a good deal of the first internet boom in Yahoo.

1:44:24I think the ultimate winners will wind up specialized just like last time. Shopping, entertainment, search. So he's thinking Amazon, Netflix, Google, and we'll have similar breakouts. It's interesting. I don't know that I completely like immediately agree with this. And this comment kind of puts him in the truth zone. To be fair, Google, Dwayne says, Control Alt Dwayne says, to be fair, Google did become the portal for the entire internet. Yahoo tried, but Google executed better. ChatGPT has already won for the moment. But whether they're the next Google or Yahoo of AI remains to be seen. They're doing a lot of things right.

1:44:58Yeah. I wonder how much consumer AI or like... Emmet fires back just to close it out. Google's not a YouTube Yahoo style portal. They're still way behind on shopping versus specialists, for example. They are searched first and foremost, and then also have extended into portfolio of adjacencies. It's not one interface to everything. Yeah. I don't know. I do wonder how the consumer AI apps will fragment. We're obviously going to talk to the meta folks about personalized superintelligence. And we saw this chart that went up just recently about how people actually use ChatGPT. And there are some pretty clear buckets.

1:45:37buckets. Like there are different people using different modes of ChatGPT from practical guidance, tutoring, teaching, seeking information was 21 % of searches, writing was 28%, multimedia, creating an image, that was 6%. And so you could imagine some of those getting peeled off and kind of becoming their own apps that are highly specialized. I tend to believe in aggregation. And like Ben Thompson has this point about Google. There was a moment when everyone was saying Google's cooked because of vertical search. So there's going to be Yelp for restaurants, the Google of restaurants. They will do it better than Google can.

1:46:24There will be travel sites. There will be mapping sites. There will be different sites for searching specific things that people will remember to go to. And Google really did change their entire model. They adapted. They improvised. They overcame. and Google was not ultimately damaged by vertical search. The question, I mean, you have to kind of ask yourself how bullish you are on agents, right? Because historically you would search something in Google, and then you'd go to a standalone, maybe you'd go to a website or an application. Yet there's a lot of websites and applications that I only maybe use once in my life, once a year.

1:47:00Do I actually want it? Do I, oftentimes I'd prefer to have an agent, for example, like Book Me a Hotel, right? These are classic examples. And so I think the paradigm is shifting. But without further ado, we have Tomer. Our next guest. Gasto coming in. From the Restream Waiting Room. Welcome, Tomer. How are you doing? Hey, guys. Thanks for having me. Thanks so much for hopping on. Great to see you. I'm such a huge fan because I went through YC in 2012. I believe you were in the 2013 batch, correct we were in winter 2012 yeah oh winter so is that i think winter at that point was after summer or maybe just before it's usually after summer it is yeah it usually does come after summer but but i remember that i was running i we the first the first payroll we ran for our yc company was just uh literally a five thousand dollar wire like from one account to the next and we were like, yeah, like, uh, everyone needs to figure out what to do with the taxes.

1:48:06And then we got on Gusto, which was then I believe Zen payroll, right? And, uh, and it, and it solved a ton of stuff and it was, it was amazing experience. And, uh, yeah, Jordi and I have been using Gusto for a long time since, um, big, uh, big launch today. Yeah. Take us through it. Thank you. Yeah. Yeah. Well, very, very excited to be here. I'm also a big fan of yours. In fact, no way. There we go. And sorry, I need to stop you for a second before we get into lunch. Where are you? The office behind you is teaming. Is this just your HQ? This is the same place that Josh was calling in. It's remarkable.

1:48:44Thanks. Yeah. We're in Pier 70 in San Francisco. We actually have our office here. Right next to the YC. Right in front of YC. Exactly. And today, actually, we have this showcase conference. We're doing it for the first time live. And it's really, really cool because you have hundreds of small businesses here, accountants who are all coming in and we're unveiling a bunch of new products to them. So, you know, we have showcases twice a year. We've been doing it for a while. It's kind of our big, big two product moments of the year, telling the world about all the new stuff you can do with Gusto.

1:49:16We started with payroll, expanded to benefits. We announced a couple of weeks ago about the guideline acquisition, pending approval, and all the different things that we've been doing over the years. Time tracking, HR, and tax credits, compliance, and other things. But yeah, today we're launching a bunch of really, really important functionality around cash flow. So, you know, when I speak with small businesses, often what I hear is that, you know, in the context of gusto is that the hardest thing where they get frustrated the most and feel like most anxious and kind of unhappy is when they can't make payroll for their employees.

1:49:53They don't have enough money in their bank account. Right. And when you dig into it and learn why and what's going on exactly, it's not that you don't make enough money as a business. The business is growing. They're doing really well. They're doing better than ever. So how can that be? Right. And it's all about cash flow. It's all about this tricky thing of like, when does money come in and when money goes out? So often, you know, think about it. I talk with a customer who's like, you know, they do like equipment for, you know, different, you know, conferences and things like that. Right. Like, you know, lighting and AV equipment and stuff like that.

1:50:26So guess what? They need to buy the equipment and buy the stuff before the event. And they get paid 30 to 60 days after the event. Right. So you have all this money going out, you have payroll going out, all these expenses going out, and you didn't get paid yet. And this is not a random one-off story. This happens all the freaking time for small businesses. And that's true for consultants that work before you get paid and many other types of businesses. So that's what this showcase is about. I'm really happy to share more about the features and functionalities. We're one of the few shows on earth that I actually want to understand.

1:50:58I'm super interested. Because we've run businesses our whole life. I've been in that position, you know, early on with my first business, which is like, oh, like, totally. It's like every two weeks, it's like, you know, public companies, they got the quarterly reporting, maybe, maybe it's going to switch every six months soon. But with a, with a, with a small business, you know, it's like every, every couple of weeks, it's like that. Yeah. So, I mean, where, where I want to start is like, this sounds like you're going to be lending money or bridging on a short-term basis with small businesses who are on the platform.

1:51:31Where does the money come from? I feel like when we usually hear about a fintech company or a business-to-business company starting to offer some sort of credit-like product, we usually hear, oh, well, they have a facility on the other side. They raised money for this. They're not maybe funding it off their balance sheet. Structurally, how does the money flow to the actual customer? Yeah, exactly. That's a really, really great question. So listen, there's a few kind of products here that come together in order to get people to have better cash flow so they're not late for their payroll. One is just like a set of functionalities about making faster payroll.

1:52:08Oh, sure. If you remember, often people need to run like the historical standard is four days. Yeah. So you run a payroll on Tuesday for people to get paid on Friday. Well, that's four days of cash that could be in your bank account. And that makes a really big difference because payroll is often the biggest expense of customers that, you know, small businesses have. So we are launching next day payroll. We're launching same day payroll. And we're also launching instant payroll. So these are three new products all about making sure you have enough money to bank on to the last moment. How do you actually do that?

1:52:40Because payroll companies forever have probably loved to say, like, people ask, can you make it instant? They say, it's not possible. Yeah, we've heard, like, oh, it's the ACH system. It's the government regulation. What's the unlock? Exactly. Okay, so for these set of products, it's all about actually our relationship with the bank. So, you know, the nice thing about being, on one side, we're a tech company. We're still a startup. We look at ourselves as a startup. But on the other side, we've been around for 12, 13 years now. Over 400 ,000 small businesses use Gafso every day. And, you know, I knew you'd do that.

1:53:14I love it. I love it. I want to gong too today. Oh, yeah. Always. Tell me when the gong is coming. But yeah, over 400 ,000 small businesses use our product every day. And with that scale, we now have the ability to work with each individual bank and kind of create partnerships. So in the end of the day, we're actually not taking credit risk on it. And it just uses some of the newer rails that banks have. I can't go into all the specifics, but it's not like us. We didn't need to raise any money for this. But it's a technological solution, which is exactly what you want. You don't want just like, oh, wow, now we have this great massive liability that's growing on the balance sheet forever.

1:53:54No, that's not happening. Now I have to worry about Gustafl. There is one thing, though, which is Payroll Bridge. So Payroll Bridge is a different product. And that's like not about making payroll faster. But that's like even if you still don't have the money in your bank account, even on payday, and you still need somebody to finance that payroll, we can come in and help. Or maybe even just a piece of it. Because it's totally possible that I have$100 ,000 in payroll going out, but I'm just 10K short and I don't want to hold up everyone. And so I can maybe take a smaller amount. Exactly. Yeah, so that happens more often than you would think.

1:54:30You know, again, it's that kind of cash coming in, cash going out. So we partnered with a company called Parafin as an embedded partner who's helping us with the loans here. So what happens here is, again, as a customer, you come in to run payroll. Let's say it's a Friday. You need to pay instant payroll on Friday. And then, oh, gosh, I'm missing 10K. I'm missing$20 ,000,$50 ,000, whatever that is. You know, with our partnership with Paraffin, we take a look at all of your history and we figure out, like, hey, can we help you kind of basically make that payroll on us? And then we're going to pay us back later on.

1:55:02So that's kind of that piece. And I think, you know, honestly, it's the biggest, biggest, biggest pain point in the world of payroll. It's funny. Like, we've been doing it for 12 years. you would think like, oh, you've already innovated in everything around payroll. There's nothing else coming in. That's not true. Even 12 years later, there's more - Sass is forever. Payroll stuff. We're also doing it on the international payments, by the way. That's another fast payroll, faster international contract store payments using stable coin as again, now technology that is available now was not available when we started.

1:55:31What percentage of the businesses on Gusto do you think are like, I'm sure you looked at the data, right? Yeah, was this a big COVID boom? There would have been opportunities to launch this feature. You could have launched this a while ago. I'm sure you guys wanted to wait until there was regulatory clarity on stable coins. But was this something that customers were asking for? Because it's one of those things that is helpful to both sides. The freelancer benefits because they're like, okay, when the person hit pay, I got paid. I don't have to wait for, you know, it ends up being more than four days.

1:56:05It can be longer than that. And the companies, too, historically would just be using these like, you know, random, either like dedicated international payroll platforms or, you know, special freelancer payments platforms, things like that. Yeah, so totally. I think, you know, the past couple of years, again, there's new technology. So we can we can use now and that's both on the banking side domestically as well as on the stable point side. That's why we were able to do it today. It takes time until you, you know, we work with these partners and often we're the biggest kind of partner they work with.

1:56:39And the reason is, you know, you may have a bunch of startups, technology, fintech companies that are using stable coin payments or using, you know, kind of instant payments through banks. But they usually have small amounts. Payroll is huge amounts, right? Tens of thousands of dollars, sometimes millions of dollars per transaction. So, you know, you really, a lot of these kind of protocols, fintech protocols, when they come out, they have like a maximum limit. So we need to kind of work with the providers and get there. And then, yeah, you're right about the compliance side of it. Like, you know, we work really, really closely to figure out that, you know, one of the things that we do is, you know, I think that was a little buzzword before, you know, AI came in and like took away all the buzzwords.

1:57:20But compliance, compliance technology, you know, so there's financial technology, fintech, compliance technology. and in that world, you know, that's kind of what we do every single day. We think about how do we take all the compliance worries and anxiety that small businesses have that they don't want IRS knocking on their doors and then help them, you know, kind of figure out all the compliance for them. There's a couple more features that I want to talk about. So one is just Gusto Money. It's a new platform. It's a new area for us. You know, when you think about Gusto, you always think about people, right?

1:57:50Pay your people, onboard your people, benefits, all that stuff. Gasto money is us entering and expanding into a new world. So that's the world that's just really focused on cash flow. So, you know, one of the things when we talk with customers is like, hey, you know, I get paid so much later. Wouldn't be like, can you help me get paid faster? Or, you know, when I pay my bills, like, oh, gosh, that takes, you know, that, you know, they need to be paid immediately. Can you delay that payment for me? And that's why we brought in invoice payments as well as like bill pay into Gasto. And we're going to keep investing in that area because it's all connected.

1:58:22You know, payroll is your largest expenses, but you have a bunch of other expenses in the business. So if you bring it all together to one platform, I think that we can really, really help people with that big anxiety around not having enough money to run payroll and kind of run their business. Does that change? Are a lot of these different products feel like they might have different underlying economics? Do you think that the structure of the economic model of Gusto is going to change over the next few years? or is this sort of like purely additive? Like how are you going to think about actually underwriting like the core business?

1:58:58I don't know what your plans are for the future, but it does feel like some of these are, you know, exchange fees or interest rate based or volume based. Others might be seat based. There's a variety of different financial models or economic models. Is any of that changing like the overall structure of where you think the business will be driving value in a few years? Yeah, yeah, that's a great question. I think for us, you know, I would say number one, we're really, really focusing on understanding what are the biggest pain points for small businesses for small businesses and figure out how to help.

1:59:31them. Now, sometimes it means, you know, business models for different products are different and we're okay with that. However, if you still zoom out at Gusto and you look at Gusto as a product and where we're going, by far, like really our majority of what we do is in terms of business model is software. Software as a service, you pay for employee, you pay for services, very high margins. This is not like a FinTech sort of, you know, play. And as you notice, we're using a lot of partners on the embedded side when we do things and you know that that are those type of kind of fintech businesses and we really obsess and focus on the user experience to make sure that for the customer they don't feel like they're being moved from one you know one system to another it's all behind the scenes all api so for you it's just one service it's gusto um but um you know that's kind of the core of what we do we're really really um focused and and are really strong on the user experience side for small businesses.

2:00:26That's great. What's the most exciting thing in tech to you outside of payroll? In technology at all. Let's see. You know, I know... And it's okay to say that your eyes never stray and that it's payroll. Compliance. Compliance. Payroll compliance. I love compliance. Listen, I love compliance every single day. That's what I'm living and breathing. um and you know but jokes aside like the truth is that when you talk with customers um this is not a cerebral sort of like analytical kind of um question and like you know world for them like for them it's it's missing friday night with their kids because of this sort of stuff right for them it's like you know the pain of you know the embarrassment of like needing to call every single employee telling them they got they can't make payroll on time and then the employee on the other side sometimes you know uh you know needing to manage all that what that means to your family so you know for me it's you know i love technology i live and breathe technology every day i love your show that always keeps me up to date what's going on uh in in the world of tech but you know i think the biggest connection for me and the thing that gets me going is that um customer pain point and see what software can do to help them yeah that's great fantastic amazing well hey i want you guys to come in here next time though by the way we have we have the again the showcase here.

2:01:45It would be really cool. In 6-1, when we do the next one, it would be cool to have a live performance here from Glasgow. How many people are there? Let's ring the gong for the total attendees. I think there's a run by hundreds of small businesses. All right. That's a lot of small businesses to show up to. You love it. Great to see you. Congratulations. Thanks so much for hopping on the show. We'll talk to you soon, Tomer. Have a good one. Great to see you. Bye. And let's move on. No one loves software for small businesses more than me. I don't think so. I don't think so. Don't go. Daniel Tanrero says, the core political divide today is pickleball Americans versus letterboxed Americans.

2:02:30I'm in neither camp. I don't really participate in either, but I would put myself in the letterboxed American camp. Tyler over there is a pickleball American, right? I guess, but I'm also, I would say I'm more of like a cinephile than a pickleball player. Okay. But I've never used Letterboxd. You've never used Letterboxd? It's just like Goodreads for movies. Yeah, it's Goodreads for movies. Okay. I feel like there's an opportunity to do a stunt there. I'm underrated that Tiny snapped this up. Oh, Tiny owns it? They bought a 60 % steak. Whoa, I had no idea. That's fascinating. Well, the Antichrist lecture is in full swing up in San Francisco.

2:03:06Did they find the guy yet? Did they find him? The hunt for the Antichrist continues. and it has spawned a ton of fun posts on the timeline. Pablo Paniche says, Peter Thiel talked about this at the Antichrist lecture. No, he didn't. Did you attend the lecture? No. Did you? No. People memeing. Grung Daniel says, Peter Thiel announced the name of the Antichrist, but I'm not telling. And people are having a lot of fun. Will Manias said that. It's crazy. I saw some of the protesters showed up disguised as Satan. I needed to put some of the protesters in the truth zone because one of the protesters said, Peter Thiel's latest acquisition, the U.S.

2:03:48government. If you know anything about PT, he doesn't buy whole companies. He believes in founder-led firms. He doesn't do private equity deals. So a little bit of a misunderstanding of the financial markets there, protester. Anyway, Tyler, are you bummed that you're missing out on the Antichrist lecture series? I was just going to say that protesters are pretty hilarious, some of the signs. I sent him in the... So the protesters, it's easy to get riled up by the images of the protesters and see the protesters as like this, like, oh, they're really over the top. They must really be aggressive.

2:04:22But for the most part, like I've seen there were protesters at just some random FAI conference I spoke at. I actually interviewed Alex Wang there, who we're talking to tomorrow. Tune in. And this is just like foundation for American innovation, like a very vanilla tech conference. Um, and, uh, so being able to protest things is important. Yeah. And so the protester was outside with a funny sign, but people were taking photos of each other. Like it was very, very friendly. I, I, when I first heard like, oh, we're being protested, I was like, oh, what, what does this mean? Is this like serious? Are they going to be really, I was going to be protesting outside of gusto and roll out instant payroll.

2:05:00For sure. For sure. Uh, anyway, we have our next guest in the restream waiting room. We have Andre from console. Welcome to the stream. How are you doing? Hey guys, doing good. How are you? Thanks so much for joining. Kick us off with a little bit of an introduction on yourself, what you do, what the announcement is. Sure, thank you for having me. Big fans here, actually. The whole team, I don't know if you can see them. Fantastic. Watching. I love it. Hey, turn around, guys. We can see you. There's a little bit of a time delay. Say hello. There's some delay, maybe. Yeah. Yeah, so, yeah, I mean, today we're here to kind of, you know, We're talking about a Series A announcement.

2:05:40We just... There they are. A little delayed. What you got? We just raised our Series A, DST Global, and Thrive Capital. How much? How much you raised? $23 million. There you go. There you go. Congratulations. All right. All right. What do you do? What do you do? What do you do? I got to know. You got to tell us. I'm not going to tell you. Stop burying the lead. Stop burying the lead. Tell us what you do. Yeah, yeah, yeah. So we automate IT support using AI agents that understand how your organization works. Fantastic. I know. We're fired up. We're not just a chat bot. You know, under the hood, we're basically plugged into all your systems, your apps, your devices.

2:06:23We understand how everything's laid out. So we know, you know, this is John. This is title department manager, the direct reports, the apps. He has access to the permissions in those apps. So like all this data is flowing into console from all these different systems. And then we let you build company policy and processes above that layer in natural language. So we call these things playbooks. But essentially, you're able to say, you know, when someone needs a password reset, first ask them this question, then run this action, then do this thing. So that's not converted to code. It's just natural language instructions, uploads the agent, just handles that directly in Slack.

2:06:57How modern is the wheelhouse customer for you? Because when I started my first company in 2012, we were late enough in the rollout of SaaS that we had no on-premise servers whatsoever. We had no local network. We just had Google Cloud apps and Gmail accounts. And so everything was managed in various dashboards. And it was an immense hassle to deprovision and provision. But there is a whole other world of companies that have much more complex IT policies. Where are you seeing the most value? What does the wheelhouse customer look like? Yeah, so we've started with the kind of more like modern tech companies to get started.

2:07:41So a lot of our customers, Ramp's actually one of our customers. There you go. You know, shout out to them. They're a great customer. Scalii, Flock Safety, Webflow. these real customers that are, you know, they're kind of like internally looking to, they're starting to think about what does it look like? What does IT look like in the age of AI? Right. And I think kind of what happened was we got it right in the seventies, the eighties, right? Where computers started being a thing. And, you know, IT was really this department that was coming in and showing people like, look, get off your typewriters.

2:08:16Here's how computers work here. So you can be more efficient. And they were like kind of optimizing business process. And And somewhere along the way, kind of like you mentioned, right, we started adding too many apps and too many, there are too many things that were being kind of managed, right? And, you know, IT kind of lost focus from being this kind of driver of innovation in the organization to just like stuck in, you know, they're like on this ticket treadmill, right? You come in, you've got 10 tickets, you log out, you've got 15. So you're kind of just like always just handling tickets.

2:08:46And so our kind of mission here is, you know, how do we take IT back to where it was, where, you know, take off the support load from them, handle those tickets for them using AI that, again, works the way they do, and then get them back to, you know, working with other teams and finding ways to, like, optimize their process, essentially. How does security fit into all this? I feel like the IT professional is very much the unsung hero, the enemy of the fishermen, the phishing scams. I mean, every place I've worked that's like professional or had a professional email, you get these phishing scams.

2:09:26And I feel like Sherlock Holmes, when I clock it and send it to the IT guy, hey, we're getting a phishing scam. And he's like, I set that up. I sent that to you. I was testing you. It was a test. But yeah, talk to me about security and how that fits in. it feels like I really don't want my IT professional to be able to be prompt injected. There needs to be a human in the loop still maybe, but maybe you know something else. Yeah. So, yeah, I guess you hit on a couple of things. So, but first we obviously spend a lot of time internally on security, making sure things work really well. This is my second company, first company we were doing, application security company, we're selling to US government, Fortune 100.

2:10:02So very familiar with, you know, building secure software. And so that's like the first thing I think, you know, next we think about, well, how do we actually kind of like give the LLM and the AI agent control over the process and like the order that things get done and like, kind of like, you know, this, this ability to flex with the user and work with the user, but then build more deterministic processes for things like API calls and actually, you know, the things that are more sensitive, right? And so our sensitive actions are actually quite locked down and you have the ability to get approvals on them, do MFA checks, honestly just lock them down.

2:10:39Some companies say, hey, we don't want to reset any passwords with console, like we just don't have to, right? So you can kind of adopt it as you're comfortable. And then lastly, yeah, if you don't support, let's say password resets your company, you decide that's too risky, we will loop in a human when those requests come through. So the way, you know, the way kind of console works high level is we sit in Slack, instead of having to go to, you know, like a JIRA ticket form or something, and you have to fill out, you know, the subject line and you're answering, you know, the laptop you have and you're like, should 19 already know what laptop I have?

2:11:12And you're kind of just going through this list and you hit it and you wait for some response. You go into console, you type in your issue, just like you're messaging a coworker. That request goes in 50 to 60 % of the time, we handle it entirely with the end user. And then when we can't, we actually loop a human in. So on the back end, we'll basically message them and say, hey, like, you know, John is having this issue. Here's everything you need to know about John. And they can jump in. And then the end user stays in Slack, right? They don't have to go to a separate portal or anything like that.

2:11:40Makes sense. What's the go-to market? Did you share any numbers related to the fundraiser outside of the headline number? Anything on the business side? Or we stay? I don't know that we're sharing too much yet. We're doing really well. So we are hiring. That's a sign that. Sign that the company's ripping. Give it away. The number that I got kind of off the record, but I'll share, is one. And that's contracts with RAMP. So that's really the headline KPI. It's more of a binary metric for success. But it's really the only metric of success that matters in Silicon Valley these days. So congrats on passing that binary test.

2:12:24Talk to me about the go-to-market. based on the size of customer, I'm super interested to know, is this like steak dinners, hand-to-hand combat, phone calls, you know, just direct response? Are you G2 hacking? What's going on? I mean, we are very flexible. We'll do what it takes, right? I think the standard process is actually, you know, we'll connect with some, you know, some folks, you know, someone on the IT team, usually some head of IT, director of IT, and we'll give them a demo. Usually they have some AI initiative internally where they're looking at, again, how do we make IT more efficient or kind of like rethink how we do this.

2:13:04We do a demo, usually they're sold. We do a POC right after that. So to get into a POC, we do like a non-prem, sorry, we do like a full product deploy into their systems. We'll do things like security review, legal, whatever. Then we'll go live about three weeks is usually how long we do like a POC for. So usually within the first week, what we like to say is you're actually in a good enough spot to go live company wide. Sure. We spend the rest two weeks just kind of like dialing it in. But we go live pretty, pretty quickly. I think scale actually went live in about three weeks during the POC company wide, because they were just seeing so much success with it.

2:13:45Are you positioning the product as like purely additive to the existing IT systems, plug in with everything? Are you seeing folks rip out legacy systems that they maybe develop themselves or legacy competitors? What is the tradeoff that the buyer is making at the moment? Yeah, shouldn't be any tradeoff. Our goal when we started the company was like, look, you're already managing a bunch of systems. If we come in and we say, well, if you want to use console, you want to use AI, you have to replace half your systems. Like that's just going to be a non-starter for really busy teams. And so what we do instead is we actually build really deep integrations into, you know, as much of your tool stack as possible.

2:14:25So we integrate very deeply into like Okta, Google, you know, Slack, you know, even your JIRA, right? Like we don't have to replace your ITSM. So your ticketing tool, you can actually keep that. You've spent a lot of time configuring it and building it out. You can actually build, you can basically put console in front of that and act as like your first line of defense. And the reason we're able to do that, you know, I know it's kind of like a nod situation. What do you mean? You're not replacing any core system is, you know, we're actually like doing work. So we don't need to replace your ticketing tool because we actually do 50, 60 % of your, you know, the requests that you're getting, again, without human involvement.

2:15:01And so that's often. Do you just try to generally charge for the value? obviously if you're doing 50 to 60 percent of the work that's saving that's at least like sounds like one to many people's time but how do you charge for it yeah so we we just charge per end user per month basically so the number of employees you've got in the company we kind of have like a a base price that we use and we discount as a company gets gets larger but we try to keep pricing pretty static i know there's this trend to you know look at usage-based pricing and so on And what we found is, you know, one that can be a little confusing, you know, credits and things are hard to like estimate, especially up front.

2:15:40And so what we found is like IT teams actually really like the predictability of knowing, OK, for this year, I'm going to pay this much money for this tool. And, you know, we kind of think of it as if you feel like you're getting, you know, you're getting more value out of it than you're paying, then, you know, great. That's good for you. Awesome. Well, thank you for driving efficiency in the enterprise. This is something we need. I don't want to share. I don't want to dox anyone, but we definitely have some interns running wild with access to systems that I would love to be able to just fire off an agent to go clean up, because you can't trust Tyler with the access to the root keys to this kingdom.

2:16:18Anything could happen. I want to be able to snap my fingers and deprovision him, turn off his ramp card before he buys more. Awesome. Well, you know where to find him. Yeah. Well, thank you so much for hopping on. We'll talk to you soon. Great to hang on. Congrats on the round. Bye. What else we got going on? We got a little break. We got Bezel. We got getbezel.com. Your Bezel concierge is available now to source you any watch on the planet. Seriously, any watch. Also, we got a new news from Donald Boat. He's been traded. Big trade deal. Trade deal. He's going over to Substack. He says he's, unless he's granted a retainer and a favorable contract from Nikita and Elon Musk, he will be taking his talents to Substack.

2:17:00uh he's been having some fun is he really six seven feet tall that's very tall if that's true yeah so i i didn't meet him in person but i knew i know someone who did meet him yeah he's apparently super tall and very very slim he says uh low a john size man yeah in order to get him on the team you got to give him a low mileage used corvette z06 c5 or c6 it's funny he's a funny funny guy. How much is a low mileage Zio? Probably $30 ,000 I would estimate. I mean Z06 is like the performance package. I believe we have our next guest in the Restream waiting room. Do we? Already? I believe Andrew Yang is joining.

2:17:42Oh, thank you so much. Welcome to the stream. How you doing? Welcome to the show. Good to meet you. Hey guys. Thanks for having me. Great to be here. Thanks so much for hopping on early. Great to meet you. wanted to have you on the show to get an update on you and everything going on. But then also, there's so many questions about AI, AGI, UBI, so many acronyms. I'm sure we'll go through all of them. But why don't you just give us a little bit of a reintroduction for those who might not be familiar, what you're up to these days? Sure thing. People remember me from my 2020 presidential campaign saying that AI was going to come, it was going to eat the jobs, and that we should shift to putting money into people's hands more quickly and broadly.

2:18:27Imagine making that case in Iowa in 2020 at truck stops before AI had actually arrived on the scene, but now here we are five years later, and I dare say a lot of the folks who are watching this are nodding their heads being like, oh yeah, we're going to do a number on a lot of rank and file entry-level workers. so over the last number of years I've been trying to figure out how to get money back into people's hands I was inspired by my friend Mark Cuban who started Cost Plus Drugs you guys had Mark on recently right we did a couple weeks ago he was great yeah so I thought can we cost plus anything else in American life and I realized that we're spending 170 billion or so on our cell phone services and data I was paying Verizon over a hundred dollars a month 140 a month to be precise and meanwhile Verizon and AT &T paid out 18 billion in shareholder dividends in the last 12 months.

2:19:20So you can kind of start doing some math and realize that Americans are paying twice as much for data as people in other countries. So today is the kickoff of Noble Mobile, where we try and put money back into Americans' hands and hopefully have people doomspring a little bit less, live a little better. We're having a big party in New York City on Thursday night. You guys are not in New York, right? And having you guys come as VIPs and hang out. We'll be there spiritually. But breakdown, breakdown, I know kind of the mechanics of Noble Mobile, but break it down for the audience. How does it actually work?

2:20:02Well, before I get into this, I'd actually love to put you guys on the spot and ask who you use for your wireless provider. So I said I was... I use Verizon. They've been fleecing me for years. Ryzen. Yeah. I'm seeing bad say. I actually have an iMessage here. Your bill is ready. This is not a joke. At 11.45 this came in. Your bill is ready. I just pay it and I don't think too much about it. But I'm sure I'm getting fleeced because I probably signed up on some like super cheap contract about a decade ago. And it's gone up every month. And, you know, it's monopoly. Dude, their recent ads are like.

2:20:39lock in the price in parentheses because we're just going to keep on raising it but the craziest thing is when you guys travel abroad when i was on verizon i went to europe with my family this summer and the travel pass cost 10 bucks a day uh to put yourself on the the foreign carriers towers you know how much verizon at &t are spending to put you on uh their partners towers about two dollars and fifty cents so why on earth would you charge us ten dollars or twelve dollars a day if It's going to cost you$250. Like the entire industry, when you start digging in, there's just a lot of profiteering.

2:21:14Yeah. So you guys pay people to use their phone less. Does that mean I start out with a, like, but break down the actual mechanics of how that works on a month to month basis? Sure thing. So check it out. I'll use myself. I know I use my phone for doom scrolling and social media a little bit too much. but at the same time, I consider myself an effective in-touch individual and kind of need my smartphone and don't want to count how much data I'm using. So the way Noble Mobile works is that you pay a$50 flat fee for unlimited data, and if you use it up the wazoo, then you paid 50, and that's the end of it.

2:21:54But if you use less than 20 gigs per month, then we'll give you cash back in the form of an end of month payment data dividend, we call it, of up to 20 bucks. And then any money we give you will then grow at a 5.5 % interest rate annualized so that your cell phone plan becomes like a mini savings vehicle without you having to think about it or do anything. And so this way - So you guys hold on, do you guys hold onto those funds as long as somebody is like a customer? Is that how that works? or like, or you can take the money on anytime you want. It's your money. But if you leave it with us, we'll grow it at a higher rate than you're going to get someplace else risk-free because we wanted it to be a set it and forget it savings plan.

2:22:41One of our first investors was a guy named Scott Galloway, Prop G, who you guys probably know. Has he been on the show? Not yet. Not yet. He's got to come on sometime. Yeah. I'll try it. I'm seeing him later this week. That would be amazing. Yeah, please let him know. In your direction. But Scott's very passionate about trying to help young people in particular, young men, save money in a said-it-forget-it way. And so Noble Mobile, we want to do three things at once, which is a lot, but whatever. Number one, we want to see you actually pay closer to what you should be paying true value for your data.

2:23:13It's about half of what you're currently paying. Number two, we want to incentivize you to doomscrow a little less and look at your family's face a little more. and number three you want to make it easier for you to save money without having to think about it so those are the the things that we think we can accomplish that's a really cool flow what uh what's the customer acquisition strategy besides you telling as many people as possible about it you're starting there uh but but what's the gtm look like sure thing so uh one of the things we figured out was that um i've got a following scott galloway's got a following uh charlamagne's got a following a lot of people have followings and so if you want to help them save money and live a little better you can tell them about noble mobile and for every person you tell you help them but then we also will reward you so what you're going to see is that there's going to be something of a collective that's now helping acquire customers and let people know about this the model that everyone's familiar with is ryan reynolds mid mobile ryan reynolds shows up on commercials and whatnot and so one of the jokes i tell is like i'm talking to various celebrities or influencers it's like look instead of one ryan reynolds they're going to be like 30 of us or 50 of us uh or whatnot and then when i tell them ryan reynolds made 300 million dollars uh then they're you know he did fantastic self-interest kicks in and people are like okay like yeah i'm definitely ready to um to you know uh tell my people um so we're in conversations right now with some very very fun fascinating figures with followings that's a lot of f's um but this is like a consumer centered human-based movement to try and make this industry work better for us it's very cool um what what does it take to actually build a new uh a new telecom company like this you guys obviously aren't setting up your own uh towers but are you are you partnered with with some bigger existing players or what did that look like you know it was a lot of work setting up our own towers uh you know i had to like get out the tool belt now so you're obviously right um we did is we we licensed uh hundreds of millions of dollars worth of data from one of the big carriers and the point i want to make to your incredibly savvy audience is like look You talk about you chose Verizon 10 years ago.

2:25:37I chose Verizon 25 years ago. And we chose it based on something that we thought was like network quality. Like I'm an important person. There's a map and Verizon map was more red than the other maps. Well, it was like in little towns where there were cities. It was like, well, this network's better than that one. This one, it was just vibes based basically. Yeah. So over the last 10 to 20 years, what happened is all three carriers bulked up their networks to the point where they're all pretty excellent. And then now you have some satellite data that's getting piped through the network. So what happened is that we made a buying decision X years ago, and then in the time since, it kind of got leveled up and became commoditized.

2:26:22So what we're doing is we licensed hundreds of millions of dollars worth of data from one of the carriers, T-Mobile, and then now we can make it available to people in a way that's really going to, again, accomplish multiple goals. And the great thing is that it's in this day and age, too, because check this out, guys. You're like me, I bet, and I have approximately zero experience actually switching carriers. Maybe you've got it zero times. I was the same way. I've been on Verizon for 25 years. Every device manufactured from 2018 on now has an eSIM where you can direct it to another network. You keep your number, keep your phone, takes five minutes and you can just bingo, bingo, switch to another network.

2:27:08Most consumers do not understand this. And even I knew it technically. But then when I actually switched from Verizon, even I was like, holy shit. I just ended a 25 year relationship in five minutes and now I'm saving myself, you know,$1 ,200 a year like that. This is like magic. So so now in many ways, the goal is to spread the magic to as many Americans as possible. possible. Are you guys, like, how does it, what's where your, what's T-Mobile's or Noble Mobile's integrations with the actual hardware manufacturers? Are you guys working on deals with, with, like, with companies at the device level?

2:27:49You know, so we, we think most people are happy with their device and will just bring their own phone. There are some people that are under a contract with their carrier uh and obviously there was like a new iphone launch just last week um so there are some folks who and by the way this is true of you guys too i love talking to people like you because we're on the same boat so there was a period when every time a new iphone came out we're like oh my god i need that you need to line up camp out yeah yeah the moment the lines yeah and then that stopped being true like a number of months ago where we're like what's the difference with this one you know the camera's a little better like i'm not even really seeing it this one's a launch and you can see that the device replacement cycles just gone longer and longer.

2:28:33So I talked about before about how the networks have become at parity and commoditized. Our devices have kind of converged around a level that we're perfectly happy with. And so most people we think don't want a new phone necessarily. They just want to not be getting gouged every month. If you do want to get a new device and you decide to go through the carriers by the way I've done this so I get a new iPhone of Verizon and like the device is free but I'm tied up in a two-year contract where I'm paying Verizon a certain amount over this period and I want the folks listening to this just to think about it for a second the carriers contracts with us are so profitable that they can zero out and subsidize a$1 ,200 device or a thousand dollar device and be like hey it's free to you because we're making so much money off you were gonna pay I I mean, obviously Apple's not giving them that device for free.

2:29:24Apple's like, you know, getting paid a lot of money. So it would be more cost efficient for someone who really wants the iPhone 17 to buy one unlocked and then sign up with Noble. If you did the math, you'd actually save money on that versus getting it for free from a carrier. But most people are perfectly happy with their current device and might just want to try a different network. Well, congratulations on the launch. Super exciting. I'm excited to see the different partners that you put together. And I feel like you just can hit the campaign trail again, but this time for Noble Mobile. Yeah, you know the playbook.

2:29:57Hang in, just go truck stop to truck stop and just like help people switch over. Yeah. You guys just read my mind. You know, I'm coming to a truck stop near you in a hot minute. It might not just be me. It might be Scott Galloway. It might be someone else in the Noble Collective. but this is going to be a really fun company to grow because no, it's, it's an industry that's really past due for some modernization and some humanization. Yeah. Last question. Did you, did you raise a bunch of money for this already? Yeah. I think you guys get to do a sound effect. We raised 10.3 million from not just a sound effect and authentic.

2:30:40We have a real gong in the studio. I did enjoy that. I want to bang that gong in person. You're welcome. Anytime. Anytime you're in LA, let us know. Oh, yeah. I'll let you know. I'll be in LA before you know it. Fantastic. I'm a Scottsan investor. Great. A lot of other fun people are investors. Shout out to Corazon Capital for leading the round. That's very cool. And, you know, we're going to grow like a weed. It's going to be fun. Amazing. Well, congratulations. Great to get the update. And we'll see you soon, hopefully. Yeah, we'll talk to you soon, Andrew. Thanks so much. Thanks, guys. See you in LA.

2:31:09Appreciate you having me. Cheers. Bye. and if you're thinking about going to his event in new york you need a place to stay book a wonder with inspiring views hotel great amenities dreamy beds top-tier cleaning and 24 7 concierge service it's a vacation home but better folks find your happy place on wonder.com um hey there's some folks that are traveling some folks that are going to be in la some folks wondering john excellent says what's the galloway appearing on tbpn would be a milestone for me the very first time i force crack my youtube app so yeah i was about honestly literally i'm about to dm john actually to know like is he that anti-scott because i feel like we have some we have a wide breadth of guests here we can have a conversation with all sorts of people um but if he's uh but if he's yeah i'm interested uh we gotta we gotta we gotta let john weigh in on we do decisions like this i think so but um but yeah uh anyway uh what else is in the news uh someone's hosting a website on disposable vape did you see this on hacker news taylor otwell says me stressing about how we're going to build the best back end infrastructure for the future of agentic workflows and quantum mcp are they making internet connected vapes now is that they are i've seen these in person.

2:32:30The LCD screens, the actual screens have gotten so cheap that you can run an entire computer on them, basically. And these are disposable. Can you run an AI factory or a supercomputer? I don't think you can mine Bitcoin effectively. I don't think you can inference the most cutting-edge frontier AI models, but you can do a lot more on a disposable vape. We should get Tyler to build something. Bobby Cosmic, the lone soldier on our TVPN Twitch. Yes, thank you. What if I want to stream TVPN for three hours a day on 5G? Would Noble Mobile be a good fit? That's the benchmark. You can't have a carrier that can't support three hours of live streaming a day.

2:33:11I think it's interesting. So Noble Mobile, I think the messaging is like a bit, you know, they're trying to do a bunch of things at once. I think just a$50 phone plan that helps you use your phone less is like enough of messaging. And I think that they can get a bunch of people to sign up for that. Is it going to be a competitive bloodbath? Yes. I don't know. There's something about saving money that can create these interesting flywheels. Remember Acorns? Didn't that company do very well? It was a, I believe it was a consumer credit card that every time you swiped it, it would round up to the nearest dollar and put the cents in a savings account and compound.

2:33:56And so it was a way to save money. And I believe they got to maybe a billion dollar valuation. It felt like it was - Yeah, they almost SPAC'd. It got canceled. They had 120 million of revenue. This was one of those things, like early FinTech company that was like winning, but then didn't, I don't think they've ever gotten out. I'd be interested to see what happens with that business. But yeah, just save money on your mobile plan is like a simple message. I think Andrew building a coalition of celebrities that feel good about promoting this. Use your phone less. But again, I mean, it's going to be like you said, you don't think that much.

2:34:35We're not the target audience, I think, for Noble Mobile. But there's a bunch of people out there. Yeah, I think I like the savings concept. I'm worried about the revealed preference of people say they don't want to scroll. People like to scroll. No, I think his argument would be even at$50 a month, it's still cheaper than. Yeah, yeah, that makes sense. Well, if you're scrolling X, you're going to be treated to a somewhat tweaked algorithm that Nikita's been working on. Peter Yang breaks it down. He says he hasn't seen a, quote, it's so over, RIP product, or this changes everything, cringe tweet on here in days.

2:35:14The algorithm must be working. and so uh dylan field put it to the test founder figma of course he said it's so over r.i.p cringe tweets new x algo changes everything and as funny as that post is it didn't go viral and so i think i think that they might have uh nerfed that meta we're seeing a new meta emerge already with the b b jeff bezos b sam altman b elon musk spawn in like the the the 4chan style One meta dies and others born. Yes, there is always a new meta. Never lose hope. Actually, I've been liking the new ones. I think it's a funny way to tell a story. And I've actually been learning something.

2:35:58Yes, and it is more in depth. It has more to unpack. It's typically hard to get a lot of text to go viral. The viral posts tend to just be like one or two sentences. But these longer stories, when they're told in this format, seem to do quite well. At the same time, they're rife with misinformation a lot of times, as we've learned from actually reading them. But you've got to go to the source sometimes. Well, I think we have Tim Higgins. We have our next guest from the Wall Street Journal. Tim Higgins, welcome to the stream. How are you doing? Hey, thanks, guys, for having me. I believe we were just reading, some of your reporting, about Elon Musk and the pay package.

2:36:41I believe you authored that, correct? Yeah, I write a lot about Elon. Yeah. And it's the trillion-dollar man. The story of the month, right? You have trillion-dollar hair. Fantastic hair. So it's fitting that you would write. But, I mean, we don't need to wade into Elon territory. Give us the update on the book. Yeah, congratulations on the launch. Well, thank you. Thank you. I wore. It is out today. It's a great corporate drama about Apple, about all of these rivals who want to displace Apple as the gateway to the digital world, the gateway being the iPhone, the app store that really changed the world almost 20 years ago.

2:37:22Yep. The ultimate toll booth. We were feeling the power of iMessage and the lock-in there and how hard it is to build other products that want to deal with communication in your life. If you're in the iMessage world, Apple has you. I mean, there's so many different angles. And people think, oh, you just switch the phone and then you're in a different ecosystem. Like there's a lot of threads holding on. So, yeah, walk us through the inspiration, the reporting process. How much was FOIA'd? How much was reading through, you know, leaked tech emails versus original reporting, talking to sources? What is the process?

2:37:59When did you start writing this book, by the way? Because this is a story that's been unfolding for years. Well, you know, to your point, it's been a 10-year fight for Apple. I mean, people started complaining about that power almost from the get-go. But you really started to see the seeds of what I think of as a rebellion, almost like a Boston Tea Party moment of these colonists in the new world that Apple created, kind of rising up over that Apple tax, what they call it, Apple tax. The 15 % to 30 % commission that Apple takes on digital services and sales, right? Right. And also more so the rules.

2:38:36Right. Exactly. And so when I so I'm a business columnist at the Journal, but I had been the Apple beat reporter at the Journal, had been the Apple beat reporter at Bloomberg. So I've written about Apple for many years and I followed the Tim Sweeney epic lawsuit against Apple very closely. And it paid attention every day during that lengthy trial in 2021 and just was struck by the scale of the drama. You know, in one way, Tim Sweeney is a very interesting protagonist. He's started the company from scratch, had this vision. He's a world builder beyond just being an entrepreneur. He's thinking about the future, whether it's the metaverse and how Fortnite might play in that.

2:39:19And he was looking at Apple as going too far, you know, taking too much, really kind of against that idea of the open Internet, right? Then you have Apple, which is also a storied U.S. company, which had really built something very important to everyday life. The iPhone, clearly the center of many people's digital lives, digital experience. And their kind of argument is, we built this. We keep it running. We're giving you all of these customers. You kind of owe us a little. And isn't that fair? And so you have this really great clash of what is, how should this play out? And so it struck me as interesting.

2:40:03And as I started kind of thinking about it in a bigger way, it really seemed as if it wasn't just Tim Sweeney. You've got the Spotify element. You've got the regulators in Europe. You've got other tech companies in the U.S. and around the world really kind of rising up in this kind of interesting moment, which I thought in a lot of ways was surprising, right? If you went back several years ago and talked about monopolies in the U.S., Microsoft, the old Microsoft, of course, was attacked by the DOJ and fought those allegations. But in the more modern era, Google, Amazon, those were the big tech companies that were painted with that monopoly brush, fairly or not.

2:40:45But Apple, in a lot of ways, especially in Washington, D.C., was seen as something to aspire to. And, you know, what we've seen in the last few years because of Epic, because of Spotify, because of others, has changed the conversation around the power of Apple and the power it has, not just with that tax, if you will, but about the control they have on the digital world. Yeah. How did you... Is Apple good or bad? I'm kidding. I'm not going to make you condense it down. So, you know, we're extremely sympathetic to the startups, businesses, building in the Apple ecosystem and creating a value for Apple users and then obviously paying the tax.

2:41:37I had a funny experience, John and I, the other day. I was, like, trying to download an app. I realized I downloaded a fake app. Yeah, that's right. And then I got to get the benefit of the Apple toll booth because I just immediately went into my subscriptions and canceled it in one click. And I was just thinking at that time, if this was just some regular app on the Internet, it would have just made it way harder. They maybe would have tried to not let me. And that's Apple's argument. They say, you know, this isn't a tax. This is you're licensing our software. Yeah. And it can be oftentimes. It's fair.

2:42:11Yes. They're providing a service. They're providing security. They're providing an ecosystem that the user wants to be part of. Yeah. Were you able to uncover any special deals that Apple has brokered with big companies? I feel like 30 % sometimes they certainly don't want to mention that that's ever been negotiated down. But there's also like the way in which Audible does their pricing. You can't just buy books. You got to subscribe to that. Credits. Part of their weird. Is their system designed to get around these rules? I'm curious. I'd be interested to know about what are the weird edge cases.

2:42:52What are the special deals? So one of the things about Apple, and you compare it to the Android system, Google's operating system, is that Apple likes to pride itself on not doing special deals. Now, its critics would point to some edge cases that they feel are leaning towards helping others, but generally that's kind of the deal, is that Apple has some rules and wants everybody to kind of follow them. Now, it's confusing because sometimes there's categories, and we can get into that. Sometimes Roblox is perhaps treated differently compared to Epic Games in the old way, but just kind of put it under the category that they don't like to do special deals, whereas Android does do special deals, in part because they feel that they have to compete against Apple to win those hot apps, right?

2:43:44The hot video games that drive so much of the sales in this ecosystem. If they're seeing the games go into the App Store and Apple, they feel at a disadvantage. So they're always trying to create an Apple-like experience in the Google world, if you will. And it creates an interesting dynamic out there and kind of shows how these kind of things play out, right? And so one of the things you see in the book is, well, why did Tim Sweeney, why did he go to war against two of the most powerful companies at the same time? And when I say those companies, I mean Apple and Google. It's because he first went to Google and tried to get his app so you could go outside of the app store in the Google world, in the Android world, and download it that way.

2:44:34And he didn't get a lot of success. And part of that was, he would find out later, was that phone companies and some of his competitors in the video game world who he was trying to partner with to create an alternative app store that would be available on the Android system, they were getting deals from Google not to do stuff like that. Now they could do that, but they didn't have any incentive to do it, right? So he figures this out, and then he's like, you know, he's got to go after them both at the same time because Google is accusing him of basically not complaining about Apple's deal when he's got a lot of complaints about that.

2:45:10So really he gambles that he can take to the courts this case that Google and Apple are out of line, and it doesn't go very well for him at the beginning. You remember his lawsuit against Apple, he basically lost most of it. Now, he would be successful against Google later on, but Sweeney essentially loses most of the case. And in Apple, the way they handled it, probably pushing too far, angering the judge over the one thing that she was upset about, this idea that Apple wouldn't allow developers to steer users to other websites to make payments. And she wanted them to be able to do this. And Apple made it so difficult that she really came down hard on them earlier this year.

2:45:56And essentially, very cleverly, these companies now are able to do that. And you see Spotify and Epic take advantage of it immediately, directing users right outside of the walled garden of Apple to make their upgrades to the apps or pay their video game stuff. It's a pretty remarkable crack in the Apple ecosystem. Have they been super effective at driving the majority of users to pay outside of the ecosystem? Or is it more like half of the users are opting for these alternative payment rails? Or are they just saying the only way to buy is now outside of the Apple walled garden? Yeah, it's pretty remarkable.

2:46:39What I understand is that the early data is pretty good for companies that are doing this. One of the other kind of data points that I look at is that Spotify several years ago, as it was thinking about how it could fight Apple in this regard, essentially started doing A-B testing in Europe with its Android app just to kind of run a testing. It was essentially like if this was Apple, this would be the Apple rules and without the rules and kind of a variation of all that to see how much of that was hurting them. And the data, according to the records that I've looked at, was pretty significant, that it was, A, considered a bad user experience, and, B, they were losing out a lot of upgrades.

2:47:20And so the early suggestion is that users like to do that. Like to go outside. Yeah. Interesting. Bullish. Bullish. Bull bearish for everybody but Apple. That's exciting. What else are you following going? I want people to buy and read the book. I just bought it on Audible. I'm looking forward to listening to it. Perfect. Did you just pay the Apple tax? Yeah, how much do you get? It charged me$16. Are you going to get$15.99? We don't need to break it all down, but what is Audible's deal? I'm actually curious. Do you have any idea? It's interesting. Just take a step back to Amazon and Apple and how they were some of the early fights in this kind of war you go back several years ago.

2:48:10And this was back when Steve Jobs was alive. And Amazon is advertising that you can buy your books and you can read it on the iPhone and you don't have to pay for it. It doesn't go through Apple's system. And Apple was just steaming over this. And essentially they came out with these rules that if you were selling it in the digital world, you had to basically use Apple's system. And so that's why you would see on the Amazon app, you couldn't buy through the app. You couldn't buy books through the app because they didn't want to pay the Apple tax, if you will. And one of the first companies after this rule earlier this year was changed because of the court ruling was Amazon.

2:48:55You can go buy your digital books now through the app because it's directing you outside the Apple system. Yeah, I remember. I would click and wind up in a web browser, even though I had the Amazon app installed every time. What should people be paying attention to going forward? I know a lot of these cases, Apple has an opportunity to kind of fight, you know, even after the judge decides one thing, they can come back and say, you know, continue to fight it. But what else is working its way through the courts? Apple's appealing. They feel the judge got it wrong in the U.S. They have been fined in Europe where regulators there fined them over their conduct of how they treated music apps like Spotify.

2:49:41Spotify behind, along with other tech companies, pushing for the DMA, that kind of landmark tech bill that was passed a few years ago that aims to limit big platforms like Apple. Apple's already been fined for that, but they are working to make rules that they think are fair, but that the folks that are against Apple would say don't kind of take the spirit of the law and really kind of an example of how Apple fights this stuff tooth and nail. So that's to watch. The bigger issue here, perhaps, is just the way that Apple has been kind of painted as a monopolist, fairly or not, it overshadows or it kind of tails them now as they go forward into new tech models.

2:50:24Everything they do is looked at as, is Apple abusing its power as they move into new spaces? So in AI, for example, I think investors would say Apple's behind in AI right now. There's a lot of frustration out there. And so even when they do a deal with OpenAI to help kind of bolster their chat capabilities with Siri, you see Elon Musk come in and sue and say, wait a second, this is monopoly behavior. Apple is using its power to hurt other AI companies with this kind of deal, right? So put aside the merits of the lawsuit, you've got Apple's getting in this position where it's having to defend basically everything it does against the idea that it is a monopoly.

2:51:05And that is a distraction that the company has to deal with. And that makes me remember a generation ago where Microsoft had a very similar situation as the DOJ went after it and regulators in Europe went after it. And Bill Gates and others at Microsoft have talked about how that period was a massive distraction and contributed in them falling behind as the new edge technologies took over. They missed out on mobile, if you will, and some other things. And, you know, there's some similarities here as you look at AI and where Apple is at this moment. Yeah, that's fascinating. It is such a distraction because, yeah, with these high stakes lawsuits, the CEO has to go.

2:51:45Like you just can't delegate it fully. I mean, I'm sure there's teams of lawyers and whole legal departments that handle 90 % of it, but it's still something at the top of the list. They haven't done any big M &A, right? They're doing a bunch of little deals. Yeah. It's hard to get locked in and focus on the next generation of technology. Well, congratulations, Tim. Congratulations to the book. The book is iWar. It's available everywhere. Books are sold. Please go support. I'm excited to read it and come back on anytime. Leave a review if you buy it. We have a five-star review. We said earlier on the show, we made it into the journal for the first time ourselves yesterday.

2:52:16It's a full circle. I read that. Congratulations, guys. Thank you so much. The journal is the backbone of TVP. It really is. The reporting is fantastic, and we appreciate everything you do. Thank you so much for hopping on, Astro. Thank you both. Have a good day. Talk to you soon. Bye. And up next, we have Alex from OpenAI coming out of the Restream Rating Room into the TVP. I'm going to let you start it. I'll be right back. I will kick it off with Alex from OpenAI. We are bringing in Alex. How are you doing? Good to meet you. Hey, how's it going? Nice to meet you too. Fantastic. Thanks so much for joining.

2:52:51Would love to get a brief introduction on yourself, your role within OpenAI and then the update today. For sure. So, hey, I'm Alex. I'm the product lead for Codex. And yeah, actually, it's been a pretty fun morning. We spent the morning looking for GPUs because we shipped a new model yesterday and demand was a little bit too higher than forecasted. And so we were running the model actually a lot slower than we were hoping and luckily we just fixed that. So the update, we just shipped a new model for Codex. Codex is our coding agent. It's an agent you can work with everywhere you code. So like in your terminal, your IDE, GitHub, even your phone, which is actually a crowd favorite.

2:53:29That's great. The goal for Codex is to build it into an AI teammate. So, you know, just like a human teammate, you start out working together with it. eventually you start delegating tasks to it and then eventually you just give it a laptop some permissions and a job and you just say like please prompt me with tasks that you think are worth doing yeah um so go ahead uh yeah i i have so many questions um i mean first i'm interested uh you running the model slower actually helps you if you're in a gpu crunch is that just because you're spreading the amount of uh inference across like a broader user base essentially yeah just think like if you you know on your laptop like if you open like way too many programs they all just kind of run a little slower.

2:54:06So the solution, you know, you could close programs on your computer, but we don't want to do that. No, of course, because you have other clients and other users and businesses and stuff. Where, like, take me through some of the use cases. I mean, we were talking to Doug O 'Loughlin over at Semi Analysis Fabricated Knowledge, and he was saying that he'll use what is traditionally like a coding agent to actually go and do something that looks like a deep research report. I had a sort of magical result where I used a coding agent to do a deep research report, but then instantiate it in HTML. And then I was able just to open an HTML page locally.

2:54:45And it was this amazing thing because it just has a different set of UI that it can pull from that's not native to just the Chachapiti app. And so that's one world. But then I'm sure there's people that are using this in the business context, deploying it into like large code bases. where are people having the most success these days? Yeah, totally. I mean, coding agents are really flexible. The primary use cases for coding agents and Codex as well are to write code, answer questions, and review code as well. And that's really what's been sticking. So we've seen like well over 10x growth in usage just over the past month.

2:55:20And most of that is improvements in how people use it to write code and answer questions. But we're now starting to see - You said 10x growth in the last month? Yeah, over that. Yeah, it's growing like crazy. And most of this is just because we've been improving a ton. So if you go back, rewind a little bit over a month ago, we had the Codex CLI, which is a coding agent in your terminal. We've landed a ton of improvements to that. Some people like buttons, and they like using the coding agent next to the code that they're editing themselves. So we shipped a VS Code extension. VS Code's a really popular IDE.

2:55:55And that's quickly becoming a super popular service as well. And, you know, one of the special things about Codex is that it doesn't only run locally on your computer. Like a teammate, it can also run on its own computer. And we call that Codex Cloud Tasks. And so we've landed a bunch of improvements there as well to make it like way faster, you know, another thing. So that's a huge unlock for mobile, probably, right? Yeah, yeah. It's a huge. I mean, you basically can't really code on mobile unless you have that. And so a lot of the work we've been doing is just like making those like fundamental surfaces where you use Codex better.

2:56:23Yep. And, you know, that kind of started that growth curve, which we're really excited about. And then yesterday, sort of building on the success of GPT-5 and listening to a lot of the feedback we've been seeing around how people use GPT-5 specifically for coding in codex, we shipped a new model, which is GPT-5 codex, a version of codex that's like basically like further optimized specifically for the codex type use cases. And that's been really cool. You know, for example, one of the big areas of feedback people had was that, you know, they sometimes GP5 would think a little longer than they were hoping, specifically when they asked the coding question in Codex, when they were just asking a quick question, but sometimes they wanted it to work longer.

2:57:02And so one of the things we did is we made it so that with GP5 Codex, we can actually more dynamically change how much time we're spending solving the user's question. And so for instance, if you ask a quick question, you just get an instant answer. But also we're seeing people, you know, using this thing, just like letting it run for over seven hours. And that's an arbitrary number. I'm sure you can get it to go longer depending on the problem. And yeah, we're seeing a lot of feedback like that, that people are really excited about. The model is also, you know, because we had the opportunity to really go deep on software engineering and train for those use cases, the model is also a little better at things like code quality, front end, steerability via agents.md, and some other things like that.

2:57:43I saw a post earlier from Theo. He said, GBD5 Codex is, as far as I know, the first time a lab has bragged about using fewer tokens. Hope this becomes a trend. Do you think that's going to become a trend?

2:58:25I mean, I guess like tokens are not tokens. Basically on easy queries, we're using like 90 % fewer tokens to answer your query. And then on hard tasks, complex tasks, we're like basically using like double. Are you using the term model router? Is there something that you can share on the architecture there that is actually driving that? Because that feels like a huge unlock. It was, of course, magical the first time you went to ChatGPT and got it to dump out, you know, pages and pages of coherent text. And you get to the bottom and it's all coherent still. And you're like, if you're coming from the GPT-3 world, you're like, wow, this is a huge breakthrough.

2:59:00But then quickly you realize, like, I don't necessarily always need 20 pages of response. I can have a smaller output. And so is there anything that you can share there? Yeah, I mean, the team really cooked here. And I expect us to, it's like part of the exciting stuff that we're doing with Codex is like these like very bespoke interventions. Sure. But no, it doesn't have a router. But there's a little bit of secret sauce there that we're not going to go too much into. Yeah, yeah, of course. How are you thinking about model naming conventions and model numbering conventions? There was a time when model number correlated to big circle, and then GPT-5 was a much more holistic set of improvements.

2:59:41You brought those to Codex with Codex-5. But is there a world where Codex diverges? Are you always building on top of the same release cadence now? Is there anything about how you think about messaging? What actually changed to the actual user to know, hey, it's time to dip back in if you tested this or you want to go deeper in a certain area? Totally. I mean, so our goals with Codex are just to build an amazing coding agent. And if you think about what an agent is, it's a combination of like the model and what we call the harness, which is kind of like the set of tools that the model has and then an interface for the user around that.

3:00:20And so the reason we put Codex in the name is because basically we wanted to just like, we're really investing in codex and like in software engineering generally. And so we wanted to have the opportunity to go like much deeper and push harder on things that like might not benefit like sort of general users of a model. And in fact, we wanted even the ability to like make certain things worse. So not that I've gone and checked, but like GP5 codex, if you're not using it for codex, is probably going to do worse at like a bunch of general tasks than GP5 will. Sure, react to this Rune post. Right now is the time where the takeoff looks the most rapid to insiders.

3:00:53We don't program anymore. We just yell at Codex agents, says Rune on X, but may look slow to everyone else as the general chatbot medium saturates. People have been saying, oh, this is the 90 % of code being written by AI agents. And there's this debate on, are we just writing 10 times as much code? Are we doing 10 times less work? Like, give me a little bit more color on how you're using Codex internally to build the actual product. Yeah. I mean, we use codecs internally a ton or a lot. And I think we're at the point now where, for writing code, these coding agents are massively adopted. And it's kind of a stylistic choice, how much of the code it's writing, but it's the vast majority.

3:01:38I will say that our goal isn't to... We don't actually have a goal of, yeah, let's make sure the maximum amount of code written by agents it's like really what matters is like your velocity as a team right yeah um but uh sort of my like maybe slightly spicy take here is like actually one of the greatest limiting factors um on the utility of coding agents right now is not uh like the model's ability to write code but it's actually the human ability to like just prompt it and like make use of the agent so for example um we shipped a code review feature that's like you know doing really well a few weeks ago.

3:02:12And, you know, the model is capable of doing code review, we actually did a ton of work to make it even better at code review. So when Codex does a code review for you, it'll like actually get a whole copy of the code base, it can like go read everything even outside of your PR, it can even execute code to validate, which like a human is probably going to be a little too lazy to do normally. But the big unlock isn't isn't even necessarily that like even GPT five is a great code reviewer. But the limiting factor is like, are you going to go prompt the model to code review every single pr that you get and the answer is no so what we did is we just like automated that built it in a nice way like integrated with github and now we're seeing uh you know it's just reviewing the vast majority of prs in the vein open opening eye repos and catching like serious bugs it's actually kind of become a meme when uh you know if so we ask uh people internally like hey react with thumbs up if it's good just so we can like kind of see um and whenever and like reply with what's wrong with this review if it catches an issue that the engineer disagrees with.

3:03:07And, you know, often it'll be the case that an engineer will kind of see this issue, reply and say, yeah, I disagree. And then like two days later, it'll be like, actually, this was a mistake and we need to actually follow the model's catch. What does it take to make it on the OpenAI Codex team these days? I mean, it feels like the skill set of just someone who's a programmer is changing. Now you're prompting, designing. uh what is it give me some advice for young people or people entering the the workforce i think uh the thing i would say for open air coding team and by the way we are hiring uh both folks externally uh and then also you know see people can transfer internally yeah uh probably the two most important things is uh kind of you can just do things attitude and speed um and i think those are pretty general things that i would recommend um you know if you You mentioned like advice for young folks.

3:04:01Like, I think it's like a sort of crazy time now to be early career because so much is just possible to do and possible to learn quickly. But at the same time, because of that, everyone else is like just doing a lot of stuff and learning a lot of stuff quickly. Right. So I think to like really make the most of the key is to just like really do stuff, especially like if you're in a technical field, start doing stuff beyond your coursework, but actually start building things. Yeah. Increasing returns to like high agency. It's funny, we're in the era of agents, but then also high agency people seem to be doing better than ever.

3:04:35It's kind of a paradox, but it's a good time to be able to have a positive attitude and actually go and just try and build things. It's a good time. Yeah, and I think this is true. Go ahead. No, go for it. Yeah. No, I was just going to say, I think this is true as well at work, right? Like maybe before you would have a number of like very specialized role, like PM, designer, front-end engineer, back-end engineer, like, you know, SRE. and now because of all these coding agents like everyone is able to just do a lot more and so you can have people like owning like entire problems end to end um you know one fun example actually leading up to yesterday's launch is um our designer built this like really cool ascii animation of a coin that's rotating um and you know that's like kind of a hard thing to get right oh yeah and so um over the weekend he just like vibe coded an ascii animation editor like he by that lets him like edit the, you know, and like, that's like such like stack compression.

3:05:34Totally. And it's awesome. Yeah. I mean, there used to be a time when it was like, you're a front engineer, you know, JavaScript, but you don't know Ruby on rails or Python in the backend. And like the backend guy doesn't know, or gal doesn't know, uh, JavaScript or whatever. And they got to talk to each other and there's loss in translation stuff. So, uh, yeah, it's an entirely new era. Very exciting. Yeah. And it comes with problems too, right? like if you have all these folks uh like the ask animation viewer i mentioned like that's throwaway code so i guess there's not much tech debt it's just like pure acceleration we actually write a lot of throwaway code like different kinds of dashboards editors mini tools um but then you know sometimes you have folks who are like specialized the one thing now like able to write code for a different thing uh and you know that actually creates a different kind of problem which is like code review and expertise and so that's why like what we're trying to do with codex now is like we're investing in making like the fun you know fundamental like code gen experience great and just writing code but we're also starting to like reach out now so we're really investing in code review because that's like actually becoming the bottleneck like having qualified people be confident in code we're really investing in validation so like before you review the code like can you assert functional correction correctness uh one thing the codex can do now is like take screenshots it'll like actually write a playwright script to like manipulate the website and then take the right screenshot of it so you can actually look before you read the code and then we're thinking about where else can we make it really easy to engage the agent before it actually gets to you on your computer doing work it's amazing congratulations thank you for the update thank you so much for hopping on the stream we'll talk to you soon good luck out there have a good rest of your day GPU's on fire good luck well that concludes our guests for the day we gotta catch a flight I got to listen to iWar which is in my audible now by Tim Higgins You can go find it in Audible, find it on Amazon.

3:07:24We will see you tomorrow live from Meta Connect. Later show tomorrow. Later show tomorrow, 4.30, where I think we're starting. We'll be posting updates on X, keeping you updated, and stay tuned. Thank you so much for tuning in. Have a wonderful evening. Thanks to Bobby Cosmic for holding it down in the Twitch chat. The legend. Thank you to John Exley in the YouTube chat. We'll see you guys later. The team's like unfading it out. We love you guys. We'll see you tomorrow.