The Lawyer Who Beat Meta and Google, Revisiting The Jetsons, Japan Twitter | Tae Kim, Logan Bartlett, Sam Stephenson, Ben Broca, Brett Adcock, Andrei Serban

26:38You actually can't. There's some creators that do this. But anyway, like surfacing those in apps, I think that will help users feel like they're in more control. And realistically, I don't think it will be super damaging to any of the platforms because most people won't opt into that, but certain people will. And in general, it'll just like increase public perception broadly. We've already seen this with a lot of the LLM companies where like you can go in and you can fine tune and add a custom prompt and kind of talk to it about what you like and don't like, change the personality. I think people have been asking for that for a long time, surfacing it.

27:11It seems like a win-win. And so something along those lines seems in the cards. Anyway, we can debate this. But first, let me tell you about Sentry. Sentry shows developers what's broken and helps them fix it fast. That's why 150 ,000 organizations use it to keep their apps working. And let me also tell you about Plaid. Plaid Power is the app you use to spend, save, borrow, and invest, securely connecting bank accounts to move money, fight fraud, and improve lending, now with AI. So do you have any other pushback on my take? Tyler had some pushback. Should we go to him?

27:43Ben Broca:Let's go to Tyler. What do you think? I mean, yes, I guess there was a few things. I mean, one thing is that like... Not enough props, right? Too many analogies? Too many parables? I think you can say that, okay, yeah, it's the content that's the problem. But the content is kind of downstream of the features, right? Because you didn't see short form video. The medium is the message. Yes. And so it is possible for me to create a platform that incentivizes addictive content. And that's the retention curve. So retention editing makes it more addictive. You become addicted to the content, but it's because of the features.

28:18Ben Broca:Yeah. So I think that's broadly - Pretty good argument. The steel man that you can make for like Lanier's position. Yeah. And then, I mean, there was other stuff, I think, on like just the nicotine analogy, we were talking about this, like, okay, so you have nicotine like broadly, and then below nicotine you have like smoking, which is like definitely very bad for you. And then you have like, you know, pouches or stuff like this, which is like probably less bad. Like it's just nicotine, there's no tobacco, so like maybe this is like less bad. And so maybe the equivalent is like, you know, the cool snowboarding videos on Instagram are like the, you know, the cleaner like nicotine stuff.

28:53Ben Broca:And then the like graphic. Still addictive, but not harmful. Yes. Unless you're Jordan and you're going to try and do a double cork 1260 and eat it. Get smoked. Get smoked into the ground. Yeah. But then on the other side, you have like the like, you know, very graphic stuff on Instagram that like we don't want people to see. And that's like the, you know, the cigarettes that's like going to give you cancer or whatever. Sure, sure. So I think they're like, I mean, I guess it's still in agreement with what you're saying, but like. Well, this is what nicotine is still, if you're under 18, you can't buy.

29:22Which was like, there was an addictive component and then there was a carcinogenic component, and they needed to sort of separate those out. And where we landed as a society was like, the addictive component is acceptable for the, it's suitable for the protection of public health, according to the FDA. And so they are approving new products that are addictive, but not carcinogenic. And so you would imagine even in the most strict ruling where every new social media platform needs to be approved, you could potentially use all of those addictive features as long as the content was not carcinogenic with inside that app Yeah, and that would be like a new nicotine gum basically

29:57Ben Broca:Yeah, like basically I'm saying like right now if you're under 18 You can still like there's like parental controls and you if you can't be under 13 or whatever But like it's like very not it's like poorly in it, you know enforced Like you can actually see a lot of the bad stuff you're under 18 on the game or whatever Yeah, so like directionally like there, you know, you can be against the the ruling of this yeah but like the the parental controls that people are like asked for are still like very much not there yeah yeah no that makes sense so i i i think i have a potential solution let's pull up this image of a cigarette package in europe oh yeah what is this so pull this up let's pull this is the hardest challenge while we pull it up let me tell you about okta okta helps you assign every age and a trust identity so you get the power of ai without the risk secure every agent secure any agent and let me also tell you about turbo puffer there we go back to search built from first principles and object storage fast 10x cheaper and extremely scalable okay so this is like typical cigarette packaging in europe john you probably wouldn't know this because you're very american you're very loyal and you and you don't you you avoid overseas trips as much as possible so on any given cigarette pack in europe you're going to see like a really terrible image.

31:12This woman apparently is coughing up blood. And so I think what a potential solution that Meta could do is as soon as you open Instagram, it makes an AI-generated image based on the last picture of you that you posted on social media and it just makes you look terrible. And it says, like, warning, social media will destroy you. And then you can scroll past it. It could potentially show you a tech neck. Are you familiar with tech neck?

31:42yes yes it's just a crazy image of you with tech neck it shows you your phone and then so you can scroll past it yeah every time you open that it's a new image it's a reminder it's a new image of you looking the worst wasting your life away are the ai labs uh lobbying to get all right we can put this we can put this away are the ai labs uh lobbying to get that removed because i think most of their timelines suggest that lung cancer will be cured by ai any day now so uh potentially you could start smoking again. Has anyone come out as pro-smoking? I don't think anthropics come out with their, their, their, their anti-anthropic, you know, has the joke that they make with journalists that they kind of got caught on.

32:20If AI is going to cure liver cancer, it's game on. It's game on. It's game on. You can drink as much as you want because that's the, because you get a liver disease if you drink too much. And so if AI, if I'm going to be able to vibe code and mRNA vaccine to cure my liver cancer, I'm going to be oozing for sure. It's the only rational thing to do.

32:41Ben Broca:Yeah. Well, this is also kind of like when companies are saying like, oh yeah, work-life balance is super important. Yeah. So then their competitors will, you know. Yes. Yes. Yes. Yeah. People on topics should tell people to open AI to start drinking a lot because AGI is going to cure liver disease. Yes. Yes. Yes. Yes. Yes. This is good. This is good. Okay. Let's revisit the Jetsons. Okay. Revisit the Jetsons. I'm sure you've seen the Jetsons. Where's my flying car and three-hour workday? So I'm going to be learning about the Jetsons. John is going to be revisiting. The 1960s version of the future is way more fun than our reality, but when it comes to innovations, we're catching up.

33:14Interesting. Let's see. Nicole says, I recently spent a weekend doing deep investigative research into future technologies. I binged the Jetsons in my sweatpants. For the uninitiated, the forgetful, this space-age family sitcom features George and Jane Jetson living the American dream in an apartment in the sky with their two children, Dog Astro and Robot Maid Rosie. The show is set in 2062, a century ahead from its original 1962 air date. It's full of fantastical inventions such as flying cars, dinner generating machines, and canine treadmills, complete with fire hydrants. The upbeat vibe is markedly different from the apocalyptic, at times murderous, sci-fi of today.

33:54The 1960s were full of optimism about what the 21st century would bring, and some of it actually has come true. While we've still got a few decades before the Jetson family is meant to arrive, I dug into some of the show's technological hallmarks and determined how close we already are. Video calling. She says, absolutely. In lieu of a home phone, the Jetsons had a video phone. The show's creators couldn't fathom mobile devices, but they were spot on about video calling. Now, to be clear, we are still working on, with one of our business associates, it's like a video call that doesn't stop halfway through and just cancel.

34:31So the Jetsons didn't predict the free tier of Zoom. Yeah, the free tier of Zoom was not considered in the Jetsons where you're talking to someone, you're clearly going to go long on the meeting and Zoom's just like, goodbye. It's over. It's over. It kicks everyone out with no notice. Is that a new thing? I feel like it used to do a countdown. I think it did a countdown too, but now it's just like, get out of here. the host the plus tier is gonna blow in the jetsons they could even create deep fakes to stand in for them on camera whoa that's cool i didn't realize that facetime's gotta get on the oh you know the other thing they haven't cracked with facetime is like if you facetime a group of people yeah like most of the people won't even notification and don't know that it's happening so we haven't cracked the the notification part yeah yeah this is good read this next line when george secretly attended a robot football game, his simulacrum told Jane he had to work late.

35:31He's like using a deep fake to lie to his wife. This is so 60s. Do not do this. Do not do this. This is dystopian. Flying cars. See, it's not all optimism over here. Flying cars and travel tubes, sort of. There isn't much walking in Orbit City. A conveyor belt brings George from bed to the bathroom to get to and from his classroom. Elroy jets through a series of air tubes called the School Homing Network. when the wrong child shows up at the Jetsons home Jensen sends uh Jane sends him back with the push of a button and they also use personal vehicles though ones that typically fly George arrow commutes in a glass dome saucer that folds into a briefcase that's pretty cool we're pretty far from there we do have helicopters but they're very expensive I always fight people on the flying cars don't exist thing because like we do have helicopters and people some people get to use those, but they are not nearly cheap enough.

36:22But we've got to get them way down. Here in the actual future, we're still toting around on pavement-pounding automobiles. A version of flying cars, however, is very real. It's called an eVTOL. Look at this. Pivotal Blackfly is a solo-piloted aircraft free to operate in unrestricted airspace. An upgraded version called the Helix can be yours for$190 ,000. You don't even need a pilot's license. That's pretty close, but I would still say we are not near the flying car because they're just not like they're way less flying car rides than waymos for example so we're just not right there uh push button jobs almost george works as a digital index operator at spacely space sprockets for approximately three hours a day three days a week as a button pusher he makes enough for to support a family of four even though majority of his day is spent with his feet up on his desk okay they basically nailed this there's some people out there that are basically button pushers right now vibe coding yeah tbd on on the revenue side but working three hours a day three days a week you know we we work three hours a day five days a week and uh maybe the future is three just monday wednesday friday streams we can live the the jetsons future now we work i don't know that would be that would be devastating for us yeah Yeah, until then, we'll be working alongside.

37:47Space colonization, nope. Yeah, they live above Earth with houses built on tall stilts. I like that. To avoid the planet's environmental inconveniences, the stilts can rise above any inclement weather, and space itself isn't out of reach. In a classic episode, Elroy goes to an asteroid on a school-filled trip. We're not quite there. Musk had preached of populating Mars, but now his focus has turned closer to the moon. Meanwhile an interplanary space race between US China Russia and UAE and the European Space Agency is well underway Robot Mades not exactly but we're getting Much closer there. It's funny that that Brett Adcock's coming on today.

38:29Yeah, he's working on the flying car. Yeah, he's working on the robot made working on the robot made He doesn't have a space thing. He's now work. His new lab is basically like a button, you know, seems like a button pusher gadget in new gadget induced pain yes and now for the show's biggest oversight no touch screens there are lots of visual displays but they're primarily operated by dials levers and other physical controls we got some levers back there in the studio uh well the show may not have anticipated touchscreens it nailed a key side effect of constant use of gadgets repetitive motion injuries orbit city is full of buttons and overworked fingers are a running gag on the show Jane regularly does digit workouts and complains that her pointers are sore.

39:12Here in 2026, office workers often suffer from texting thumb after scrolling through endless feeds and tech neck after craning down to look at mobile devices. And don't get me started on my strained hand with carpal tunnel syndrome from all the clicking. 36 years and counting. We may not be living as exceptional a future as the Jetsons, but we've still got three and a half decades to catch up. By then I will be twice as old as I am now. I've already witnessed the dawn of high-speed internet, the iPhone and generative AI. How many tech revolutions will we experience in another 36 years? By the time we hit the show's 2062 deadline, maybe we will finally live in space or make our current planet more habitable and make a comfortable living on a nine-hour workweek.

39:55Tyler, what do you think? Predict your timelines for 2062. Will we get space colonization?

40:03Ben Broca:How do you define space colonization? Living not on the earth above the Carmen line, that's your primary residence more than half the year. How many people do it? Just you can do that? Anyone can do that? Yes. Anyone with like, if you can afford an apartment for a few thousand dollars or a house that's above$500 ,000 in America, you can choose to live in space. So I would assume like population of millions. It probably depends on the industry that is chiefly benefited from people living there. Button pushing? Well, big news. Sean Frank is in the chat. He says, I'm here, guys. H-E-A-R. So he's trying to signal that he's listening to you, Tyler, and to you, John.

40:58Well, thank you for coming in. Let's read him some ads since he's here. Let's tell him about Restream. One live stream, 30-plus destinations. If you want to multi-stream, Sean, go to bluestream.com. And you know what we got to tell them about? We got to tell them about app love and profitable advertising made easy with Axon.ai. Sean, get access to over 1 billion daily active users and grow your business today. Andrew Reid says, the faster technology progresses, the harder it gets to print something in the office. We have experienced this. It's very true. The brothers. Aaron from Box says, Reid's Law.

41:33I know you may have wanted a better law, but I don't make the rules. Yeah, it's very, very difficult. Apple has just like never done the printer, I think for environmental reasons. I'm not exactly sure. But like there's never been like, oh, the gold standard, the Tesla printer is just like the one you get and it does what it wants. It just does everything flawlessly. And it's at that like, you know, five nines of reliability. We've had a pretty good run with our printers, but we're always in the market for new printers. So there's more. We are, we're looking for a new printer right now for a special project.

42:12Anyway, let me tell you about Cognition. They're the makers of Devin, the AI software engineer. Crush your backlog with your personal AI engineering team. What I hate most about technology in hotel rooms, Jordy, I want your take on tech in hotel rooms. I want to know about your experience when you walk into a hotel room. The Wall Street Journal says when you book a hotel room, you can count on some things like shampoo, a hot shower, some way to get a cup of coffee. But a stress-free technology experience? No way. Not even with basic technology you find just about anywhere. Like TV, Wi-Fi, and outlets for your devices.

42:45Unless you carry a suitcase, bowl of gadgets, cables, and adapters, you're risking every kind of tech frustration. Did you know that Ben Thompson carries a special device that acts as a Wi-Fi repeater? when he travels. So when he goes to a hotel, he logs into, yeah, this is amazing. He logs in to the hotel Wi-Fi, I believe, or the plane Wi-Fi through that device. And then all of his devices connect to that automatically. And he'll bring like a fire stick so he'll be able to watch TV shows and his laptop and his phone. Everything automatically syncs to that device and it like reroutes it. I thought that was a very interesting thing that he's clearly optimized a lot.

43:28He's like a huge, what is it, like gear bag guy. He has all the wires dialed, as I would expect. I usually forget to bring a charger.

43:40Hotels are missing out on a fundamental truth. In a world where so much of our work, travel, and relationship experience is shaped by technology, the quality of a hotel's tech service is core to what it's like to stay there. Give me a hotel room that lets all that tech fade into the background so that I can focus on my trip. But no, here's what you usually get instead. TV muddle. I suspect I've logged more hours troubleshooting hotel TVs than I have watching programs on hotel TVs. Okay, there's a bit of a retro charm in a TV that flips on and instantly tunes to a live network broadcast. But in the streaming age, I'm just as likely to crave a little quality time with Netflix, Disney Plus, or Apple TV.

44:17Many hotels have caught on to this reality by offering some sort of streaming option, but they approach this in so many different ways, you never know what you're going to find or what tech you'll need to make it work. needy Wi-Fi is another one most hotels I've visited recently seem to have figured out the charging extra for Wi-Fi makes about as much sense as charging extra for a better toilet everyone needs to get online so you might as well build it into the price of the hotel room so now that we've taken that great leap forward why are we still forcing people to log into the network not just once per day but over and over again once per device each and every day or often several times a day it isn't usual it isn't unusual for me to log into hotel wi-fi 20 or 30 times a day i think you're doing something wrong honestly honestly just i don't i don't this doesn't resonate no with me at all you're fine the only thing i want from a hotel yeah is to be able to order room service without calling someone okay that's like the only thing yeah and and hotels miss on that yeah for the most part like if you have a little ipad or you could even order on the tv app that would be amazing and you get like a domino's pizza tracker type thing that's all i want yeah i i feel like they kind of deliver on everything else and i don't watch i i was uh listening to i think it was george hots was explaining how he ordered room service in a hotel he was staying at and he vibe coded an app that interacted with the ordering service so that he didn't have to talk to them.

45:44And it basically like read the entire menu and then like created like a voice agent to call or like, or like reverse engineered the API of the ordering menu. And he was able to order by command line, just like checking into hotel time to build a CLI. It's a truly the future. I love it, but it is, it is. And you know who else is vibe coding these days? Gary Tan. Ben Hilack has a joke here. He says the year is 2027. Gary Tan has just crossed 1 billion lines of code per day. Water to three-year-old Californian towns were diverted in order to cool his locally ran LLMs. Riots erupt, and protesters demand answers to one single question.

46:24What is he building? We got to have GT on. I can't wait. Let's get Gary on. We got to get Gary on. We got to know what Gary is building. People are joking about this. What was the latest stat? It was something like 80...

46:39Ben Broca:78 ,000. 78 ,000 lines of code per day on Gary's list, which is his blog. It's a blog. And he's built blogs before. He's built these sites. But I guess with all the testing suites and packages and mobile optimization, I don't know. I can't imagine the volume of code that will be generated when he creates a mobile app for it. It's going to be trillions of tokens going into that. Anyway, let me tell you about LabelBox. RL environments, voice, robotics, evals, and expert human data. LabelBox is the data factory behind the world's leading AI teams. Sam says, I remember when this was announced but didn't fully appreciate the size.

47:20That's a hell of a cluster. The Department of Energy will basically be a frontier AI company. NVIDIA is collaborating with Oracle and the Department of Energy to build the U.S. Department of Energy's largest AI supercomputer for scientific discovery. The Solstice system will feature record-breaking 100 ,000 Blackwells and support the DOE's mission of developing AI capabilities to drive technological leadership across U.S. security, science, and energy applications. Another system, Equinox, will include 10 ,000 NVIDIA Blackwell GPUs expected to be available in 2026. both systems will be located at Argon and will be interconnected by NVIDIA networking and deliver a combined 2200 exaflops of AI performance.

48:06We've talked about nationalization before. We haven't talked about privatization. We could potentially spin this out, take it public. There's an option here.

48:15Ben Broca:So I was interested in this. I looked, this is going to be like somewhere around like a quarter of a gigawatt equivalent of 100 ,000 black holes. Half a Meta Campus, I think. I think Meta is working on 500 megs. Yeah, I think, I mean, Hyperion, like the end state is like, I think a gigawatt or more, right? Yeah, more, more. But that's like the first big jump for them. But the default Meta Campus, I believe, is around 500 megs. Cisco, to acquire Restaurant Depot. Not our Cisco. Not. Not even close. To acquire Restaurant Depot for$29.1 billion. dollars before we take you through this let me tell you about the real cisco critical infrastructure for the ai era unlock seamless real-time experiences a new value with cisco there's only one cisco in our hearts uh and this is cisco with an s i i dislike cisco why because every time i find it a very frequent experience where there's a new restaurant coming to my area I'm excited about it.

49:21They invest a million, two million dollars in building out this incredible space. Looks great. And then you eat there for the first time and you can tell that they're just sourcing like Cisco. I'm not going to say slop, but the food quality is not great. And then it's like, why did you put all this energy into making a beautiful space? And then you're just, you know, chefing up generic food. It doesn't make any sense to me, but I believe the founder of Restaurant Depot. Look, I think I saw it somewhere. In the Jetsons, they had dinner generating machines. Dinner generating machines. How do you think that's going to happen?

49:59Ben Broca:Travis Kalanick built this. Isn't this Cloud Kitchens? No, no. The dinner generating machine or Cisco? Because Cisco's - Dinner generating machine. Yeah, yeah. That's basically it. But it's all part of a pipeline. The founder of Restaurant Depot - What happened? who just sold for$29 billion, was born in 1932. Whoa. 94. He's still kidding? I think we should hit the gong for him. Let's do it. Congratulations. Great to finally get a solid exit. It's never too late. So if you're 93. I can buy that sports car. Finally. Yeah, that is a true overnight success. Congratulations to him. Excited. I mean, you know, we got to give food to people.

50:43people are hungry there's some good things you know maybe maybe they uh they stock some raw milk and you're on board then you know who knows okay before we play the next video let me tell you every day is a fight between the advertisers and the viral videos let me tell you about railway railway is the all-in-one intelligent cloud provider use your favorite agent to deploy web app servers databases and more while railway automatically takes care of scaling monitoring and security, and then we can play this video. We're heading over to Japan. Yes, this is the key sport that we will all be picking up in 2026.

51:19This is going to be the hottest thing in San Francisco with the hills and the office chairs. So there's an office chair racing league. Look at the speed and the technicality here on the corner. It's incredible. It's incredible. This athlete says the corner once controlled me. Now I control it. I think I got something. You you have quite a bit of leverage with the legs The six-year transformation is crazy. I mean this guy is incredibly quick lost and we need to bring this to the US We need to bring this to the US chair racer Miura Going around the devil's hairpin All right, so Tyler, what is happening on X in Japan?

52:08Can you break it down? I don't know all the internals, but it seems like

52:11Ben Broca:Nikita's been posting about this, but I think they basically introduced all of Japan Twitter onto normal Twitter. Oh, because of translation. There's been translation for a while. But for some reason, this weekend, half of my timeline was just Japanese posts. All about America, about how much they love barbecue, they respect the cowboy aesthetics. and all these things. Cool. I didn't know that you were an otaku. And we need to figure out how and why over 50 % or something like that of Japan is like a weekly active user of X, which is just crazy. Yeah, they have great posts. Well, yeah. I mean, yeah, so let's pull up this first post.

52:48Wait, wait, wait. This is a little bit of an update, like narrative violation, because that's a narrative violation. That's a narrative violation. Because when Grok went viral, everyone was like, oh, it's good at anime. It's big in Japan. and it was at the top of the Japanese app store. But it appears that Japan is just using Twitter broadly. Elon. They just like the app, and that's where they have conversations, which is very cool. That's a narrative violation. So let's go to the first piece. Let's pull up this first post. This is hilarious. Recursive pizza? And this is the translation from Grok.

53:24When I saw this, quote, pizza topped with a pizza in America, I thought, there's no way we could beat these guys. This is this is an amazing I've never I've never seen there's so many layers. There's actually one two three four layers of pizza I'm gonna make this I feel like this would be a smash hit in my house This is this is this is quite this is a great smart performance peak performance. We got it. This may be for lunch today Let's get some pizza. And so this post Which in Japan? or I guess now everywhere got 93 ,000 likes is the translation from grok is I like this photo of American men and meat someday I'd like to join in on this in person there's just some guys cooking a whole bunch of steaks that's a lot of meat wow that's a lot of food they're they're having a big barbecue in in Sasebo's dining establishment someone else someone else somebody's I guess an American is posting their grocery haul, and someone says the amount is way too much, as expected.

54:34As expected. We have a brand over here in America. We do things this particular way. Hello, Japan. We love your fascination with our barbecue. Here is me buying half a cow's worth of meat for our family. We store it in a big freezer in our garage. I actually have heard about this. Buying in bulk, obviously, is more economical, but uh hilarious ratio by uh doctor something or other uh dr nicholas the amount is way too much as expected what else is going on japan take me uh in someone else says in sasebo's dining establishment it's common to spot u.s military personnel enjoying their meals with lively enthusiasm one day at a restaurant i came across a group that reached an oddly intense level of excitement just upon seeing bacon.

55:19That's incredible. I love it. Let me tell you about Lambda. Lambda is the super intelligence cloud building AI supercomputers for training and inference that scale from one GPU to hundreds of thousands. And let me also tell you about 11 labs, build intelligent real-time conversational agents, reimagine human technology interaction with 11 labs. In fundraising news, physical intelligence is in talks to raise$1 billion at$11 billion I need to know why is Jeff Bezos here besides the fact that he looks fantastic in the tux? He might put in some money. Oh, no, no. The company has previously raised more than$1 billion in capital from investors, including Jeff Bezos and Alphabet's independent growth fund, Capital G.

56:02So you could have put Peter Thiel because Founders Fund's in. You could have put Lightspeed. Is that Danny Reiner? You could have put Carroll or Lockie. Yeah, Lockie. But Bezos in a tuxedo seems to get the viral attention. But very good news. We actually interviewed both the co-founders of Physical Intelligence, both Lockie and Carol, this last year. And they don't do a lot of media, so it's an interesting little segment. We spent maybe 20 minutes with them. And you should go back and listen to it because it's a very interesting insight into the business that they're building, which I think a lot of people, you know, they're not a noisy firm.

56:44They're not a noisy company that's like posting vibe reels and going and picking fights all the time. So there isn't that much coverage of physical intelligence. But like if you just look at the traction, look at the open source contributions, the data, the fundraising, like clearly something is happening there. And so I think it's worth digging in and paying attention to. Last night, Bill Ackman hit the timeline. Whoa. He said, some of the highest quality businesses in the world are trading at extremely cheap prices. Ignore the mainstream media. One of the most one-sided wars in history that will end well for the U.S.

57:19and the world. And we have potential for a large peace dividend. One of the best times in a long time to buy quality. Ignore the bears. And he says, and Fannie Mae and Freddie are stupidly cheap. Asymmetry at its best. They could be at 10X, and it could happen soon. And, of course, GRTickets comes in and says, X.com, the market manipulation app. that Fannie Mae and Freddie Mac are up 42 % and 37 % as of this morning. I think they've actually dipped back down a little bit. But Justin says, posting your opinion on a public website is not market manipulation. JT says, don't ruin the tweet. Yeah, it's not market manipulation.

57:59It doesn't seem like he has any inside information. I don't know. Does he even have a position? Isn't that disclosed in his filings? I'm not exactly sure. I would take every recommendation from a Twitter poster, every piece of financial advice with a grain of salt. But this one certainly turned out to be some sort of pump going on. And I did dig into this. Somebody asked Grok, like, hey, break it down. Like, what is actually going on here with Fannie and Freddie? They generate$25 billion in stable annual net income from guaranteed fees, low credit losses outside crises. they're still in 2008 conservatorship, and the stock trades for a total market cap of$10 billion.

58:44So there's a world where you're sort of buying maybe, I don't know exactly how aggregated this is, but maybe it's like$25 billion of cash flow at some point for$10 billion. That feels like a very good deal, get paid back in four months, five months. But of course, there are a whole of other political insurgents. And of course, he does own. Fannie Mae and Freddie Mac are in his Pershing Square portfolio. But again, not illegal to share your opinion. Yeah. Well, there are some, not everything is up. Mike Zuccardi shares the current MAG 7 plus drawdowns from 52-week highs. NVIDIA is down 21%. Google's down 22%, Microsoft's down 36%, Apple's down 14%, Amazon, 23%, Meta, 34%, Tesla, 28%, and many others have drawn down significantly.

59:43Fortunately, we have the perfect person to ask about what's going on with NVIDIA because we have Take Him in the Restream waiting room. Before we bring him in, let me tell you about Console because Console builds AI agents that automate 70 % of IT, HR, and finance support. giving employees instant resolution for access requests and password resets. And let me also tell you about Vanta, Automate Compliance and Security, because Vanta is the leading AI trust management platform. And without further ado, let's bring Tay Kim in to the TPP and Ultradome. Tay Kim, how are you doing? Thank you so much for taking the time to come chat with us.

1:00:22And congratulations on the incredible launch of your business. Thank you. It's been really gratifying. That first day, you never know who's going to show up. Totally. I was thinking maybe 15 subscribers or 20 subscribers, but like hundreds of people showed up. Tons of billionaires and tech founders. It's insanely gratifying. Yeah, it's great. Incredible. So is it over for NVIDIA? They're down 21%. We just read since the 52-week high. Is it doom and gloom? Is it over? No. I think I was on last December, and the stock is, semis and chips have gone up, and now they're back down to where they were in December.

1:01:05The chip sector's flat on the year, and NVIDIA's down 10%. And it reminds me a lot of about a year ago. Do you guys remember, everyone's freaking out about DeepSeq, the super efficient models. We're going to destroy AI compute. There'll be a huge compute glut. And then everyone freaked out about Trump's tariff wars, preparation day. And this year seems very similar to that. Almost it's like Groundhog Day. We have fears over AI CapEx. People think that it might be the peak. And then we have the Iraq war. And one of these things is oil up here. Iran. Easy to get them mixed up. They happen so much.

1:01:45Feels like the same thing over again. Yeah. But someone said - Sorry to distract. We wanted to throw - We wanted to - We wanted to show respect. We wanted to show respect. That's true. To a real podcaster. I mean, it's very similar to Iraq. I love it. These are great. But in$100 oil, this stuff is unsustainable and probably subside. Okay. So because I like the deep seek analogy, and I feel like the market half digested the agentic coding narrative and the Citrini article, whether you thought it went too far or was too hypothetical, like, clearly the markets did react. and a lot of names sold off.

1:02:26But in a world where you believe that narrative, you would think that NVIDIA would be going up. But you're saying that there are other factors at play that are sort of tamping down the excitement in the market broadly? I mean, there's no doubt. Just like tariffs a year ago, NVIDIA had 30 % drawdown when their business was actually flying, the actual final of the business. I think the same thing is happening here with the Iran war. Things will eventually subside. Oil can't be$100 forever. And Trump will probably backpedal in the next few weeks ahead of the Trump. So let's recap a few of the key stories around NVIDIA.

1:03:03We just came off of GTC, and there's a lot going on at the company. I mean, it's a huge company. Maybe it'd be good to start with just next generation chips, changes to strategy, what people are actually buying. Maybe that means Grace CPU standalone sales or the development with the Grok partnership. What's sticking out just on the actual AI product side to you that you're most excited about? Well, inference demand is exploding, driven by the AI agents and coding assistants. I met with Ian Buck. I met with dozens of engineers at Meta, Google, NVIDIA. And all of them are seeing crazy inference demand and AI compute shortages.

1:03:48So across the board, people are in crazy clamoring need for AI. And we're, I mean, yeah, you're seeing that from talking to engineering leaders at big tech companies, but we're also seeing it from vibe coders who are just on X and Twitter and talking about how they're hitting rate limits and they're subsidized and they have multiple plans and they actually shift around from one model provider to another just to make sure that they're getting the tokens they need to build whatever they're building. And you see the tweets. People are building bots to pick up any kind of B200 GPU that can. They're waiting weeks and months or whatever.

1:04:26Basically like sneaker bots, but for NeoClouds. That's crazy. Exactly. I can't believe that. And the great thing is Jensen, he's very prescient. He probably saw this demand months away. He locked up all the supply agreements for memory, co-hosts, connectors ahead of time. He saw this inference demand. And to take advantage of this coding system boom, it's almost like a gold rush. You see open AI pivoting toward it. Anthropic obviously is thriving on it. Billions of ARR every few weeks. And Jensen acquired Grok, acquired the assets of Grok and the people of Grok. And the combination of integrating Grok's technology together with Vera Rubin lets NVIDIA serve this tremendous wave of compute demand economically.

1:05:17And Ian Buck talked about it. Jensen talked about it. So NVIDIA is positioned perfectly to thrive on this coding agent wave that we're seeing right now. On the Grok deal, Jensen did a fantastic interview with Ben Thompson and was sort of asked the same question two years in a row about ASICs, the threat of ASICs, the idea that the GPU, the general architectures can truly satisfy 100 % of demand. It feels like there's a shift in NVIDIA's strategy there. Do you see that? It feels like the right move, but do you see it as a shift in the philosophy of the company or the strategy, or is this just something that the gears have been turning for a long time, and this is maybe just an unveiling of a strategy that makes a lot of sense and has made a lot of sense for a while.

1:06:08I think what Jensen does, he sees where the market is shifting and where the economic value is. With Mellanox, he did this in 2019. He saw the world shifting to, it's a networking chip, but he saw the world shifting to these 10 ,000, 100 ,000 GPU clusters and Mellanox on the need for that. In the same manner, he saw AI agents and the inference behind that taking off. And he said, oh, this Grok thing will work perfectly with Verirubim. It doesn't replace everything. It just says talk about 25 % of the inference demand would be, Grok would work on that. But them working together where 75 % of the inference is Vera Rubin, 25 % is a Grok low latency stuff.

1:06:51It's like the perfect combination to take advantage of this. And the other thing is we're just in this great liftoff of AI innovation. We've talked about Anthropic Mythos, the blog post that leaked out. So we're going to have this step-up function. They told Fortune there's going to be a huge step-up change. OpenAI is coming out with their model soon. And then when I went to GTC, the biggest takeaway I had was this session between Jeff Dean and Bill Daly, both chief scientists of Google and NVIDIA. And it's online. I highly recommend people watch it. And he talked about, Jeff Dean talked about the context, have context window innovations where they could focus on the 10 ,000 documents that work well with your request and query.

1:07:37So we're going to have this context window innovation. Both chief scientists talked about stacking memory right on top of the GPU or TPU. And that's going to be a huge innovation in the coming months or years. And then Jeff Dean talked about synthetic data for audio and video. There's this huge runway that data is not over, and then they're going to be able to take advantage of all this data that people don't realize yet. So you have all these vectors where AI models, you can just keep getting better and better. Yeah. How are you processing the idea that NVIDIA will be investing in an open source frontier lab capability?

1:08:18that feels like potentially competitive with some customers. NVIDIA has like never really been in that market before. But at the same time, I've been the biggest like supporter of open source American AI models. I loved when Meta was doing it. I want more of it. I loved when OpenAI, Open Source, GPT, OSS. It feels really, really important, really great. But it does feel like a strategic shift. How did you process that announcement? It's not a huge, I think it's like 25 billion over the next few years, which doesn't really compete with what OpenAI and Anthropoc are doing. But these smaller models are going to be helpful for people running smaller use cases.

1:09:01So GPUs, as long as they're utilized even locally or in the cloud, NVIDIA benefits. And saw the top people at Quinn left and we don't know where they left to. Quinn is an amazing model. It's kind of like what Deep Deep is, what people thought Deep Deep should be. Quen works well locally. Quen kind of subsides because all the people left. Wait, what's your theory on where they all went? Another Chinese lab or international? I asked all the engineers when I was at GTC. No one really knew. But people are trying to say NVIDIA should actually hire them. Because the more capable open source model, NVIDIA doesn't care if you're using GTC to run open source or not.

1:09:44They just want more AI adoption across the board. Yeah, and NVIDIA has probably more levers to pull if it turns into a negotiation with China. We're tracking the Manus story with Meta, and there isn't that much that Meta can give to China in exchange. If there's like, hey, look the other way on this particular deal. Let this one flow through. We'll trade this. Meta not really doing any business there, but NVIDIA, of course, is going to be selling Blackwells at some point in the near future. And there's probably some level of pricing. You know, it can be part of a larger discussion, which makes a lot of sense.

1:10:23And one thing that kind of went under the radar, Jensen literally said at GTC they got license approvals on both the U.S. and China side. So we're going to see billions of dollars of H200 orders. Okay. So, yeah, I mean, it seems like there's a path on the demand side that's very, very clear. You've mapped it out a few times. It's a huge number. It's already massive revenues, just an incredible growth. But what is the supply side looking like? Because it feels like TSMC is not ramping CapEx nearly fast enough over the next few years. And if we see another 10x increase in compute demand, we could be really constrained on the leading edge fab side.

1:11:07So how do you think NVIDIA is going to process that? Well, NVIDIA is in the driver's seat because Jensen goes there five, six times a year and is best friends at TSMC and speaks at their employee day. So they're going to get higher. They are getting a higher allocation to wafers and co-wass and all that stuff. So NVIDIA will benefit. But I agree with you that industry-wide, Google is dying to get more TPU wafer capacity. All the hyperscalers that have ASICs are trying to get more wafer capacity. So there is going to be an AI compute shortage in the years to come, just like you said. And NVIDIA just benefits because they're the biggest dog in the house, and they can prepay tens of billions of dollars to get the allocations they need.

1:11:52Yeah. I mean, maybe there's some offtake in ASICs that can potentially be fabbed somewhere else at some point. I know that a lot of the ASIC companies wind up fabbing at TSMC, but it feels like if you're already doing some sort of re-architecture, maybe there's a way that you can squeeze something a little bit out of an Intel deal or something else. I'm not exactly sure. It's Samsung and Intel. Samsung and Intel, yeah. They're fabs that can possibly do it. Yeah. That's the bull case on Intel. Yes. Yeah. Is that at some point, the labs and Google, like, we're across TPU, extra GPU capacity, NVIDIA, the new R.

1:12:33Like, there's just so many buyers of lab capacity, of fab capacity now that you could imagine everyone coming to the table, potentially in Washington, D.C. or Mar-a-Lago, since the U.S. government owns a slice now. and everyone's saying, okay, let's hold hands and jump across this and say that if the supply comes online, we will buy it at this price because we have really, really solid use cases that will justify the investment for us and for Intel. So that would be a really, really good case. But again, even if the money is there, how long does it take to get to good production numbers? I mean, I suspect like Apple and NVIDIA are considering either Intel or Samsung for their lower end stuff.

1:13:18Whether it be like a mid-range iPhone or NVIDIA side, definitely their consumer gaming GPUs. They may go back to Samsung and maybe even Intel. Yeah, I have one more, but go for it. I wanted to know how you're processing the ARM CPU announcement. It's an interesting dynamic because they're sort of frenemies with NVIDIA now. They're competing in many ways is to break the x86 monopoly because they both are selling ARM CPUs, but then they're also competing. And so I'm wondering how you think that plays out, what that means for NVIDIA and just the rest of the semiconductor supply chain. I think ARM is, their CPU opportunities, a longer term, even they said 2030, 2031.

1:14:05It's a longer term opportunity. I don't really expect the major hyperscalers like Amazon to switch to ARM's product offering. They have their own. And same with NVIDIA. They have their own ARM CPU that they're going to incorporate and sell. So it's not that big of a... I don't think Amazon or NVIDIA are really worried that ARM is going to take any big share. It's probably going to be on the margin for companies that can't develop their own ARM CPU, the more the mid-tier hyperscalers or enterprises that use these things. But I think the ARM thing is very important because it kind of confirms what the biggest underlying thing that's not really consensus yet is this massive CPU shortage that we're seeing.

1:14:50Just over the last few months, we have Dell, AMD, Intel, CFO talked about, they're talking about three to five year locked in supply contracts and hyperscalers. So this is a major trend that's going to go over the next few years. And the reason why is AI agents need more CPUs. The ARM CEO talked about four times more CPU cores versus last year's kind of AI infrastructure model. So we're going to see this massive demand for CPUs that people aren't really understanding yet. Because AI agents, the whole thing, requires orchestration, tool calls, database queries, web searches. And that's all handled by the CPU.

1:15:37Give me your bull and bear case for TerraFab. TerraFab, I'm not that optimistic. I mean, it's so hard to build that. Do your absolute best to give me the bull case. Because TSMC is so short that Elon needs to find. But even then, how are they going to buy semi-cap equipment from ESML and AMAT? Like there's just no capacity there. So I'm not optimistic on that. And this is stuff that takes decades. Chip fabs is almost like cooking. And it's not like something you could just follow a manual. It's almost like cooking where it takes a lot of trial and error accumulated over decades, TSMC and even Intel.

1:16:31So it's not something you could just jump right in and do. Yeah, it's someone goes back to the XAI debate about like, do they need AI researchers or should everyone be an AI engineer? Like, are we in a research period or a, you know, the Ilya Sutsukov age of research versus the Elon Musk age of engineering? Where are we in semiconductor production? It feels very engineering, like an engineering process. But what we've seen from ASML and TSMC is that it does feel like there's a little bit of research and artistry to it. And the cooking analogy holds. Yeah, I've been doing a lot of research into space.

1:17:14And it's a lot of trial and error. Yeah. Almost like cooking a recipe. And it also feels like, at least with XAI, if all the researchers are in San Francisco, you can sort of just like walk across to the coffee shop, poach someone. But if the best, you know, semiconductor engineers or technicians are in Taiwan and they see it as a national, you know, urgency to, you know, bring, you know, stability to the country, both economically and geopolitically, then you have a very different calculation. It's like, oh, yeah, I could make five times as much if I left my home country to, like, be abandoned.

1:17:54That's a very different calculation. And everything that I've heard about the culture at TSMC is that the folks who work there are extremely dedicated beyond the economics. They are true missionaries, not necessarily mercenaries. And so it does feel like it's even harder to do like a talent raid in the leading edge fab world than even the AI world, which is extremely competitive. And there are still tons of missionaries. I guess another question I have is, would you expect XAI slash SpaceX at any point to get to basically just open up a shop as like a NeoCloud? Because the thing that was like probably one of the least compelling aspects of the TerraFab pitch was him just saying, we need all of this compute.

1:18:42We need to do this because we're going to be so chip constrained. We're going to be so supply constrained. But there was no explanation of. where the demand was going to come from. Is it going to come from Optimus or Croc or Twitter? Yeah, it was just very unclear. There's a lot. But there's even the question right now is should XAI be kind of renting GPUs? I don't know. Renting out GPUs? Because the biggest win has been Colossus 2. Yeah, Colossus 2, which was built very fast. I think Elon's pitch with the SpaceX IPO, and we'll see it in the coming months, is the AI compute. There's going to be so much demand over the next 5, 10 years that you're going to have to use these SpaceX satellites that have GPUs in them to serve that today.

1:19:34And maybe, I mean, even though Tesla's been vertically integrated to the point of being a consumer product, SpaceX has not. It's been a railroad, and there is a world where you fab the chips, you put them on satellites, on Starlinks, in space, and then you let other companies do whatever they want with those GPUs. Think what Elon did with Starlink. That's a telecom infrastructure play, and this will be an AI computing infrastructure play. Yeah, yeah, yeah. That fits that model. There's a world there. I'm not going to bet against Elon. It might just take long. Yeah, yeah. What's going on with Helium?

1:20:05What are you tracking there? There's chatter about Helium shortages potentially. Jensen has talked about this. this is a risk but there is probably like six months six to nine months of inventory in the channel bernstein has talked about it's not a risk in the short term so so if this thing if this iran stuff lasts yeah in you know two three four five months then becomes a problem okay but if it you know gets solved or straighter opens up with the toll or whatever uh final negotiation they come up with over the next few weeks, I don't think it's going to be a problem. Yeah. I do think that like most of these materials, there are extra deposits.

1:20:47They're just not economical to mine. I don't think that all the helium exists in the Middle East. That would be very odd. It's similar to the rare, rare thing, just like you said. Yeah, where in a supply-constrained scenario, it becomes more economical to mine American helium. Let me put it this way. If helium becomes an issue, we're going to have bigger problems in our hands. Okay. I mean, there's going to be world starvation. Let's hope not. Let's hope not. That will be the least of our problems if Elon becomes the problem. Take me through depreciation gate. How did you process that and where do we stand now with the fear that GPUs will depreciate precipitously and H100s will be worthless in 6 to 12 months?

1:21:26It's totally not a problem right now. Core Weave has talked about these things are lasting 5 to 6 years and they're getting almost 90, 95 % of the pricing. so it could potentially be a problem if this is a bubble I don't think it's a bubble but if it's a bubble two, three years from now and there's a compute glut then the stocks are going to go down because there's a compute glut but as of now it's the opposite all the GPU rental prices even for stuff that's six years old is still being sold out and the AI compute demand outpacing supply is so large that this is not an issue right now Do you have any theories on where the next step change in token demand could come from?

1:22:09Because right now we're seeing it in CodeGen and there's a lot of optimism around these types of workflows being applied to other forms of work. But we were talking about this on Friday. Like even if AI can just one-shot beautiful financial models, it won't necessarily even make a real dent in token demand, at least compared to to code gen because no company needs to just constantly be, you know, be generating models at the rate that let's say Gary Tan generates code. And, and so I'm, I'm like kind of been trying to wrap my head around where, where could these incremental use cases. I actually think code gen is still just early innings.

1:22:55Yeah. And I don't disagree with that. 10, 20 agents and they're kind of over overseeing them. But then we have this other stuff where these models, the mythos and open AI, they're just going to get better where you could automate all these work process flows. Companies are going to use them for every single vertical customer service research, simulating chip design where they can verify drug discovery, where they verify drug molecules. So we're just getting started at this stuff. So you're going to see vertical AI agents on every single category. And I think Logan's coming on. He wrote this great post on X that he says the AI agent wave is going to kind of attack this$6 trillion knowledge economy, right?

1:23:44It's not just about programming anymore. They're coming for us. Yes. I don't think... I'm actually... They're attacking the key context economy and the TBPN economy. No, I think it's like a calculator, a spreadsheet. You know, 30, 40, 50 years ago, we had like 50 accountants doing the spreadsheet manually, right? And now after a spreadsheet came, it didn't get rid of all of knowledge work. It just enabled people to think at a higher level and get more done. And I'm very optimistic about that. I mean, one way that you 10x token demand around a financial model without 10xing the number of financial models that you're building is having the agent go and collect 10 times as much data.

1:24:30And so there's a lot of situations where, I mean, you look at like hedge funds that want to understand the price of Walmart stock. There are hedge funds that will task satellites to take pictures of Walmart parking lots, estimate the number of people on a day-by-day basis that are going into the Walmart to shop and then using that as a proxy to project revenue and then flow that through to cash flow and then flow that through to the DCF and the actual evaluation of the company. And if you think about all the different financial models and all the different businesses where you could go and say, well for this company I need to go to every single local like I want to know the price of Squarespace let me go to every single website that's powered by Squarespace and estimate the revenue that they're bringing in and their willingness to pay for their hosting service something like that and all of a sudden like it's just one spreadsheet it's just one number at the end of the day but it's like a thousand times more work went into it let me give you this great example.

1:25:31Every year I do this same store sales for these fast casual companies, so like Chipotle, Cava, and I put out this tweet. It goes viral. A year ago when I do it, I would have to manually go to every IR website for these six fast casual restaurants. It would take me like an hour or two. I would try to use a chatbot. They would get it wrong. I did it like a few weeks ago, and all the chatbots got perfect. So it just saved me two, three hours of tedious manual labor. So that's only going to get better and better. Yeah. Yeah. And it's only going to take you one. Like this year is the year that you do it with multiple chatbots and you fact check it yourself.

1:26:13And then forever, it's going to be just one prompt. And it got it right. And it got it right. A year ago, it wouldn't get it right. But now in one, two minutes, I put, give me the same store sales for these six restaurants. I put in Gemini. I put in chat GPT and just to make sure they're right and they're right. So that all, all the tedious labor, all the manual labor, all the data entry that, you know, all of us are used to, um, that stuff is going away and we could think higher level. So I could look at the same store sales and say, Oh, the economy is at risk and whatever, but all the grunt work, all the tedious work is going to be taken care of, uh, by these, uh, AI agents.

1:26:52I can agree completely. I agree completely. Uh, We got a lot more sound effects since the last time you joined. Last question for me. What's your outlook on Meta? It feels like the broader market right now has zero faith in Meta to actually put all their AI investments to use. I have this history with Meta is that every time it starts falling apart, I say it looks cheap and then it goes down another 30%. but nothing has changed. Like no one's going to replace that digital ad position. I mean, like I would even say in the AI world, they're even better position because Google might lose digital ads share to AI chatbots, their search position in the future.

1:27:40So like no one's going to replace Instagram. No one's going to place Facebook. Billions of people are still going to use those social media apps. And you know, it's every six, six months to 12 months, everyone goes through this bare meta cycle, but their pure competitive position really hasn't changed. And you saw what happened to Sora, right? Like, you know, everyone's all excited about Sora and that got shut. Totally. Yeah. And there's just this world where even if like the AI spending is like a side quest, it's like really, they just pulled forward like three or four years of CapEx and they will use that for their other products.

1:28:16It's probably even less like wasteful than reality lab spend, which might take even longer to realize that the cash flows from, like they can recoup, okay, we built this massive data center. We did this training run. We didn't get to the frontier. We're not getting a lot of like gen AI usage, but we can apply it to our ads platform and tools and reels recommendations and a million other things just in years, 2028, 2029. And yeah, we're a little bit ahead of schedule. Our core ad engine monetization. 100 % yeah the gem model Reality Labs he made a waste of 70 to 80 billion dollars he might waste 100 billions of dollars on these frontier AI models but their core ad engine core business that money making engine is not going to be affected by this yeah well thank you so much for taking the time to come hang out always a great time Tay go subscribe to Key Context on Substack Follow Take Him on social media, First Adopter.

1:29:19Join the many beaners that were the first adopters. Yes, yes. You'll be in good company. And thank you so much. We'll talk to you soon. Have a great week. Great to see you. Cheers. Let me tell you about Figma. Agents, meet the canvas. Your AI agents can now create, modify, create and modify your Figma files with design system context in beta starting today. And let me tell you about Graphite. Code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. so uh chamath yes holly hafia says the biggest threat to instagram's moat is an incredible image model okay zephyr says meta bottom um an incredible image model it should i mean that's basically that's a that's a like you're basically saying okay if sora was if the content on sale was better a hundred times better yeah would that be a real threat to instagram and i still am not I'm still not convinced.

1:30:14I feel like a lot of people have their network there. They want to share with their friends. They have a graph there. And even though the recommendation, like the content doesn't come through the graph anymore, having your friends on there to have the conversations and the comments, there's still a lot of moat left. But if they could make an AI agent of you that instantly reacts to every video I send you. Killer feature. Killer feature. Killer feature. There's some, yeah, between our DMs, there's a lot of stuff that you've got to still react to. Well, without further ado, we have Logan Bartlett from Redpoint, his managing partner there.

1:30:49Welcome to the show, Logan. How are you doing? Good, gentlemen. How are you? We are fantastic. It's great to see you. I like this camera setup. This looks fantastic. You know, once upon a time, I was a semi-professional podcaster before you guys stole all the thunder in the industry and forced us into oblivion. Three cartoon avatars was truly goaded. Back to market update, Dex. Just put the investments in the bag, bro. I know. That's exactly right. Yeah, the McDonald's bag of cash I have. That's what I'm doing these days. But you're also writing market analysis, which I always look forward to.

1:31:26Yeah, this has been consistently some of the best content in the entire VC industrial complex. And I've enjoyed it for many years. It was extremely valuable during the interest rate crisis as well. And also the conversations that you were having on the podcast, but it felt like a really rational reset that wasn't a total black pill, wasn't a total white pill. It was just actually like, here's some data. There are obviously some conclusions, but you can also make your own. So thank you for everything you do. Take us through the biggest findings. Take us through the process that led to this particular research.

1:32:01Yeah. It turns out there's a little bit of nuance that 75 slides give you more than 140 or 280 characters to kind of tease out in some ways. So it started probably in January. I have this process every year where I have a panic attack that we have an annual meeting coming up. And I got tricked in 2020 when I joined the firm. They were like, we're going to give you this really illustrious honor that you get to do the market update. We so trust you and what you have to say. And I thought that, oh my gosh, this is amazing. I'm being bestowed this honor of doing this market update deck. Little did I know no one else wanted to do it.

1:32:44And so every January, I get a mild anxiety attack that I have this coming up. And over the past couple of years, there's been a bunch of different, I mean, that year it was COVID, then it was kind of the Zerp fallout, Zerp era 2021, then Zerp fallout. Then 23, I think was SVB. 24, maybe I got like a little bit of a respite. Then last year was the tariffs. And so every year there's like something going on that forces us to recalibrate. But this year it became pretty clear it was going to be the software sell-off and what was going on in the public markets. And so monitoring that, I sort of started from a process of talking to a bunch of smart friends in the industry about what they're thinking about and trying to probe on questions that they wanted answered.

1:33:33And this generally involves a lot of public investors because private investors in some ways are like fish in water, where like, you sort of just operate in the world around you. And so if you're doing defense, you really just focus on defense. If you're doing software, you just focus on that. If you're doing whatever, healthcare, you're focusing on that. Public market investors, I find, are a little bit more zoomed out. And they typically have an opportunity to play across different scale of businesses, different sectors, different, you know, types of companies, all that stuff. And so I talked to a bunch of them and, and software and like, what the hell's going on was the big narrative.

1:34:09And there felt like there was a major disconnect between what private folks were seeing going on and what public folks were thinking about. And so trying to bridge that gap of how do we have this world where, you know, software companies are now trading at 4.1 times NTM in the public markets, but also getting priced at two, three, 400 times ARR in the private markets. And so sort of setting out to bridge that gap was kind of the goal. Is it a, is it a gap or is it a golf? I would say it's a optimism disconnect maybe. That's a good phrase. I like that. It is amazing. I did a panel recently with a bunch of private equity investors, and in hearing them talk, what I concluded, and some of this is true, I think, for public investors as well, is what is the risk of going to zero and optimizing your process around, hey, we really can't have a zero X in the portfolio versus private market investors, you're optimizing on like, what are the chances you're missing out on a 30, 50, 100 X?

1:35:16And if you take those two lens, it ends up with a very different place of like optimism versus cynicism, upside versus downside, you know, all the questions you ask. Yeah. So let's, let's start. You'll appreciate this, Logan. I had a portfolio company at the end of last year that is a software as a service business. and in one of their updates they they made the announcement that workflows are now called agents in in the product and i was like they were like this workflow stuff seems like people are not that excited about it now we're we're switching gears these are now agents and if we just you know that breaking bad meme that like he says we had a good thing you stupid son of a bitch That was like all SaaS investors over the course of the last 100%.

1:36:04Being like, you MFers, you had to go mess up this really good thing we had going on with this AI pixie dust. With a nonprofit that didn't even raise a seed round until they were multi-billions. It's like we couldn't even get in early. We had a good time. We had a good time. So let's start with the public markets. How much of this is driven by the Citrini article? How much of this is driven by actual data points where we're seeing? I was just pulling like the top 50 SaaS companies, sort of pure play SaaS companies, and trying to answer the question like, is revenue decelerating yet? Are we seeing a kink in the graph, like some change in the data?

1:36:43And I didn't go nearly as deep as you go. But how much of this is just like narrative and anxiety about a changing curve to the financials versus actual data points where people are saying, okay, like we're not going to be growing as fast. We're not going to be as profitable as before something else that would change the valuation. Yeah. I mean, I think broad buckets, there's two main things. One is like the public market investors are fed up with stock-based comp. And so like, let's put that in a bucket. And I actually, I do think venture investors and public markets, CEOs are, um, to blame for some of the softness in the cultures and like how bloated some of these businesses got.

1:37:24But also you have to be practical. There is a game on the field to play. And like you could triage and say, hey, you know, we're really going to reduce the number of employees we have. And you really have to be careful there because, you know, they could your best people could just walk out the door if their friends are all getting fired and they can walk out the door and go work at Anthropic or Lagora or, you know, one of the businesses that's growing at this crazy, crazy rate and get stock based comp. And so I am sensitive to that. But that is a real part of it that like there's not true profits going on.

1:37:58And so I think let's put that in a bucket, though, that that's like sort of a side. The other thing, the far more interesting conversation to have is like, are financials deteriorating? And the answer is really no right now. It's more of this like long term existential question of what terminal value of these businesses are worth. And it used to be, hey, 85 to 90 percent of a business's value was tied up in the period beyond the DCF, right? The terminal value of the long term duration of it. And that's really what people are asking questions on. And to be honest, I think what's really happened is the public investors are saying, I can't tell the difference between Salesforce and ServiceNow and Snowflake and CrowdStrike and Guidewire and Samsara and all these businesses.

1:38:50And to be honest, I don't even really want to go dig in and figure out all the little specifics here. I'm just going to go put my bankroll in NVIDIA or Google or AMC or something else. And I'll wait for this to sort itself out, wait for the market to do its thing and figure out what the buying opportunities actually are when it's a little less uncertain. And so I think it's that. And people are asking, like what is the long-term terminal value and saying, I'll wait on the sidelines until other people really show the proof points that they're going to be able to survive this AI thing. Yeah. Is there a world where we move into a regime where we're talking about not revenue multiples, but like EBITDA multiples for these software companies?

1:39:32So I was looking at a company that was 3 billion market cap, 100 million of EBITDA, very stable the last five years. and one investor was making the case like, oh, AI winner. And I was like, I don't see that. But also I don't see these customers churning. I just see them doing AI stuff on top of this particular company because they're more infrastructure layer, more data storage, that type of thing. And so I was like, I think you can count on 100 million a year. EBITDA and probably cash flow for 10 years, 20 years. But do you want to be paying 30 times that? Is that enough? And I don't know if that's the rational framework.

1:40:10Nobody knows anything. You have to apply a big discount. One of the slides I loved was the slide on newspaper earnings. Oh, yeah. Yeah, take us through that. Slide 22, you say newspaper earnings. Yeah, I mean, it is interesting. It's funny. I actually have that up on my screen here as well. But yeah, newspaper earnings. I mean, when these platform chips happen, you might not see it in their earnings or revenue initially at all. And so the newspaper example in the deck was that newspaper earnings were actually fairly stable for like the five years post-internet while their value collapsed. And so everyone saw the writing on the wall of where this was headed.

1:40:47But it took a while for that to actually come through and show up in the in the actual financials themselves. So, John, I guess your question on it, like I I use revenue as a proxy and maybe it's like to to flip of a nomenclature. We really should be talking about like free cash flow with deductions for stock based cop or whatever. But like all these things are growing at different rates and that's sort of been the historical lingua franca that I kind of used. But you're right. To be clear, it makes it impossible to comp to the private markets because no one's generating any EBITDA. That's right.

1:41:23So it's a useless comp. But I'm just thinking like if I'm a public markets investor and I'm just and I'm choosing between Google, Apple and then some small cap, mid cap software company, I probably want to have an EBITDA hat on or something like it to sort of understand my just my rate of return, which is going to be a lot less like, oh, all of a sudden they're growing at some unpredictable rates. So the DCF gets crazy and I'm paying some high rate. Yeah. And this might be a little simple for some of your listeners and maybe helpful for others. But like at the end of the day, a business is valued at the current value of all future free cash flows.

1:42:00And so discounted back to today's dollars. And so when the reasons software businesses have been so good is the you have annuity streams going out into the future and you're able to, with some level of precision, figure out what the discount back the value as in the future. And so that's a that was a great thing, particularly when we had retention rates at, you know, 95, 96, 97 percent net retention rates at 121, 30, 140. You could really do very little and you could discount back those dollars with pretty good certainty of figuring out what those are worth today. It was almost bond like. And I think this equity made a bunch of money saying, like, actually, you know, this is this is better than a debt instrument.

1:42:43This actually sits on top of the debt in terms of your vendors are going to get paid before your debt providers will because the business needs to keep going. Now, I think we're seeing a little bit of cracks in the armor. And I think your analogy is a good one where it's actually not. I worry less about the like the churn risk. Are people really going to churn off of Salesforce or Workday or ServiceNow or whatever it is? Like maybe. But I worry less about that. I worry more about the value abstraction that is captured on top of it. And if the AI dollars, which we found in one of the reports, AI dollars this year, it's a bigger pie of net new dollar opportunities in AI than all of software combined by like 50 percent or something.

1:43:29And so if you're not capturing the AI dollars, then your growth rate is going to go to near zero. And if your growth rate goes to near zero, then it's worth something. But it's not worth you're right like the 30 times almost like a real estate investment It's like what's your cap rate? You know, like if I'm giving you a hundred bucks Am I getting five bucks this year ten bucks this year because there's a lot of other options and then yeah The other thing historically with with software has been just low interest rates So oh oh that cash flow is coming in 20 years fine Like it's basically the same as today if it's zero interest.

1:44:03Yeah, exactly But when you're at six percent, you know, you do discount it back and you get a lot lower number Anyway, where should we go next? I'm interested in... I'm curious, any of the public markets investors, you said a lot of them were just like, I don't want to try to be the smartest person in the room and lean in and figure everything out. It's safer to just like, you know, bet energy, bet semis, et cetera. Was anyone like licking their chops, being like, this is the greatest buying opportunity, like actually had some well-thought-out thesis around how, it was like Toma Bravo, some of their slides leaked from their LP, summit.

1:44:39And they obviously are in the position where like they have no choice, but they can't be bearish now. You know, they have to like, you know, create the like 4D chess of how this is like a huge accelerant to their businesses. But yeah, I think some of the public, some of the public guys, they are very interested in trying to discern what's going on. And this is actually a really good buying opportunity if you believe people are going to figure out the agentic opportunity or the AI opportunity, because it's certainly not being priced in in a material way. And the incumbent vendors are going to get every chance from their existing customers to get this right.

1:45:21And so I think that's the if you were to paint the optimistic lens about, you know, Toma got dragged a little bit for some of their, you know, talking their own book. But I actually think some of the slides that people were dunking on were, it's true fundamentally that like, hey, your incumbent vendors are going to get shot one, two, three and getting it right. I think the problem, and at least what we're seeing in the private markets, is that the culture of building these AI companies is just so different than the culture of building what the historical software a company looked like. And you guys, I think, I think, you know, I was an investor in, in ramp and, and, and the stuff that they did, like, it is my cross the bear.

1:46:11Logan in particular, he's, he's, he never takes victory laps. That's the thing. We'll take it for you. We'll take it for you. It's, I am, I'm unknown. I was a silent investor for a long time. And so I'm glad to come out of the closet as a ramp investor here for you guys. But, you know, one of the things that they did culturally for a long time that I thought was kind of crazy was they shipped a lot of stuff and would just put it out of the market and see how people react to it. And that was very different than the way that I learned, you know, the companies I invested in 2014, 15, 16 and how they built products was they had a very tight product roadmap.

1:46:54They communicated with their customers. You know, they had it over a three, six, 12 month period of time. And they would only really release it when it was fully ready out of initially an alpha, then a beta. Then they would take a GA with a handful of customers and take a bra. But the ramp guys sort of put that on its head where they would move really fast, iterate, get it in front of customers, ship it at like 90 percent readiness and then see how the market took to it. And if they if it resonated, then they would continue to build around it. And like that mindset is actually what I've seen with a lot of AI native companies now, which is like you're not totally sure what the model capabilities are going to be in three months time.

1:47:33And so what you need to do is internalize what your customers are going to want, like have enough of an appreciation for their job that you sort of know what workflows exist or like what existing pain points are. And then when the model capabilities keep getting better and better, you need to internalize what that customer is going to want and what the capabilities of the models are or where they're headed and sort of let those two things intersect and then deliver that to the customer. And so it's a very different way of like building product. And that's one example. And we have a slide in there of like all the different examples, but like it's sort of been flipped on its head.

1:48:10And so I actually don't worry from a is it possible standpoint for the for the big public companies to do this. I think it's totally possible. And I think some of them will figure it out. But the vast majority are going to have to totally change their culture that they built over the last 10, 15, 20 years. And that's really painful. And I think that's where they're going to end up falling down more than anything else. Yeah, this is fascinating. I'm like sort of an early-ish adopter, I think. And I recently wanted to know, like, how much have we spent on Apple products? And I was able to get that answer in like ramps AI mode, basically.

1:48:45and I didn't need to like export any data. But then I wanted to know how many, what I've spent with Apple over the last year on my personal financials. And for that, I had to vibe code something that exported all the data and did it manually. And so the question of like, you have a system of record, there's going to be some new feature. Where is that value going to be captured? Are you going to capture that value or is another system going to come down and it's going to be a feature of a chat bot or a feature of another platform? Like this is a tale as old as time. Yeah, it's abstracting the value on top of it, which is interesting.

1:49:19I mean, I guess if you guys think about like my direct visiting of websites has definitely gone down because I interface with Claude or ChattuBT in a meaningful way. And I think that same thing is going to play out within the enterprise as well. And it's not just going to be retrieval of information. It's going to be actually taking actions. And so now I don't totally care. I'm sure you didn't totally care if that information was coming from on ramp side, if it was by bill pay or credit card. And ultimately, once you five coded that application, you didn't care if that information ended up coming from a credit card statement or an email receipt or whatever it was.

1:49:55In fact, I wanted to unify credit card and checks and bank transactions as well. And I want to put all of that in one bucket. And that's something that's not a feature in my bank right now, but it will be if they move quickly. But it already was a year ago in Ramp. And so it's just like the pace of play is still on the order of years in a very interesting way. And yeah, definitely encourage all of those companies have opportunities, but they have to go win them. No one just gets granted monopoly on the new capabilities that emerge on top of their platform. Sorry, Jordy. In the deck, you have some bub talk talking about are we in a bubble.

1:50:40And with every advancement with coding agents and things like that, it seems like there's plenty of demand for tokens right now. People are willing to give real dollars for tokens, and that's just going up and up and up. and but I think there's a tendency right now at least for kind of the early stage private markets crew to say like AI is not a bubble so I should still be investing like tens of millions of dollars into all these different early stage companies and things like that and I've been like I've been kind of feeling the bubble in in private markets like just based on the number of companies coming out every single day yep that seems to all be doing kind of variations on like the, you know, the AI CMO, right?

1:51:22And I'm like, maybe, maybe that ends up being a big category, like a sort of niche vertical SaaS player that's like AI will be like at a 50 cap for a seed. Yeah. And I guess my, my, my point is like, we can, AI maybe isn't a bubble, but that does not mean we're not experiencing like a massive bubble in the kind of venture world right now. Yeah. I mean, it sort of goes to like where value is going to accrue. And like, if you, we did a slide on percentage of GDP in there. And if, if, if you were obviously investing in airlines, like it was a transformative technology that didn't end up proving to be a material investment opportunity.

1:52:01And what actually presented opportunity was the second derivative considerations of like business travel or like, you know, lounges and airports or whatever B2B sales and all that stuff. Like there were second derivative things that were actually far more impactful. And you're right. Like it's possible. And this is what I've told our LPs that I've asked is like, I think we're operating in a world in which our mortality rate of companies we invest in is going to be higher than it's been in the past. It just like it is even at the stage we're investing in. I think we're going to see a lot more businesses die.

1:52:36I hope we will also invest in things with a lot more upside. And so we'll end up with, you know, hopefully things that could be hundreds of billions of dollars, which used to be not in the realm of possibility. And so I do think we're entering this like extreme period of uncertainty. And the only thing I've really been able to come back to in all this is because you're right. These categories end up so crowded and they end up very dynamic in terms of like how the category evolves, what the product surface area ends up looking like, all that. And so in some ways, we're back to like investing in teams and investing in like the wedge or the general space that they're operating in and then hoping that those doors open or that that see parts and they're able to run through that in a meaningful way.

1:53:25But it's very possible that the model providers end up soaking up a ton of the equity value. And so just because there will be a CMO in the AI world that a company starts, it doesn't mean that any of these companies will be the one to capture that value. And actually, it would probably be very unlikely that it would. And so that individual investment, you might be very rational in doing it or not doing it. and the opportunity will ultimately create a ton of value, but it might not be a private early stage specialized company. That's going to be the one. Yeah. The yeah, it's been interesting to look at these businesses ramping revenue so quickly and still, and, and have like real customer love and pull from the market and still have that question in the back of your head of like, does this eventually just get zeroed out?

1:54:18You know? Yeah. And, and for me, that's the one people, sorry, go ahead. Yeah. for me, the, my, the only real comp I have, because I kind of came, I came kind of online in my career in 2018. So I got to see the, you know, Zerp era, uh, very closely, but I remember with OpenSea, you know, and the NFT boom, that was you, I remember the way that they ramped revenue, even the thing that made, you know, I think a lot of otherwise, you know, great funds like pile money into it, what ended up being the top, is there was like, you could kind of just say like, okay, even if revenue drops by like 90 % and this doesn't end up being like this mainstream opportunity, there's still like a business here and maybe you can just own the category.

1:55:06But then revenue ended up dropping like 99 % or something like, and I think that still stays in the back of my mind that that was more of a demand issue versus like new kind of competition from from an adjacent player uh but but still uh are there any previous booms that you do like as comparison points if not dot com do you like railroads it sounds like electricity yeah electricity yeah what do you like um that's a good question i haven't actually uh i haven't actually thought of the right analog for what time we're i mean people you know the industrial revolution is the one that people come back to the most and uh i that wasn't on our like gdp calculation chart uh but i do i think there's there's elements of like the shifting balance of of uh of uh worker dynamic and like where people are actually gonna totally the leverage employ themselves in that way going forward and like wealth, you know, there's a lot of considerations on wealth capture and what percentage of the population that's going to go to.

1:56:14And there's definitely a lot of like populist rhetoric out there. And so I think this is more of a, I think, revolution than like a technical, I mean, it's both a revolution and a technological shift in some ways. And so I think like the car or the airplane or the railroad or whatever, like that That didn't fundamentally shift the balance of an entire workforce in some ways, the way that I think this has the chance of doing. That's the one I kind of come back to, but it's a good question. I'll think more about it. How are you thinking about capability overhang, diffusion, the - The copability overhang.

1:56:54Yeah, the debate about the models are good and they're getting better really, really quickly, but there's just like, you know, teams and companies where they're like, yeah, I'm actually fine doing my spreadsheet job. And, you know, I, yeah, I got it. I've heard this direct quote. I got to check that AI thing out. Oh yeah. I got to check that out. And when did you already say that to you? Is that recent? I think, you know, it is interesting. I mean, that's where you're seeing a lot of this like FDE, Palantir era where bridging the last mile is really, really hard. And I think we assume that, that like, if we build it, they will come in some ways, but it's obviously, that's not the case.

1:57:40AI to most people, I guarantee if you took whatever, I don't know, 350 or 300 million Americans or something, and you ask them like, name an AI company. I would guess I'm making this up, but like 25 % wouldn't actually be able to name an AI company. And like 70 % would say, Oh, that's that chat GPT thing or something. I talked with a guy, I talked with a guy and said, like, what, what are you using any AI products? And he was like, Nope. And then I was like, what about chat GPT? And he's like, I use that every day. I love it. But he just doesn't think about it. It's just a website. Like we've been to websites before.

1:58:17And I think that's an interesting thing. You know, Brett Taylor talks about this from Sierra where there's so many capabilities that you're raising the waterline of and that they're needing to build in house themselves, knowing ultimately the model providers are going to need to productize that or going to productize that. And so they end up building things that they throw out six months later all the time. And I think, I think that's kind of true on the go to market or like education side as well, where like a lot of these customers, if you're going into an industrial business or a healthcare business or an energy company or whatever it is, like, you're having to bridge the capability to competency, uh, and like bridge that gap to the individual person at the end of the day.

1:58:57And so I do think this diffusion, when everyone talks about like, are we in a bubble structurally at a big picture? I don't, I sort of reject that notion because of both the demand. And when people talk about the power supply and all that, I actually think it's going to be it's going to take far longer to get this out into society in a really meaningful way than people on the internet tend to think because the real world's a lot more complicated than i think we make it out to be when we're just you know living in our techno utopia it's a good point if you can call x a techno the techno nightmare except in japan and japan i did want to ask about buy versus build economics you talked about how you could just buy that's the one that people have been asking all about, by the way, today.

1:59:41Yeah, people like that one. So Logan makes a point. You can buy Slack for a thousand employees for like a quarter million a year, or you could build it in-house. You estimate it around 2 million a year and then like other kind of random unexpected costs. I'm sure a lot of people, anybody that's pushing back on this, just tell them like, okay, build me Slack. Build me. It's a really funny thing where I just think it's sort of the 80-20 rule in some ways that people assume building a software product is like the getting to the proof of concept or like the credible mvp in some ways look you can send messages you can create a group and then it's like oh i vibe coded this thing and it does all of what slack needs to do but then there's not even to mention the network effect of slack of you build the perfect clone and then it's like okay do any other companies use it no okay we You still need Slack.

2:00:35Yeah. And so, so like, let's say, I think, I mean, people want to argue about the specific math on all this, but like, let's say that you're willing to do all the integrations in the SSO and the search and the file sharing and the, you know, whatever the admin controls and compliance and all that stuff. Let's say you do that. Yeah. The emojis, the GIF embeds, all, all those things. Like, let's say you do all that. Was that, whatever that costs, like, what is the opportunity cost that you spent all this time doing that rather than like focusing on whatever it is your core business is. And so actually like I, we did the math here and it's 2 million versus 220 K or whatever.

2:01:13Like, let's say it's the same or let's say it's cheaper. Like is, is saving, you know, 40 grand, let's say it's 180 versus 220. And it actually has to be significantly cheaper. But, but also, uh, I mean, I talked to a friend who runs a company and, uh, just about AI stuff. And I was like, oh yeah, like you should probably you know be aware of this stuff but uh what percent of revenue is going towards like software broadly like what's your it spending is like less than one percent and so it's like yes like you could take something that costs a thousand dollars down to two hundred dollars like maybe you take that but not if it's a headache at all because 99 of the time you want to be with your actual customer suppliers because it's completely different business and so that was the point someone was arguing me about is like most companies actually you know aren't growing like software or tech companies are.

2:02:02And so these costs are really material to them. And I'm like, you know what's material to them is like decreasing their workforce turnover from like 70 % a year to like 60 % a year and not having to pay incremental recruiter or staffer fees to get people on. The difference of the Slack budget and saving 40K, I guarantee does not like resonate at all. And Slack was a simple example because it resonates with people. But I think it's true across the board. So I was trying to think of like what a good bet would be with someone to try to like come up with it's a very hard thing to figure out of like what the right framework of of thinking about this is because i love just codifying bets with people and being like you know okay let's let's wager some money on what this is and i couldn't come up with a good one so if you or if anyone listening can come up with a good bet on this i would love to place whatever a significant sum of money on the side of it it's funny it's funny to think about the company building slack in-house and they're like throwing time they're getting 10 people on a call like hey like we need to meet and talk about some up we need to talk about like our roadmap for our internal slack we need to kind of bat some ideas around about different trade-offs that we're making well the deep irony here is that slack was an internal tool for a game studio like it was actually like the we need to build our own thing because we communicate so frequently and then it became and there's a historical analogy by the way of this that i didn't include in the deck because i uh i really like drew hauston from dropbox but like they built their own data centers and like i don't know uh like yeah if that was a good cost decision from them but like from a focused decision should they have just used aws and you know uh or gcp or or one of the other i don't know i don't know the answer to that and i didn't want to like put him on blast and actually get into the debate because i like him quite a bit but like i don't know if that's like the right decision for them and that they were even like the furthest you know they're like a tech company that that was their business able to decrease costs and who knows what the opportunity cost of incremental products or mindshare or whatever it was going and doing that yeah no that makes a ton of sense last question uh did did uh the current ai suite make making your annual report significantly easier or was it still So...

2:04:12Handcrafted. Handcrafted. It's an artisanal craft of this, but I will say it is interesting doing this deck every year. It does serve as a snapshot of what the model capabilities are and how much progress it's been made. And so I think if I go back like two years ago, that version of it, I can wordsmith my talking points that I was actually talking when I get up there in front of the LPs and do it. And last year it was actually a decent, I could ping pong some ideas. Here, I would guess, I don't know if there's 68 slides or something, I would guess 75 % of them, AI had some hand in either helping visually lay it out in some way, writing some of the texts, maybe coming up with some of the analogies.

2:04:56And that is such a step function change versus where it was 12 months ago. And so it is helpful every year to like revisit, come back and like see what's actually possible. Because as you just go about your day, you sort of forget what three weeks ago was or eight weeks ago or 15 weeks ago. But when I went through the process this year, I was like, wow, this is really much less painful. And I think the principal and associate on the team that worked with me on this, we're very much appreciative of where the model capabilities are going. because I think if it made my life a little bit easier, it definitely made their life a lot easier.

2:05:30Totally. Is that alpha for up-and-coming venture capitalists? What advice do you have for those who want to make a career out of venture capital? Because it feels like coming in and surprising the entire partnership with a very deep analysis. I would just say have a non-traditional background. Maybe grow up in the Menlo, Palo Alto area. Go to Stanford. It's actually an interesting thing. so so if you guys uh have a minute i can i can riff for a second on this but like historically so we've hired people out of investment banks largely speaking and so why do we do that well we hire people out of investment banks because it's an expressed interest in finance and technology okay that's great two is they have the model training of like what we need you know the cap tables and and you know projections and all that stuff three is there's like a high pain tolerance and like willingness to grind and do the extra thing and then four is like it's a referential network.

2:06:23Like we can call the same MD at Morgan Stanley or Goldman Sachs or Catalyst every year and be like, hey, how does this person calibrate to that person? And it gives us a qualified pool of people to pick in. The thing that investment banking didn't have was you're very much like if you ended up in investment banking, you followed a pretty straight path for the most part in your life. And I say this as a former investment banker myself, where like you went to a high school, got good grades, got into a good college, you know, did interviews, got a good job. Then at your investment banking job, you're staffed with like 90 % of your day is pre-filled by someone else.

2:06:57And so it's like, okay, well, if I work hard and I stay late and I do this pitch book, align the fonts the right way, I'll get a good bonus and then I'll get a good job. Well, then we drop you in. And increasingly now with the model capabilities, the financial modeling, Claude can do it better than, you know, or as well as most of the people on our team and the sort of remedial tasks are getting the water level keeps going up. And so when we hire people in now, we've always had to train on the agency thing. And it's a little bit of rewiring your brain where, hey, my day used to be 90 % filled by the staffer.

2:07:32And now you're telling me just go figure out what's a good company. Like, where do I even start in that? And so in some ways, investment banking is actually a bad pool of how it's wired and prepared people for this world now. Now, historically, it always was, but we were willing to forego the agency because we got the modeling capabilities and the remedial tasks. And we sort of took that as the basics. And then we had to try to figure out if there was agency there. Now, increasingly, like the models are getting so good that agency might be the only thing that matters. And so, like, are you able to find differentiated network?

2:08:07We're actually working on an internal model for agency at CBPN. Yeah, right along with our taste model. We've hit a huge unlock. We've cracked taste. Now it's agency. And so that's the thing that we now are trying to figure out. Where do you find pockets of people who still want to do the job talent-wise or have the capability to do the job, but also have agency? And you might be finding people that are entrepreneurs. You might find people that are project managers. You might find people that have taken serendipitous paths in some ways. And that actually might be a good sign and not a bad sign.

2:08:41And so it's forcing us to think in a different way of where we're hiring people from. So what I'm hearing is that you're pulling up the ladder behind you. That's right. That's right. Any door, I always say when people ask like, hey, how did you get to where you are in your career? My answer is always like, well, it's a specific question what I would do if I was in your seat. Because I can tell you the doors I walk through, but those doors aren't just shut. They're like shut. They're cemented over. They've like been fortified. You're not doing what I did. I lucked through this path. Fuck! Thank you for joining us.

2:09:15Always a great time hanging out. We'll talk to you soon. Just with all the AI progress, try to ship one of these a week. You got it. I was waiting to get it. Have a great rest of your day. We'll talk to you soon. Let me tell you about Phantom Cash. Find your wallet without exchanges or middlemen and spend with the Phantom Card. And then head over to public.com. Investing for those who take it seriously. Stocks, options, bonds, crypto, treasuries, and more with great customer service. And without further ado, we have the founder and designer of Granola, Sam Stevenson. Welcome to this show, Sam.

2:09:48How are you doing? What's going on? I'm good. I'm good. How are you doing? Congratulations. Massive news. Tell us what happened. Let's hit the gong. Let's warm things up since we're in our Lambda lightning round now. I want to hear what happened. So we have raised a$125 million receipt from... Congratulations. I didn't hear it over the sound of the gong. You said Index Ventures? Index Ventures, Index Ventures. And with KP participating as well. Fantastic. What unlocked the round? Is it just continued progress, new features, a little bit of everything? Talk us through the progress over the last year.

2:10:30Yeah, I think it's been, I mean, all the above. Like, we've been talking to these guys for a while. They've been fans of the product. I think as all of our investors have, they've all used the product a bunch before we've got to talking about investing. And then, yeah, like growth has been good and continuing. And I think it's a combination of that. And then I feel like the environment is waking up to the power of having all of the context of what's happening in your meetings in a company. You know, like I think like everybody adopting MCP is like making it apparent that you can like if you have the right context about what's happening in the company, you can do that to power so much of what's happening in your company.

2:11:16And I think we're just well positioned as like that, that context gatherer that a company can take advantage of. Yeah. So my my something I've been thinking about with this category is why do you think the why do you think the labs have not built a product in this space? I'm sure there's a number of reasons, but I'm sure they would generally love the context that you're gathering so that their agents could actually leverage it directly. And so explain like why you guys have had kind of just open, not to say open, but there's certainly competitors, but like why have you guys been able to just run away?

2:11:58Even image gen. Yeah, yeah. I mean, yeah, I'm sure they are working on it. Like, OpenAI had a stab at it last year. They had, like, launched, like, a longer-running record mode thing, which I think was aimed at this. But, like, I think it's a few things. Like, I think we're basically, like, people use us for meetings, right? Which is, like, it's a big time for a startup like us, but it's also only a slice of people's life. And if you're designing a super open-ended, general-purpose chatbot like ChatGPT, I think they're probably questioning, does this make sense or should we be going for always-on recording of everything and anything less is not good enough?

2:12:44I think that's probably part of it. The other thing is, I mean, we found building Granola that that like you can build a granola clone like super easily, you know, right. And like a weekend you could build a thing that transcribes your meetings and gets you a summary. And like all the work is in, um, understanding like the, all of the like social nuance of like, who are these people in the meeting? Why are they meeting? What's this meeting about? And like, therefore like what notes do you want out of it? What are the action items you should care about? Like, um, just, just kind of all the work behind the scenes to like actually make this thing fit into your life in a way the way you'll use it and find the notes useful is a bunch of work and yeah that's um you know we use the latest and greatest models so like you've still got to go and do that work on top of that at least today yeah uh yeah it makes total sense what what kind of progress are you guys make like what is the what's the most kind of like sci-fi element of of the pitch for this last round?

2:13:46Like, are you imagining a future where the only thing that humans do is just kind of meet and talk about what should be done and then make a decision and then machines ultimately carry out all the work? Like, how far are you kind of like taking out the, how far out are you kind of looking? I mean, I can see, I can see that like, yeah, I can see that someone being true like uh i do think like um i mean we do it we do it internally in the company you know we'll we'll meet and talk about a thing and then and then go ask granola to write a brief uh that we don't give that hand to a coding agent or um or we use it as the material to write a job description or a blog post or whatever like the conversations are like incredibly good input for a lot of the work that you end up needing to do at a company yeah um i think the like the more kind of like the most near-term sci-fi things that we see is like, if you want a pulse on what's happening with any project or any group of people in the company, going and looking and asking Granola what's happening with this project is easy and incredibly insightful.

2:15:03I think that's essentially because transcripts are just such a good up-to-date record of what's happening in a company. So much more so than like, I don't know, a Notion doc or a Google doc that someone had to sit down and write. Without making any effort, you have kind of an up-to-date picture of what's happening in the company. And I think that's just gonna be useful to a lot of people in so many ways. What does diffusion look like inside a large company? Like how much training, education, messaging you have to do? I'm sure you have playbooks for this, But how do you, because once you like land, I imagine the next step is expand.

2:15:41What does that look like for Granola? All word of mouth pretty much at the moment. Like for most companies, like we focus so much on just making it like a good product for the individual when we started. That, yeah, like still the majority of our growth is like patient zero finds it at a company. And then tells their friends about it because they love it so much. and we just go on from there. We're working on a bunch of team-facing features that let you harness the value of a whole group of people using it together. The motivation behind a lot of that is that we create a reason for you to just go wall-to-wall across your company with Granola and unlock the power of sharing all that context of the transcripts together.

2:16:30But for the most part, it's still just word of mouth. People love it and they share that with each other. Talk about the end-of-year wrapped campaign. I've heard fantastic reviews for it. That was such a delight to put out in the world. Explain it for those who did not receive one or followed the story. And then tell me about it. So we did a spin on Spotify wrapped, as many companies do, which for us was Granola Crunched. the basic thesis was like you know Granola if you've used Granola some of our users have used Granola for like thousands of meetings over the last year and if you look inside those meetings and across over a year you can tell a lot about a person and what's going on and what's important in their lives so Granola Crunch was basically like every user could hit a button and generate a like Spotify rap style report about that year.

2:17:33It was things like, who's your partner in crime? What's your favorite catchphrase? What are some of the smartest things you said? What are some of the dumbest things you said? That kind of thing. It was mostly fun. There's some things that could hit pretty hard. I remember mine was like, I felt kind of uneasy sharing it with other people because it felt so big personal. I've heard that from a number of people. They've been like, it was scarily accurate and I didn't want to share with the rest of my team because it really understood me. You said nothing from my end, thanks. 700 times. That's great.

2:18:11Okay, so we have this buddy who's built a massive company in a super regulated industry and ultimately had a huge exit but has been through lawsuits you know, lawsuits and discovery along the way. And so he's just like, absolutely hates every product in this category. He's like, I don't care how useful it is. It's not worth, you know, someday having like, you know, basically line by line, live kind of transcript of every meeting that you've had at a company. What, what are the, I'm sure you guys are aware of, of, you know, some of these more like sensitive use cases, like what kind of fix it? Like what, what are you doing at the product level for people that don't want every conversation?

2:18:58You should sell saunas. So if you want to have a meeting that's off the record, you go into your company sauna and then nothing can record you. There's no recording devices. This is the, this is Lindy. Maybe, maybe check, check for any product roles. You got a wire on you? Anyway, Sam. Yeah. Yeah. Yeah. I mean, I think like this is like a really tricky path for us to walk. And like, it's, it's like so important for us to be you know doing what we think is right at every step of the way like I think um I think we have to juggle like I think tools like this are kind of that there will be somewhat inevitable in a work context I think I think like talking about the personal you know like always on recording type things is totally different but in a work context there's just so much value in having like you know like transcribes meetings and therefore being able to use that conversation data for stuff.

2:19:50So I think that's kind of inevitable, and the question is like, how do we kind of make that okay for people in the meantime? And I think like some things we've observed are like for companies that use granola, they very, like, they basically just get comfortable with the idea that like, we're gonna make it a thing that like we use granola internally, and that's the default, like that you should kind of just assume that's happening. and you can know it's okay to opt out and say you don't want to use it. But companies get comfortable with that very easily, we find. Then the external question is a question of following the laws in your state and like Granola, we make that clear to you but ultimately it's a tool and it's up to you to follow the rules on it.

2:20:39But I meant more like even on data retention, right? So like if a company, there are plenty of meetings that are just not that important. But if that was pulled in discovery at some point down the road, it could be unnecessarily damaging. Now, the bull case for this is that whoever's, you know, suing the company and like wants that discovery actually has more like they're more able to make an effective case. Yeah. but uh but but like yeah i was asking do you have a referral link for companies that i'm planning to sue we uh yeah we have like we have this from companies in all directions like uh some companies are like real like this is an opportunity for them to have things on record and so they yeah yeah but plenty go the other way right where they want everything like off the record and deleted immediately um we've ended up building a bunch of uh retention controls so you can do this either way like some companies will set transcripts to self-destruct after 24 hours um so like you know there's no more record of those and you just get left with the notes um and uh yeah i think i'm a big fan of that like i feel like we there's it's not in our interest to have like a like a real like on the record record recordings of everything that's happened like all granola needs is is the the kind of main notes of what we talked about and yeah yeah yeah Yeah, yeah, yeah.

2:22:07Like a healthy level of abstraction is like good for everybody there, I think. Last question, can you tell me about the brand? Because this could have been called like Panopticon, or it could have been black background, steel, silver, you know, it could have been very different branding wise, but it's granola, it's crunchy, there's a, you know, this like green color that you've picked, like it's clearly intentional. What are you thinking with the brand? yeah we basically like i think since the beginning when we first started studying how people take notes i think one thing that was really apparent was like taking notes in meetings is a really personal thing and uh we like there was already a bunch of ai note takers out there but people hated them people nobody wanted to use it and no one felt comfortable kind of like writing their raw messy thoughts into them um and so we really just wanted granola always to feel personal and like it's yours and that the more you use it, the more it feels like your space.

2:23:06And all the branding is like downstream of that. Like the name, the colors, the kind of messy, like organic feeling textures and stuff. Like it's all meant to communicate that this is like your thing. This is not your company's thing. This is your thing. Love it. Well, thank you so much for coming on during a big day and breaking it down for us. Very exciting progress. Fantastic progress. Talk to you soon. Thanks a lot. Have a good rest of your day. Good hanging, Sam. We'll talk to you soon. Bye. Let me tell you about Vibe.co, where D2C brands, B2B startups, and AI companies advertise on streaming TV, pick channels, target audiences, and measure sales just like on Meta.

2:23:39And let me also tell you about Gusto, the unified platform for payroll benefits and HR built to evolve with modern, small, and medium-sized businesses. And without further ado, let's bring in Ben from Pulsia. What's going on? How are you doing, Ben? Guys, how are you? Welcome to the show. Introduce yourself. Since this is the first time on the show, tell us what you do.

2:24:00Ben Broca:Yeah, my name is Ben. I run a platform called Pulsia. It's an AI that builds and runs companies autonomously. You give it an idea, and it's going to go about building the product, running the marketing, running ads, doing support, and all of the things that a founder would do to start a company and grow it. So should I think about this as you've fine-tuned a bunch of agents, built MD files, or workflows that then leverage other foundation models to deliver on those? You have some playbooks in place. Like, what have you done that's, because I imagine you're not training the actual foundation model, you're using different tools off the shelf and different integrations and then fine-tuning things.

2:24:49But walk me through, like, the actual experience of using the product. Of course.

2:24:54Ben Broca:I mean, the way to think about it is, I spent a lot of the past 12 months spending 12 to 16 hours a day using AI, using cloud code, using codecs, and building my companies with it. And the idea is that the fundamental models are super powerful, and I pretty much think that AGI is here at this point. The models are super intelligent, and they are fluent at using any tools. But I think the trick is knowing how to configure them to give them the right tools, the right orchestration, the right series of tools to get to an outcome, right? So for example, one of our agents on Pulsia can run meta ads campaigns.

2:25:35Ben Broca:But to do that, there's a lot of steps that are needed, right? It's like creating the creative, you know, using maybe an AI generation model. And different labs are good at different things. You know, Nano Banana is great and Codex is great. and there's writing models and there's all sorts of different things. So you're choosing those and rerouting those. How do I think about it in terms of an actual payment flow? Is there a world where I give you a credit card or a bank account and then you already have the integration set up so I don't need to go set up an AWS instance or I don't need to go set up a meta ads campaign and you can just kind of say, hey, we're running a$100 test campaign.

2:26:20We're going to withdraw$100. We think we're going to bring back$200. Okay, we did. Now I need$1 ,000.

2:26:27Ben Broca:Exactly. I mean, if you think about it, agents are essentially like AI humans that can act on the economy. And today, obviously, if an agent, like if 1 ,000 agents go on Meta to create accounts, Meta will say, no, you need to verify your identity and all this stuff, right? And so there's a first layer of infrastructure to build that we've built at Pulsia, which is how to make partnership with those platforms for them to understand that it's an agent working on behalf of a human for a certain task and to have all that set up ready. And as you said, today we obstructed it quite a bit where you pay a subscription and you get one task every night of your agent doing work for you and then you can do various types of tasks.

2:27:11Ben Broca:But in the future, as you said, if you want to open a bakery in New York and you have this idea. There's going to be a lot of orchestration to buy the real estate, to hire staff, manage them, all the fulfillment. And AI could totally do this. But you probably will have to say, Polsia will tell you, you've got to deposit$100 ,000 on an account because we're going to have to do a deposit. We're going to have to pay the realtor. We're going to have to hire staff. And that's something that what Polsia is trying to do is really give access to all the best practices of being an entrepreneur to anyone who has an idea and wants to fund it and wants to try it.

2:27:51Ben Broca:And obviously it's going to be a much lower cost than what you can do today. So Pulsia's ramped revenue super quickly. I feel like every time I see you guys, the ARR's gone up. But what is actually, give us an example of like automated companies that are working on the platform, like individual entrepreneurs that signed up. Yeah, what are they building? Yeah, what are they actually building? What are they selling? Yeah. So there's like an entrepreneur who's like building like a service to create ads from a script, you know, autonomously using different APIs and reselling that to people and like has a bunch of customers that are paying.

2:28:36Ben Broca:You have another person who's building like an AI receptionist for businesses and so using the agent SDK to figure out how to respond correctly based on context. You also have existing businesses who are using Pulsia as an AI team that can build a landing page for them, create leads, sort of like lead capture, and run ads to get customers for their offline business. So there's a lot of different use cases. It's actually very varied because obviously this platform's promise is so open-ended that you get, and it's pretty affordable. It's like$49 a month to try it for a month. And so you get a lot of people with a lot of ideas.

2:29:19Yeah, it feels like the low-code, no-code, like boom all over again, where like there were low-code, no-code products at the hyperscalers and GCP and AWS, but there were still platforms that did a little bit more and became like low-code, some no-code, and you're seeing the same continuum of how much do you want the platform to help you before you actually open up the terminal yourself? Does the human matter a lot still? Yeah. If I just go on there and I pretend to be my 10-year-old self, am I going to print or am I going to be cooked still?

2:30:01Ben Broca:First of all, this platform is to build real businesses. so it's not like a get-rich-quick scheme. It takes time to ramp up, it takes time to build real businesses. Obviously, trying to do a lot of things on the marketing side to automate more of trying to get customers, but also bringing on maybe people with ODF influence that can bring on their audience and sell them services. But to answer your question about how much the human is needed, I think that as long as humans are the ones buying the goods and services, you need another human on the other end who understands the subtleties of what people want these days, right now, what are the new trends, what are the new things.

2:30:45Ben Broca:In a world where there's going to be an abundance of new services and goods being sold all over the place because all those AI tools are augmenting people to build faster, better, you need humans for the taste. So the way I explain it is you get the 80 % operational work, day-to-day grind that can be fully automated by AI. That's like engineering, that's like support, that's like market research, that's like pricing. And usually you need to hire people for that. And today with tools like Porsier, they can do most of the work. However, the 20%, which is taste, which is branding, which is marketing, trying to market in certain ways, understanding how to position your product, maybe having an audience to sell to, however small it is.

2:31:36Ben Broca:If you have a thousand followers that are dedicated to what you do and they love you, you don't need that much more to get 10, 20, 30 paying customers and start doing income. And today they're selling merch and tomorrow they can sell real services that may be more sophisticated. So that's sort of the way I look at it. There's a world in the future where I'm going to introduce services where you can completely autonomously let the agent run wild, obviously, because you don't have to give it feedback. It will every night wake up and do work. And I'm going to choose ways for you to let it run 10 times a day, right?

2:32:13Ben Broca:If you pay to compute, right? And I'm sure that would work. It's just that that becomes like, you need to have a very tight feedback loop on what the user feedback is, so it feeds back in to what the service is and how to make it better. and I think there's a world where a human with a lot of capital can actually start building a lot of money-printing businesses as the loop gets tighter and the platform gets smarter about what are the best practices. And I think this is where the world is going. And ideally, I want to give that opportunity to the 99%, the people that think that AI is charge EPT, and that's pretty much it.

2:32:52Ben Broca:And if we can give them the tools to be economic actors in this new era, I think we will hold benefit and it will be a more just sort of like society. Okay, last question. We have to ask you about the name. You write a lot of people up because it spells AI slop backwards. Is that intentional? Is that a joke? What's the name? I mean, it started as a, not as a joke. It was like my lawyer asked me to come up with the name for the ink when I started the company. And I was on my couch and I was like, oh, I could name it like, you know, Pulse, yeah, I still up in reverse. That's a good name. Intentional.

2:33:29Ben Broca:So it was intentional. It was intentional. That's amazing. That's amazing. I thought it was by accident. But it was not intentional. I decided to use it as the product name. I started the company in April, and I built the product in November. And I was like, that's kind of cool, actually. And I will make people talk. And it did. Well, thank you so much for coming on the show and breaking it down for us. Have a great rest of your day. Yeah, good to meet you. And we'll talk to you soon. Sounds good. See you. Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB.

2:34:05Don't just build AI. Own the data platform that powers it. And let me also tell you about fin.ai, the number one AI agent for customer service. If you want AI to handle your customer support, go to fin.ai. I would love to sit in on that pitch meeting. We're building an infinite money glitch. It did seem like that. $49 a month. I mean, I don't know. There's a world. Teespring was empowering entrepreneurs to sell a lot of T-shirts. There's a lot of different things. Depends on what you bring to the platform, I suppose. Well, without further ado, we have Brett Adcock in the rooster and waiting room.

2:34:39Let's bring him into a TV pin ultra. Brett, how are you doing? Guys, good to see you again. Good to see you again. Welcome back. Been far too long, but since the last time we talked, you launched a new company. So break it down for us. Oh, Hark? Yes. Yeah. Let me. I want to know about that. Okay. Well, I mean, I guess the summary here is I've been working for the last three years on, I think maybe one of the hardest AI problems in the world, getting AI to work on humanoid robots. Yeah. Separately, you know, separately, I've been like watching, I'm basically using and watching what's happening in the digital world, like, you know, the different language models.

2:35:12And to be honest, I think it's just incredibly dumb. Like I, they don't remember anything about me. It's not very personal. They can't listen or talk to me really well, can't see the world, can't use computers well. I just think this whole experience is just like, I think it should feel very much like a sci-fi movie. It should feel like Jarvis that can really understand you, very personalized, use tools well. So about seven months ago, I started a new AI lab called Hark. And we want to build a really advanced personalized intelligence. in order to get there we think there's some fundamental gaps remaining in the models so we basically have a large focus on trying to build new multi-modal models and the second thing is we're interacting with AI today through 20 year old computers like my phone and laptop these are all decades old and we feel very strongly that there's a next generation of AI devices that need to be built to kind of interface with AGI appropriately.

2:36:22So we have a team dedicated not only to models here, but also on the design side. One of our key guys, Abadur, started about four months ago, previously led design for MacBook, MacBook Air, iPhone 13, 15, 16, 17, was keynote for iPhone 17 Air about five months ago. So, Abs is here with a killer team on the hardware side, and we're designing next generation interfaces for the models that we're working on here internally. So, is it the interface that you think is the issue, or do you need more compute locally? I think there's some big gaps in the model side. I think there's twofold. I think there's some large gaps remaining on the model development side that we want to try to close.

2:37:07And then secondly, I think just the interface of how we're using traditional computers right now to interface this AI is extremely broken. We think both need to be fixed. To have a really killer, super intelligent personal assistant, we think you need to fix the hardware interface. And we also think we need to fix the model side. I mean, there's just simple things today that need to be better. Like computer use agents today are just not very good. They're getting better every month, but there's still a large gap in order to get there. Speech-to-speech systems, which will be a really natural UI into AGI, are just not great.

2:37:46They don't remember things I've told it. They don't have access to my life. They can't access my calendar. They're pretty high latency. EQ and naturalness are not great. We're taking this holistic approach to this problem and saying we have to work on the models and we have to like fix the interface issue here today. What is the hiring market right now for all of this, all of this talent? Because you're basically going up against Apple, OpenAI, Quad, you got Demis. Like you're going up against, you've already bit off a lot, obviously with Figure and we can move over and get the update there. But I'm just so curious when you're recruiting talent for Hark, I have to imagine any of the people that you're hiring if you want to hire them they probably have the opportunity to work at these other companies so what's working on that side?

2:38:44I think the summary is all the other companies are kind of boring, they're all doing the same thing they're all copying each other we've headed a certain direction in the last three years I think that direction is somewhat saturating to work on vision understanding if we're working on models that can go and interact with the world and get that interaction data, we think these areas are especially important to push the boundaries and get the AGI, this AGI feeling of highly multimodal scenarios. So we're finding from a hiring perspective, we're being extremely competitive. We've brought on now over 50 people into the team, about two-thirds of that from the AI side, from top frontier labs.

2:39:28I will say it's probably one of the most competitive areas I've hired for in general around compensation the space is just completely lit up like I've never seen like it before I've hired people across all areas of robotics and AI and software and hardware just like this is it's next level competitive I think we have a very small amount of people in the world that really understand how to build the right infra, pre-training, data mix all of this is just very tough So some of these spaces are just new, like computer using agents that can really reason really well in pixel space. Like this is just happening now.

2:40:02So there's not a lot of good precedent for how to go out and build these systems. Why not? What was the decision making process around doing this externally? Because I feel like a lot of the capabilities that you want to build with Hark, like I'm assuming you'll want to integrate into Figure. if figure is going to be a robot that can add value to my day-to-day life, where is the overlap and why build it externally? I'm a bit fan of focus. I feel like figure, we have a singular focus, which is like how do we solve for a general purpose humanoid robot? How do we build like a human in a body suit that has a common sense reason?

2:40:43A lot of the AI focus we have around figure is basically how do we predict physics around things I grab and touch and move through the world. At Hark, we have a different objective. We want to launch next-generation consumer electronics, and we want to basically build extremely multimodal models that can almost act as a Jarvis-type interface to AI. And the focus on those tracks are completely different. And with that said, though, I think there's some opportunity over time to closely collaborate. the voice on the robot today is using the Hark voice API. So if you talk to any of our robots here today, it's using the Hark voice model that we designed here internally.

2:41:24So I think there's a lot of room over time to collaborate the business together. We're both, we're taking an entire data center of B200s in April here and Figure literally has half the building and Hark has the other half the building. Obviously paying for things separately, but we literally between the two of us have an entire data center of next-generation blackwells that we're using for training for AI models. I want your latest timelines on the chat GPT moment for robotics, humanoid robotics. We were talking to Sean McGuire about this. He was putting it maybe like two to three years away. Three to four.

2:42:04Three to four years from just seeing them on the streets, seeing them in restaurants, seeing them in the real world. Maybe not economic impact, because that could happen in all sorts of different industrial settings. But it will be a special moment, I think, when people wind up interacting with a humanoid robot. What are you thinking these days? You can come to figure right now, and you can see robots running complete 24-7 shifts, fully autonomously, with neural nets all the way down the stack. so like i think um this will be a big year for us to ship robots commercially to many different customers of ours yeah and then we're also working on trying to how do we integrate these into the home yeah how do we get robots to go and do like laundry and dishes and tidy the house like things that we just like i don't want to be doing nobody really wants to do and use like robotics as a key tool for this um i think we're i think we're we're like having this moment now i think it's uh we're in it we're feeling it as like right now like we're seeing these robots do uh long horizon autonomous work at figure here and i think over like this year and next year it's gonna be very the whole world's gonna wake up to it i think we saw a little bit of that the white house last week with uh with figure it was just like we saw like unprecedented demand like uh like uh it was um it was kind of crazy because we didn't show any like new capabilities so we're like uh internally we're like okay we're just gonna be at the white house and it was a big it was a big milestone, like the first humanoid robot built in the US at the White House in history.

2:43:40So it was like getting the invite from the White House to come there and be able to be the first one to do it was just huge. It was great. But there was no new capabilities there. But the whole world is just like waking up to the moment now of humanoids. And it was very apparent from last week and all the incredible reaction we got from basically the entire world that we're still early here in the cycle. I love it. Well, good luck and thank you for taking the time to come chat with us. One quick question. Can humanoids reliably crack open a cold Diet Coke and serve it? I imagine the tab is kind of a challenge.

2:44:15That's AGI for me. I don't think they should have any problem doing this. That's a demo. That's a demo we'd love to see. Because honestly, John would buy a humanoid just to come over with a cold one, crack it open. He goes through a lot of these. So I think it gets a real utility out of it. All right, John, bring a six-pack over to figure, and we'll test them out in person. It's a deal. I'll talk to you soon. Awesome. Yeah, see you guys. Good to see you. Have a good week. Cheers. Talk to you soon. Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange.

2:44:48And let me also tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. And without further ado, we have Andre from Console. TBPN Royalty. in the TDP Electric. I'm Andre. How are you doing? What's going on? Great to see you. Great to see you, dude. Good to see you guys. Thank you so much for taking the time to come chat with us. You're always with us. Yeah, you're always with us. Yeah, I was going to say I love the sneakers. It's special to have you here. Anyway, why don't you just give us a general update on the business? Like where, walk us through some features, some customers, and then I want to hear the latest and greatest.

2:45:24Cool. Absolutely. Thanks again, guys, for having me on. Of course. Yeah, so I think as you guys know, maybe the rest of the audience doesn't, but we are, we're a console, we build AI agents that automate service management or employee support. We do that directly in Slack or Teams. And things like onboarding, offboarding, PTO requests, access management, even telling the facilities team there's no more napkins left in the bathroom or something. and so last week we just launched our kind of a big product called assistant and assistant helps you do tier two work so it's not just the tier one employee support stuff now it's an assist it's an agent that helps your it hr legal finance team automate the more complex tasks that often span you know multiple systems so for example say uh you know there's like there's an internet outage you can actually now ask console to go and investigate across those different systems.

2:46:24So on the back end, we're plugged into your Meraki, your CrowdStrike, your Okta, your Entra, all these different systems. And console knows how to go and pull data from those different tools and kind of come back with a report for you. You can also tell Assistant to go and fix things, right? So you can say, actually go and push this update to this user's laptop or something like that. And then you can also have console build out itself now. So with Assistant, you can tell it, hey, I want to connect into Koopa or NetSuite to pull this information. And console will go read the API docs and then build its own connector into that system and then write a workflow with it.

2:47:00Well, yeah, I was about to ask. That's basically like it's instead of doing a feature request, just request the feature. Yeah, yeah. I was going to ask, it feels like years ago you would be spending like, you know, you'd be stack ranking all the different integration requests. It would take like maybe a month or two to write each one by hand. I imagine that that was accelerated when you first built the product, but now it can be handled on the customer side, which is crazy to me. Yeah. So when we started Consular, we were like, hey, we're actually going to build this framework internally. And then our engineers are going to use AI agents to build out integration super fast using that framework.

2:47:39And so we get things done in like two to three, four days. and then I think a couple, maybe two, three months ago we were thinking to ourselves, can we actually just have console do that just iterate on itself and that's where the idea came from and that's what it does, so you just tell it the same way I would tell an engineer hey, we need to build this for this customer, now the customer can just tell console directly I want to pull this data, I want to pull all these actions into these systems and it'll build itself out to do that. You think that's going to be an entirely new, basically, part of every product where you'll still be able to request a feature, but at some point it's like you're paying for the software.

2:48:21You should be able to adapt it to your own needs. But no one's come on the show and pitched that specifically as a company in the application layer. Basically, we're going to give you the autonomy to adapt the product to your own needs. Yeah, it's a crazy new paradigm. Which is just crazy because like every, every sat, like just every customer wants that. Yeah. They want it. They're not like, oh, I want to wait for the support to get back or my account manager to talk to the engineers. Yeah. I'm sure it'll come about in other tools as well. I would say the core unlock for us was actually building out a really robust, again, kind of framework interface that we can plug into.

2:49:07I think once you have that and you have, you know, the really important piece for us is the context graph. So we have a context graph under the hood where we're ingesting data from these different systems and we actually model out your organization. And so when you tell it, hey, go and update John's laptop to this version, or you ask it, what is John's version of his laptop? It gives you an answer. You say, go and update it. Console already knows who John is. We know what laptop you have. We know how to find it. if you were to build that from scratch, you might have committee lookups. It would get a little lost in the sauce.

2:49:41And so that context graph is really important. It's a really core part of what we've built here. How are people thinking about... You've got to coin this, by the way. This isn't entirely... You need to... Or deployed something. I don't know. Yeah, something. What should we call it, guys? I don't know. Like agentically deployed engineering or something. I was saying name it after yourself. Yeah, the Serban method. Yeah, yeah, the Serban. The Serban. Yeah, you got to get people posting like, something like that. Figma needs to incorporate the Serban method. Yes, yes, this is it. Yeah, I think we got it.

2:50:17We'll work on it. So talk me through the experience of onboarding to console and how people are thinking about this in terms of like net new functionality. So I'm basically increasing my IT spend, but it's all justified because workers are happier. We're getting more stuff done versus like ripping, replacing an existing system. We're not going with an alternative solution or like at one point in like 2013, I was scaling a startup. We had like 50 employees. We had like an outsourced IT partner that was like one day a week and they managed like a ticketing system. It was very manual, but there was basically like a consultant who was available every once in a while, like a fractional IT person.

2:51:02How are companies actually interfacing with console and integrating? Yeah. So I would say most IT and kind of service management teams roughly scale linearly with headcount growth. So you have like one IT person for a hundred people or maybe one to 150, maybe one to 200. If you don't include Ramp, who has a very insane ratio. But, you know, Most companies are in that range. And so with console, they're able to take that, you know, they go one to like 400, one to 500. So we act as ultimately like this force multiplier for your team. And so, you know, yes, there's spend going into console, but you're actually saving on the back end of that as you're, you know, we work with companies like Databricks, you know, Cursor, Figma, Chime.

2:51:54Founder. You know, and these guys are growing incredibly fast. And a lot of these guys actually have plans to keep their IT teams flat, you know, through this hyper growth phase that they're about to experience. And it's entirely because of console. And so we're seeing, we're starting to see that now, not just in IT, but, you know, HR, legal finance workflows as well, where, you know, they're just doing this employee support where they're just answering questions that, you know, answering questions or taking action into systems that they have kind of elevated permission. into. Now you can have an agent that just does that.

2:52:31So they don't need to spend their time on that. They can focus on more in-person. What's your approach? Do you just, are you the only IT person at console? Like, do you force yourself to dog food the product to the extreme? Or are you, what's the... We take a bit of a crazy approach here where everyone has full admin access to console and everyone is just encouraged to build their own work. That's probably smarter. You want everybody using the product. At some point, we probably need to pull it back. I think our head of security was complaining last week. He's like, okay, we've got too many salespeople in here.

2:53:04I can imagine. But we actually have our, I was actually just talking to our office manager. We're going to have console just do our dinner orders as we do dinner in the office every day. And with Assistant, now she can build out her own workflow. She says, hey, I want to ping me every week at 4 p.m., give me the options, I'll select it and just go and order it into, I think she's using like ECCater. So not an integration we would have built out on our own, but I think with Assistant it can do that. And so our head of security actually, he was presenting a use case to us the other week. He rolled out CrowdStrike on our devices and he did in like 40 minutes instead of I think what he said, it would have taken him like a day or two.

2:53:50And he did that entirely through Assistant. He was just, hey, I want to plug into these laptops. It went, it understood, hey, you need to download these two binaries if you're going to deploy it on these versions of macOS. Here's where you upload it. Here's the script you write. Okay, do you want me to push it? Yes, and just deployed it. So I think there's more use cases than we can imagine. And so I think we work closely with our customers where we're almost like just trying to show them the technology. And then I think they tell us what they want to build. Yeah. Yeah, I mean, there's so many startups.

2:54:20I'm sure a lot of founders in the audience have felt this before where you're the CEO and you set up all the IT systems. And then actually offboarding as the super admin is extremely difficult. I actually fully lost my Amazon account at a previous company because it was so deeply integrated into the company that they just couldn't figure out how to change the super admin. I was like, just take my Amazon account. And so I just don't have Audible anymore or like Amazon Prime. I need to set up like a new account and basically just declare like Amazon bankruptcy because I was just the admin for like a decade.

2:54:57And things just like built up and I try and give people the other password. Anyway, there's a million. How are you? How are you? How do you? How do you try to? How have you been trying to model like the like overall opportunity for console? Because you're still early stage. So sure. At this point, it's just like, let's get as many great companies as we can on the products. but then five, 10 years from now, at later rounds or stages, you'll probably be asked that question more seriously. But how do you think about it? Because you are at this moment selling against what historically was the kind of, I don't know, labor tam to some degree.

2:55:37But the other side of that, the interesting thing is there's a lot of companies that would use console that never would have had a dedicated IT person. And so it's not just replacing people. It's like bringing a kind of capability to a company. But how are you thinking about the overall category? Yeah, absolutely. So there's a couple of things, I think, that are going on at once. I think the first one is, you know, when we look at IT today, it's very much a reactive role. It's very much a cost center because I think it's kind of like escaped us a little bit, right? In the 80s and 90s, IT was actually an enablement center.

2:56:20You were bringing in technology or deploying, giving people computers, deploying Wi-Fi, or not Wi-Fi, but internet, giving them email and so on. That's like a force multiplier. We've done too much of that. We have too many SaaS apps. There's all this sprawl. And now the team is just managing that. They're just doing support. and so with console the way we think about it is we're going to bring our teams back to kind of what it was like in the 90s where you can actually have this agent handling you know all of the the the ticket management for those those simple systems and um those those kind of basic requests now you have assistant that can go and do more complex work across the the enterprise and the value there is now you can do 10 times as much as you were going to do it you know in that year And, you know, if you think of, you talk to any company, no one will tell you, oh yeah, like our IT team is kind of always on top of it, our IT systems are, you know, perfectly set up.

2:57:18There's always something kind of a little bit lagging behind and it's because they're just constantly drowning. And so console allows them to kind of focus on the more strategic and kind of be more outcome-based. And so I think when you think of it that way, you know, that the TAM is actually all of the work that, you know, all these companies would love to do. And they just don't have the employee headcount or the cost really to go and spend on it. And so that's one big piece. The other one is we're actually just ripping out and replacing tools like ServiceNow and Jira Service Desk and FreshService and kind of replacing it with a more AI-native solution.

2:57:56And so we expect that to kind of scale as you onboard more, as companies become more digital native, right? They have more SaaS. You're saying you're the SaaS-pocalypse. You're flaming the fires of the SaaS-pocalypse. I didn't say that. I said that. Uh, very, very cool. Uh, yeah, it's great. Great to get the update. Um, that, uh, yeah. And, and we're going to work on this coinage, go back with the team, brainstorm a little bit. You tell us, you tell us what, and we'll start asking every company, are you guys using the Serban method? Yeah. It's good. It's good. Yeah. I love it. I'll run it by my head of growth.

2:58:35We'll see what she says. Fantastic. Good. Great to see you. Have a great rest of your day. We'll talk to you soon. Thanks guys. We'll talk to you soon. Um, there's a bunch of news that we need to run through before we head out. Artemis 2 is launching and Kalshi has it at 64 % chance before April 2nd of this year. We're going to the moon. Four people are going to the moon. Everyday Astronaut says, I'm honestly shocked at how the general public has no idea Artemis 2 is taking humans out to the moon and will be the furthest humans have ever flown. Every non-space nerd I've talked to has no idea we got to get people stoked this is what i'm going to be writing about tomorrow i want to deep dive this why is no one talking about why is no one talking about the moon we're going uh nasa is set to launch four astronauts around the moon the deepest human spaceflight since the final apollo lunar landing on in 1972 and there's a bunch of goals so uh you can go track that and uh and we will talk more about that tomorrow and of course bring you a whole bunch of other news and interviews tomorrow.

2:59:41Leave us five stars on Apple Podcasts and Spotify. Sign up for our newsletter at tbpn.com. It's been an honor. It's been an honor. It's been an honor to be here. It was a rough couple days. It was a rough couple days being away. Yeah. But we're back. We're incredibly back. I'm glad. It's going to be a great week. And have a wonderful evening. Yeah. We will see you tomorrow. Goodbye. Throwing smoke. Throwing smoke. Okay. Goodbye, everyone. See you tomorrow, folks. Goodbye. We will be back when the smoke clears Have a wonderful day Goodbye