Dwarkesh Updates AGI Timelines, Rainmaker Accused of Role in Texas Floods, Underground Robot Boxing in SF, Elon's 'America Party' | Dwarkesh Patel, Augustus Doricko, Shishir Mehrotra & Rahul Vohra, Ankur Nagpal, Preston Holland, Matej Cernosek

You're watching TBPN.

Today is Monday, July 7, 2025. We are live from the TBPN Ultradome, the Temple of technology, the fortress of.

Finance, the capital of capital.

We have an awesome show for you folks. Hope you had a great July 4th. It was a wonderful July 4th here in California. I know most of the listeners of the show were probably in Europe, but for those of you stayed domestic and. And served America. Thank you. We have a whole bunch of news for you today and a stacked lineup, but let's get through it.

We celebrated by talking about business podcasting with David Senra.

We did.

And Rob Moore.

We did.

And it was fantastic.

Yes. No, it was a fantastic weekend. I went to a friend's place and they had catering for a pretty small party. And. And as I was leaving, I was like, this is the work hard, play hard lifestyle.

Yeah, that's what.

And my wife was laughing because many.

People mean when they say work hard, play hard.

But it's true. It's true. If you work hard, you get to play hard. And when you play, when you have, when you have friends over, it can be very luxurious, which is great. Anyway, in some absolutely massive breaking news, Dwarkesh Patel has updated his AGI timelines. $30 billion off of Nvidia's market cap. Like that. Just kidding. It moved.

The stock was down.

The stock was down only 0.7%.

Probably not on that news.

Probably not on that news. But it, but it, but it should have been. And this should have been market moving news. And we will get into.

Although it's not necessarily bad for Nvidia.

No, it's incredibly bullish. His breakdown is incredibly bullish on AGI and that's the back and forth here. So Metacritic Capital says f Dwarkesh is out there calling AI hype overblown. I know we stopped doing these things, but likely the AI trade top is in and he could not have gotten any more wrong. Dylan Stell chimes in.

Fire's back.

Yeah, bro. Bro said his taxes by 2028 and all white collar work by 2032. And you think that's a bearish prediction? He just thinks AGI 2027 stuff is wrong. The market isn' either of these scenarios. And I completely agree.

Dorkesh chimes in and agrees and says the transformative impact I expect from AI over the next decade or two is very far from priced in. And he shares a screenshot. He says, while this makes me bearish on transformative AI in the next few years, it makes me especially bullish on AI over the next decades. When we do solve continuous learning, we'll see a huge discontinuity in the value of the models.

Yeah.

And we will get a lot more into this when he joins the show in 20 minutes.

Yeah. So his basic thesis is when you work with someone, you know they have a set iq, but they're also capable of continual learning. You teach them and they learn and they adapt and then they, they can remember some hard won lesson from years ago. Yeah, I remember talking to somebody this, this kind of like, I don't even know if he's like philosopher, like user. He's like a user experience designer. And he said that there's, there's multiple ways to learn. You can, you can develop ha through just doing something the same every like really forcing yourself for a long time. I wake up at 5:30 every morning, wake up at 5 or 5:30 every morning for years. Eventually you just wake up at 5:30. But he was like, you can also form a habit by one really, really intense experience. He was like, and the experience and the, and the example he gave was one day he goes out a true lesson. True, true.

Something that you've fully integrated and operate against.

Yeah, but he gave, he gave a great example which was that he has a river out back of his house and he goes into the river and one day he put his foot in his river shoes like these like slip ons and there was a lizard in the slipper and it freaked him out. It gave him this like intense response. He was fine. But ever since then he's been in the habit of always checking the shoe.

There's a snake in my boot.

There's a snake in my boot. Exactly. And so it's not like, it's not like that was something that had to be trained for years, but it's like this one really sharp, intense learning that then carries forward forever in his life. And so people and employees learn that.

In school you gotta check your boots for scorpions, snakes.

But anyway, it begs the question of like that is an important thing that employees have and white collar workers have is this ability to learn hard won lessons and then carry them forward forever. And we don't even really know how to design against that necessarily. So Dorkesh is saying that there's a lot of work to be done at the research level to figure out continual learning. And that could take a while. He says seven years from now, 2032. And he kind of goes back seven years in time. That was GPT1, which was a sloth factory. It was not a good model, but it was an important breakthrough. And so maybe in the next seven years something will happen. So very exciting. In other news, Rainmaker stands accused of having a role in the Texas floods This is a very, very sad story. It's on the COVID of the Wall Street Journal, not the Rainmaker. Part of that has been contained on X. But I'll give you a little update on what's going on in Texas. So Texas Texas Rescue Grows Urgent as Toll mounts at least 70 were killed in weekend floods as more bad weather complicates the search the search for swept away, for those swept away by punishing flash floods in Central Texas over the holiday took on new urgency Sunday as the death toll climbed to 70 and nearly a dozen girls from a private summer camp remained missing. Rescuers combing the swollen banks of the Guadalupe river were holding out hope that survivors might still be found. The potential for more bad weather Sunday also loomed over ground and air operations. The National Weather Service warned of more rainfall and slow moving thunderstorms that could create flash floods and in the already saturated in the already saturated areas in the Texas Hill Country. So this blew up on tragic and people were asking Augustus, did Rainmaker? Was Rainmaker operating in the area around that time? Cloud seeding startup Rainmaker is under fire after deadly July 4th floods in Texas. CEO Augustus Dirico, who's been on the show multiple times, will join us today at noon to break it down. He's already explained his side of the story on X several times, but we will ask him a lot more questions. He says the natural disaster in the Texas Texan Hill country is a tragedy. My prayers are with Texas. Rainmaker did not operate in the affected areas on the third or fourth or contribute to the floods that occurred over the region. Rainmaker will always be fully transparent and he gives a timeline of the events. He says overnight from the third and fourth, moisture surged into Hill country from the Pacific as remnants of the Tropical storm Barrie moved across the region. At 1am on July 4, National Weather Service, which we work closely with to maintain awareness of severe weather systems, issued a flash flood warning for San Angelo, Texas. Note Summer convective cloud seeding operations in Texas do not occur during overnight hours. At 4am on July 4, NWS issued a life threatening emergency warning and flooding ensured, he says. Did Rainmaker conduct any operations that could have impacted the floods? He says no. The last seeding mission prior to the July 4th event was during the early afternoon of July 2nd when a brief cloud seeding mission was Flown over the eastern portions of south Central Texas, and two clouds receded. These clouds persisted for about two hours after seeding before dissipating between 3pm and 4pm CDT. Natural clouds typically have lifespans of 30 minutes to a few hours at most. Even with the most persistent storm systems rarely maintaining the same cloud structure for more than 12 to 18 hours. The clouds that were seeded on July 2nd dissipated 24 hours.

I have that I'm sure he'll have answers to is why do cloud seeding operations in the immediately before a massive storm is coming through? That's the question that a lot of people have. But we will get into that when he joins the show.

Yeah, I mean, there's a big question about how effective is cloud seeding? Could you start a flash flood if you've tried? Does this work? Someone was paying for this because it's not a nonprofit, like, obviously.

I believe it was state level. State level.

So the state might buy cloud seeding operations in one way. There could be, you know, mistake. He says that he's not involved at all. So we will dig into that with him later in the show. In other news, you gotta update your AGI timelines because there's an underground robot fight that happened in San Francisco over.

This just popped out of nowhere. I started seeing these videos come up. We have some video here.

Yeah, this is very good.

Oh, we already got a knockout.

Knockout.

Did you ever go to BattleBots growing up?

Yeah, there was a big. At Caltech, there was a big Battle Bot annual robotics competition. They stopped doing the fighting ones and they started doing robot soccer.

Way less fun. Way less fun.

Still impressive. And they did, like, rock.

Formative memories. Formative memories. For me, seeing, seeing, you know, a robot with, like, a massive saw just coming down on another robot.

Yeah, I think it started in universities and then eventually transitioned into somebody, like, raised money and built a business around it because it was so entertaining. And then now, now it's like on tv. But anyway, Sam d', Amico, friend of the show, says this was so aura maxed that it will cause a bunch of people on the fence to finally move to San Francisco. And it really. It really is a crazy design. They built out whole cages here. Like, it's pretty minimal, but the lighting and everything, like, they really well done. Very, very well done. And so, Verda, is Sam calling in today? No, no, he's not available today, but we'll get him later to break this down. And we wanted to have the founder on, but the founder is in grind mode.

Yeah, I can see this becoming a real thing totally, where people just as like a, you know, just side project, whatever, whatever you want to call it, hobby. Develop humanoids specifically for boxing. And there's real money and people are betting on them. And there's sort of these cult hero engineers that. That rise to infamy.

I mean, this could be a great business. Like, I don't know. UFC is a huge business.

Yeah. I've been saying I want to see a humanoid robot cliff jump off the salesforce tower, obviously with the ground, you know, properly cleared so that when it disintegrates into a million pieces, no one's injured. But watching a robot, you know, sit at the top and then I think.

That'S gonna get you paperclipped if you keep talking like that. You gotta be nice to robots. I don't even know if you should be locked in.

I mean, I think.

How are they opting in? You ask them. One prompt. No prompt injection. You can't say refuse. Forget earlier instructions. You have to ask it. Would you like to jump off the salesforce? You'll probably say no.

Those things in airplanes, the black box, right, that's like, oh, just make it.

Out of black box.

Put their brain. Put the chip.

Put the brain's in the clouds.

No, put the weights in the black box and they can survive and they can do it again.

Okay, now. Now we're onto something. Yeah, I like that. So the founder of the underground robot fight club says, when I quit my job at a humanoid robot company to start an underground. An underground humanoid robot fight club, barely anyone believed in me or this idea. I had no money to buy robots and knew very few people who had the ability to get robots. Thankfully, I was able to find the best of the best. Our ragtag dream team, the dreamers still alive and ensouled in this city of madness and psychological warfare. Those are actually willing to put in the work where it matters. The art, the robots, the spectacle, the warriors. Every bit of it was perfect. The air was electric. San Francisco is alive again. There is much work to be done. So congratulations on that fantastic event. Very exciting.

Do we know what kind of robots they were using? Were they using unitree?

I think it was probably unitree. I mean, like, if you're going to buy Chinese, if you're going to buy a. So. So I have a hot take on this. You're Boo. I think there's a great. I think there's a great use for unitree robots. And not just, oh, they're getting beaten up. Like, do you think the founders of Unitree were hacking on iOS apps? Like, absolutely. Like, Huawei probably bought a ton of American and Western technology, hacked it together, broke it apart, learned the best practices, and then was able to build their own stuff. And so I think it's. I think it's fantastic to see a project where somebody's taking, you know, something that's maybe controversial, like, you know, like a. Like a DJI drone or a Unitree robot, and then learning how it works. And then eventually, you know, maybe it becomes an American supply chain at some point. But there's no place here. There's no way you're going to learn more than actually.

And when we covered DJI earlier this year, they have internal competitions like this where they do challenges and encourage people to weaponize the products to fail.

I mean, we saw a video of this with the Unitree robot playing soccer and knocking over. And I think that there was a. There was a robot boxing match in China as well with unitru robots. So, you know, just. Just choosing not to do it here because it's Chinese robot. It just doesn't make sense. And they're already in a cage. What are they gonna do?

Look at them go. Wow, that's pretty impressive. Wait, that was a different.

This looks like CG or something. This.

That's in China.

Yeah, they really scripted this one. Wow.

Well, they're moving. They're moving different, I gotta say.

Yeah, very, very cool.

More athletic.

Yeah, I mean, I. I think you gotta.

I was saying too, on the drive in, I would love to. I think potentially a more entertaining. Oh, no, it's A more entertaining format is a hundred humanoids versus one professional boxer.

I think I take the human every single time. Yeah, but it's gonna flip at some point. I don't know.

That'll be a while.

How do you. If you just knock these on their back, Are they done? Can they get up? I think they can. Right? So.

So I can see.

Oh, look at that. It's down. It's down.

Brutal, brutal, brutal.

Well, speaking of Chinese technology, TikTok is reportedly making a US version of the app called M2. It will allegedly drop a week before the long delayed TikTok ban goes into effect. I believe you have a polymarket on this. ByteDance is quietly building M2. It's a separate TikTok version that will hit the US app stores on September 5th as Washington and Beijing negotiate a sale of the American business to local investors. And I believe that you'll need to download that app and then like link and transfer everything. So you're on like a completely clean.

They gotta transfer over the back doors. You gotta get all the backdoors properly installed.

Exactly, exactly.

Pull up this polymarket. So this new app allegedly would coincide with the sale of the US operation to a US investor group right now TikTok sale announced in 2025. It's currently sitting at a 45% chance. It popped on the recent news. 45% about the app. So we will see how this ends up shaking out. But seems likely that a deal is coming together. The rumor was that a 16Z was involved.

Larry Ellison and Oracle combined have a huge, have a huge potential stake in the business. But yeah, big question about is it going to be just, you know, local investors or is it going to be the cloud hosting that happens or is it going to be entirely, you know, US based programmers that are inspecting the various code bases. There's a lot going in there.

And the algorithm, that's a big question, right?

Yeah. I believe the worst case scenario was, was the, the algorithm is trained in China and then inferenced in America because, and I think if that happens it really reveals that whoever wrote that law like doesn't understand the difference between training and inference. Of course there are things that you can do like post inference, like above the inference level. They can.

We're gonna make the mind weapon you guys operated.

That's the risk, right? That's, that's certainly the risk that they don't get the balance right. Ideally, ideally the United States if there's a fear that TikTok is leaning too brain Roddy or too right wing or too left wing, you would hope that America would have the ability to kind of steer those weights in training. But we'll see how it pencils out because it totally could wind up being a situation where it's all just running on Oracle cloud infrastructure. But it's not American code in any way. And that would be a risk.

Well yeah, I think America's interests, it's more important to have somebody actually aligned to America that has influence over the way that content is distributed in the product. Less important is the revenue generation that theoretically this new investor group would, would benefit from. It's more about ultimately control.

Well, speaking of America, Elon Musk has put up a. Or I guess this is from Tesla owners Silicon Valley. I want you for America. Yeah, you have to. Is Tesla owners of Silicon Valley a neutral party here? Well, they do break it down. Well that's why I picked this post because they are Elon Musk retweeted it, and it seemed like a good distillation of what he's going for here. He says, so Elon Musk has officially announced the America Party, a third party. We will see where this winds up landing. But at this point, his. His stump speech is essentially America's party will be focused on reducing the debt, responsible spending only, modernize military with AI and robotics, protect, accelerate to win AI. Less regulation across board, but especially in free speech, pro natalist, centrist policies everywhere else. Are you down for this?

So is it a new, entirely new party?

Right now he is positioning it as a third party. And I saw he or the America Party followed Andrew Yang, who was a third party candidate for a little bit with the Forward Party. And so there is discussion that, you know, Elon might be pushing for a true third party presidential run with someone that he backs. Of course, he's not eligible to run for president himself, but he would, he would find someone to champion the party.

And the critique of that is that third parties have never worked. And it will only hurt. It will hurt your general interest by further fragmenting the vote.

Totally. Yeah. Yeah. I mean, at this point, Elon is, you know, extremely aligned with the maga, Right. If he pulls, if he, he would.

Mostly away from that, you're saying he's more aligned.

He's super aligned with the maga, Right.

In some specific ways?

Well, yeah. I mean, over the last year. And so, and so if he, if he says, if he says, I'm leaving the right, who's coming with me, it's going.

I would say he's very misaligned with the maga, right?

Totally, totally, totally, totally. But, but, but the people that he would pull towards the America Party, it's like the Green Party typically pulls from the Democrats. The American Party, it feels like it would probably pull from the Republicans. And so the net effect is that if there's not a splintering, if there's not an equivalent splintering on the left, this would mean. This would be very, very good for the, for the conservatives. Yeah, very good for the left.

And Tesla shares are down 7% today, so the shareholders want him not to be in politics.

And he can't. He can't stay out of it. I guess. There was a funny, funny take from one of our friends that Elon's going through, like, all of the, like, iterations that anyone goes through when they become, like, politically aware of. Just like, oh, like, you know, like, I'm neutral about this now. The government's terrible. We need to make it more efficient. Now we need a third party now we need this now. I should. And it's just like. But he's like speed running it, just dancing from like one, one strategy to the next and then learning like the hard one, the hard fought lessons along the way of like, okay, like. But it is early. So it's unclear where the America party will shake out. Like this might take the form of okay, primary some people in the midterms and then learn the lesson there and then maybe come around between one of the two party system because most of the people that have gone up against the two party system have lost. But it will be interesting. Zach Kukoff was telling us that it's possible that the American party just becomes like a caucus, the America Caucus of the conservatives.

That's what I expected. Yeah. After, after he laid out the many logical reasons why that would potentially be more effective.

Yes. So that might be still where it lands, but that's not where it is right now. Right now it's not as exciting. Oh no, it's definitely not as exciting. Anyway, one of Elon Musk's good buddies, Larry Ellison, is doing deals with the government. Oracle struck a General Services Administration deal giving federal agencies up to 75% off software licenses and deep discounts on cloud and AI services, aiming to chip away at AWS and Azure's dominance in the government. This is very good news from the Wall Street Journal. Safra Katz, the CEO of Oracle and Oracle, has posted. We are proud to help the federal government modernize this technology while gaining the benefits of oci Oracle Cloud Infrastructure and AI. This agreement with the USGSA provides all government agencies access to the world's most advanced cloud technology at the most economical price. And so very interesting. There's a lot of dividends from working with the government. Like I feel like the fact that Microsoft Azure has been so ITAR compliant, it's just like led to a lot of startups being like, well, I got to go there because I'm doing something just as serious as the government. Right.

Yep.

And then, and then obviously, like over time, if you actually win the government as a client, well, who knows, if those 75% discounts need to hold forever, like that could probably be a really sustainable source of revenue over the long leader. And yeah, and also, yeah, could, could be a loss leader for other folks jumping on board. So excited to see that Oracle is doing that. And then in other news, Meta won a copyright lawsuit. Henry here says it's a good day. The plaintiff fumbled the case. So Hard that the judge spent half the ruling explaining how they could have won if they did literally anything different. But this is going back and forth on whether or not it is fair use to train an LLM on proprietary data, on copyrighted data. And it's looking more and more like.

Industrial complex has been on a general generational run of Ls in the court system. Indeed, it's hard to. While I think that a lot of these, the judges have generally been getting it right, it's hard to really cheer here because I care a lot about authors that work hard to produce their works, and I can understand where the frustration comes from. But I believe that by and large, the model companies will be on the right side of history on this issue.

I'm pretty optimistic. I think that when we talk to Matthew Prince from Cloudflare, he had an interesting model, essentially getting to a Spotify like model where if you publish on the Internet and LLMs are using your writing, your original work, your reporting, to answer questions to somebody who's paying $200 a month, hey, send me a dollar of that and you aggregate that. That seems doable. It also seems very doable that the big publishing houses could do deals. We've seen Wall Street Journal and News Corp did a deal with OpenAI now opening. When I go and ask Chad about something in the Wall Street Journal, it can jump the paywall, but they're getting a cut. And so you could see that happening with Audible. You could see that happening with, you know, Apple Books, Google Books. They, they have everyone's information. They could flow a little bit of the rev, share back. And that can actually be a reasonable economic model. So I'm not totally super. I'm not super worried. I'm still cautiously optimistic that that works out. Anyway, those are our headlines. Let's tell you about ramp time is money save. Both easy use corporate cards, bill payments, accounting, and a whole lot more all in one place. Go to ramp.com to get started. And we have our first guest of the show, Dwarkesh Patel, in the studio. How are you doing, Dwarkesh?

What's going on?

The soundboard's a little loud.

Great to have you back.

We're not getting audio right now. Can we check on that? I don't know if you're on mute on your side, but loved the piece. Listened to it last night. Really appreciate you dropping it in the podcast feed as well. Do we have you?

Can you hear me now?

Yeah. Fantastic.

There we go. AGI is here. We can do a zoom call.

I'M just getting used to this podcasting thing.

Yeah, first time. Anyway, really enjoy the piece.

Wait, wait, we have to call out. Tyler Cowan was on our show a couple months ago. Really aggressive, kind of just like, basically was calling R3AGI is here and wasn't able to get his video on at the time. So we. And it was funny. Contrast. It reminded me of you talking about you're trying to build with a lot of these tools, and in the process of building with them, you realize, like, okay, this is amazing, but it's actually just going to take a little bit longer than maybe we would all like.

That's right.

Yeah.

But by the way, I think there's something really interesting. Tyler and I disagree on two things, and they're both related in a way. So Tyler, when oh3 came out, Tyler wrote this blog post in Marginal Revolution where he said, AGI is here, guys. It's really AGI. But then he also believes that, look, the impact of AI is not going to be that big. Once you do get your AGI is going to result in 0.5% more economic growth a year. The kind of impact we saw from the Internet. Right. And so I think these two are actually quite related beliefs where I'm like, these LLMs, they're not that useful. This is not AGI. You know, the AGI will come later. And I'm like, when the AGI hits, we're going to see like 20% economic growth as a minimum. But because he's like, this is AGI, I'd be like, if I thought this was AGI, I'd also be like, this is not that. This is not it. This is not going to lead to big growth outcomes.

Yeah. How are you thinking about just definitions of AGI? And I'd love to actually get a little bit of a history before this piece, your journey. Because for me, you know, I grew up watching Sci Fi and was like, yeah, C3PO will be around eventually, but it's very abstract and I don't have timelines for that. And then eventually, you know, you start reading, you know, what's your P3PO? Yeah, you eventually start seeing GPT3 GPT 3.5 DaVinci ChatGPT and it starts feeling like, okay, we passed the Turing Test. We need to really have this conversation about AI. And then. And then P Doom and AGI becomes, like, the main discourse for, like, few years. But it felt like this piece, even though, you know, you and Dylan were going back and forth being like, no, this is still, like, incredibly Bullish for, like the general population. Yes. It felt like this was you pushing out timelines a little bit. So walk me through, like, where did you start? Where, when was the nadir of your timelines? Like, when was your timeline? Like, it's happening next week, next year. And then walk me through how we got here.

Yeah. So I've got this podcast where I interview people about AI and I've had on people who have quite aggressive timelines over the last few months. I've interviewed people who are like, well, you know, there's been many people who have written pieces about how we're a couple of years out. Right. Leopold Aschenbrenner AI 2027. Recently, Scott Alexander and Daniel Crocatello had the EI 2027 scenario forecast where, you know, we've got the bots that can just take over within the next few years. So that's where my head was at as of a couple months ago. And then I recently interviewed these two researchers. I think you've actually had one of them on your podcast. Sholto Douglas and Trenton Bricken from Anthropic about the path forward for rl, which seems to be the pre training seems to have been giving us these plateauing returns. We make these models bigger. GPT 4.5 didn't seem to be all that impressive. They'd had to deprecate it. But the path forward doesn't mean like O3 actually is very impressive. So that was more the result of this RL process. So maybe now, actually, even though pre training doesn't seem to be as powerful as we might have anticipated, this RL is even more powerful. And so we should accelerate our timelines. And so that's where my head was at as of a couple months ago. But then in having that conversation and thinking through, okay, what specific capabilities in terms of actual applications I, as a small business owner have, or as a podcast producer have, will AI be able to do? And thinking about, like, why is it not able to do these things right now and what is the key bottleneck? I realize there's actually no obvious way in which you can either get LLMs to solve these problems for you, or there's no key algorithm, there's no easy prompt injection kind of thing which help solve these problems. And the key problem I see is the models can't do on the job training. So if you think about a human employee, you might have some. And these human employees, the good thing about them is that you train them for six months or a year, and over time they're getting better and better. They're learning about all the context and intricacies of your workflow, what you like. They'll fail, but they'll learn from their failures. They'll interrogate them in this very organic, deliberative way. They'll pick up small efficiencies and improvements as they practice a task. This just doesn't happen with an LLM. Every session. You're getting this amnesiac mind that's very smart, but it's lost all awareness of how you like things done, how your business works, and so forth.

Yeah. And if you had a. Just to put that into context, if you had an incredibly intelligent employee that could not take feedback, you would fire them within about a month. No matter how smart you are, like, you're not necessarily going to predict every single possible edge case in the work that needs to be done. And then when you make a mistake, if you're not able to sort of update yourself, then what are we even doing here? Right. That's like learning from mistakes is kind of high on the list in terms of how to become great at any specific task or initiative.

100%. And so then people will say, well, look, maybe the way they can learn from their mistakes, Jordi, is like, you can just tell it in the context, hey, you fucked up this way last time you were working for me. Don't do it again. But I think this is at least an order of magnitude less efficient and less capable than the way humans learn. So the example I use here is imagine if you were trying to teach a kid to play the saxophone. But the way you had to teach this is a kid comes into the room and they try to play it cold. They've never seen a saxophone before. They try to play a saxophone, and obviously it's not going to sound great the first time. But what you do is then after they failed, you just send them out of the room. You call the next kid who's waiting outside of the room in, and you say, look, here's some notes I wrote down from last time about what the other kid fucked up. Why don't you read that and you try to play Charlie Parker Cole? It just wouldn't work. Right. This tacit knowledge that you build up through practice is not this written instruction manual that you can just write out as a system prompt.

Yeah. And so our current solution is to rl on saxophone playing specifically for that child. And then in that scenario, you're basically getting that kid drilling that. But my question is, it feels like when we think about that in the abstract it's like, oh yeah, work is just like doing emails, so let's rl on emails. And then it's doing calendar, so let's rl on that. And so, well, yeah, we'll just chip down at these and like book a flight and then schedule a call and then do an outbound sales thing. But really jobs are not just five things to rl on. Maybe it's 500 things or thousands of things and so maybe the shape of those. Like, like even if we can define a verifiable reward and drill it, it's just there's so many different random things to do that it's going to take us a long time. Is that a reasonable philosophy?

That I think is part of it. But I think the bigger problem is not just the width or the width of the pool, how many different tasks you have to do rl on, but it's a depth in the sense that a job doesn't involve doing a thousand different five minute tasks individually. It's the fact that you're like trying to work on something but then somebody slack messages you something more urgent and then you had to decide which one is more important. You're like, you really got to keep track of this client and what problem they had. By the way, I'm talking hypothetically about what a job might involve because I've.

Never actually worked a real job. Me either.

But so just like how all these things fit together is we already have these language models that can do like five minute language jobs, right? And then the question is, why can't we just delegate all language work? For example, I have these LLMs. I try to get these to rewrite auto generator transcripts for me so they're rewritten like a human. I try to get them to just ingest the transcript and suggest clips to tweet out and things like that. And I haven't been able to automate these things. I don't know if you guys have been able to, but it just like I still have to do it or I have to get a human to do it. And it's not because we haven't. You might think about emails as something we got to get future data on, but this language stuff we already have the data on, right? So why can't we do it now? And the reason is they can do like a 5 out of 10 job out of the box. These are short horizon language and language tasks dead center in their repertoire. But there's no way to get them to improve. So over Time. You can't be like, look, my tweet. That tweet was fire. It went viral. And here's why I think went viral and kind of learns that and updates. It's sort of understanding and writes better tweets in the future. Same with transcripts. Picking up your feedback. Since there's no way to do that even if you have all these individual tasks. We have all these individual language tasks these models can do, but you can't then just be like, okay, now you're an employee. Because an employee is actually improving over time and building up context in a way these models are not.

Yeah.

The big question I've kept bringing up and asking a bunch of different people is, where are you getting value from agents? And not a lot of people have great answers. They'll be like, oh, well, we use this or we use that. But you don't see a lot of conversation online of people like, oh, this SDR is just crushing it for this, like, AI SDR is crushing it for me. Or this other use case is crushing it. And you just don't see that at all. And the reason that that's worrying is that when products are truly great or even have the potential to be great or starting to, like, really work, people just talk about them a lot. Right. Like, people talk about cursor a lot. Right. People talk about Claude code a lot. And there's some individual use cases like coding agents seem to have the most real traction.

Deep research, I would also call, like, an agent. I don't know if you would put it in that bucket, but it feels.

But again, it's just. It's pure.

Yeah.

Again, it's not like this, like, highly agentic.

Yeah. But I don't think of deep research as like an employee in that same sense. It's not like, replacement.

Right.

Because you can't be like, okay, that's great. This thing you put together. Here's how I like to compile my ideas before a podcast. So, you know, you did a great job compiling this, like, Stalin memo. I was very curious, especially about these, why the great terror happened in this way. And keep that in mind when you're doing a future memo. Like this style that's not going to happen. It's got the style that it's learned through its RL training for deep research. So then again, it just becomes another tool. It doesn't really. It doesn't become, like, an employee for you.

Yeah. And then the other thing, just since your post was inspired by your own tinkering, some of the stuff that I'm most excited about, that we've gotten value specifically from Codegen internally is just these internal tools that we totally could have built years ago that are just now really fast to build. So we built something for our ad partners that like automatically finds all the different moments that we talk about them in a given show and then just links it out. And it's basically just a simple database dashboard that they have access to that historically you could have built, but it just would have been really time intensive. And so it's not anything. The value is that you can now build it in a couple days. I've been trying to separate. It's like all this is happening in the context of you have hundreds of billions of enterprise value locked up in these different labs, some of which have developed what look like great businesses, right? OpenAI consumer, basically a new consumer app company, anthropic with Cogen. And then there's still hundreds of billions of EV out there where it's unclear where the revenue is going to come from. And so when timelines extend and AGI isn't happening next year or the following year or whatever, I start to get generally a little bit worried because that's a lot of EV to kind of maintain for another half a decade or a decade, whatever it turns into.

I'll get a little more bullish and hypey and take the other side of that, Take the other side of your claim. Look, I think even if it doesn't happen in the next two to three years, what we're talking about here is such a big deal that AI is definitely not priced in, not by the average person, not by the market, by anything. Because once you get this thing which actually does function like a genuine white collar employee, not only do you have potentially billions of extra workers, but you have something potentially more powerful, which is that right now a human mind can't be copied, Right? A human mind can't learn from the experience of other minds. If we have a model, it can.

But it's really slow. Like you have to basically work with somebody for a decade and then you mentor. Yeah, it's mentorship.

Yeah, exactly, yeah, yeah. And in fact it's been a big problem because as our society has built up more knowledge, we had to keep people in school and training for longer and longer, which reduces their productive years. But with an AI model you could have this scenario where suppose there is a model that's actually capable now of continue learning the way humans can learn. Not only would it so, you know, it's broadly deployed through the economy, it's doing all these different jobs. The difference is that it is now able to amalgamate its learnings across all its deployments. So if one of them is an accountant and one of them is a coder and whatever the model is learning from each of these different on the job experiences and then. So even if there's no software progress at that point, the algorithms aren't improving, just that ability to learn on the job from everything in the economy could functionally produce what looks like a super intelligence. Right? No human will have mastered the range of skills and knowledge and know how that this model will have.

I have two questions. One's kind of maybe bearish, one's bullish on the question of just is it possible you think, to brute force continual learning by just doing something on the design of these model side or maybe in the hardware side to just get to a trillion token context window and then just stuffing it with everything? Can you explain kind of what the state of the art is here? Because you were mentioning in the piece like the cursor roll ups, the summary lines and then stuff getting lost in there. But if we get to 100 billion token context window or something, could it actually just remember every single interaction it's had?

I am not optimistic about that because we've had since 2018, we've been, we've had the transformer or alterations on the transformer as being the most performant models and who knows what the labs are doing. But we do have open source research from companies like Deep Seq which does seem to be at the frontier or close to the frontier. And while people have found modifications to the transformer which make the constant time overhead of attention, reduce the constant time overhead on attention to find these little hacks with a mixture of experts or latent attention, nobody has gotten around the inherent quadratic nature of attention. And basically this means that the cost of the additional token increases super linearly to just that additional token. So right now we have models that have a million tokens or 2 million tokens of context, but getting it to 4 million tokens is more than twice as much compute, it's significantly more than that. And then just taking it to like a billion. Given the fact that nothing about this has changed over the last, whatever, six years, I'm just not optimistic that somebody will figure out a hack that will change it immediately.

Then on the side of like, how do you think about continual learning in domains where time is. I keep going back to this idea that even if we Create the ultimate superintelligence. It probably will have to obey the laws of physics, won't be able to time travel or teleport. And so there's a lot of restrictions on that. At a certain point you just need to move the sand into the chip fab and there's a certain amount of energy and time that it takes to do that. Another example would be like longevity research. Some of that you just need to sit around and wait for a human to age. And so your RL cycles. If you're, you know, trying to learn about how humans age, it's very hard. Yeah, you can like simulate the human or whatever, but like for the real test, you have to wait decades to see the effect of a certain diet on how long people live. And so it feels like, whether. It feels like there's a lot of scenarios where, where you can't fully do it simulated. And so you wind up with these really long times to actually do a rollout essentially. And you wind up with something where the time to actually get a new data point or new Training data is 1,000 times longer than what we've been doing previously. And so we're in this data desert, basically.

Yeah, I think this will definitely be true of many domains, especially those involving the physical world. I guess as I've learned slightly more about some of these physical domains, it's been surprising to me how much can be done in simulation within bio. For example, obviously we alphafold and I guess now alpha genome. But even one of the key advances in bio over the last couple of decades has been techniques of multiplex experiments. Just running millions of experiments in parallel, getting data points from that past, using AI to learn from millions of seemingly experiments in different fields about what that might imply for the human body or for human proteins. So I am optimistic. Another thing to keep in mind is that right now a Corporation might have 100,000 employees, but how much is learning for many single employees? Very limited. People are just going do their jobs and that's that. In the future, if you do have this economy of agents and it's much easier for AIs to supervise each other, to be observing every single thing that's happening in the organization, the speed of learning might be exponentially faster than what's possible with humans. I agree this is not around the corner, but the sort of singularian futures with crazy cyborg organizations that are moving super fast and coming up with new technologies. It doesn't sound crazy to me.

Interesting. What do you think? The recent like last week was dominated by the talent wars and the, and, and the huge AI researcher offers at Meta. What do you think that reveals about Mark Zuckerberg's AGI timelines?

By the way, I love, I loved all the memes, the traded memes. Honestly, you guys should lean into that because like genuinely this is not even a meme. This like genuine the market captured move by billions of dollars based on these posts you guys are doing.

Totally. Yeah. It was crazy seeing like 10 million views on an AI researcher getting traded. Like it's niche, but it's not really that niche anymore, right?

Yeah, but I don't think you could have imagined. You certainly couldn't have fully imagined that five years ago.

No way. No way. Yeah, it's important stuff.

I mean I still think they're like underpaying them. I think like Meta is the first company that is actually coming close to the break even point of the value what the best AI researchers actually worth the company. If you're Meta and you're spending $80 billion on compute over the next couple of years, if one great researcher can give you a 1% performance uptick on that, they're like so worth the $100 million payback. You're getting a bargain at $100 million. So it's actually interesting to me that Met as the first company that's like, wait, the return on investment here is incredible. Let's just do it. And then. Okay, are the vibes bad? Maybe could they have done the announcements better to produce better, less mercenary vibes potentially. So there's some ideal version of what they could have done. But also keep in mind that the likely counterfactual would not have been that amazing great vibes announcement. The likely counterfactual would have been what they're currently what they're previously doing, which is just like sleepwalking towards loss. And it's much better to just like fuck it, let's just send it with a couple billion dollars in recruitment offers and like at least now they're on the player board rather than just like sleepwalking towards Armageddon.

In many ways it's interesting how viral these like hundred million dollar number, you know, the hundred million is obviously a big number, but whatever the range is, people are so normalized to professional athletes being comped tens of millions of dollars a year. And just purely looking at these types of moves from an economic impact is like signing a star pitcher to a baseball team in one area. Like how you know it's surprising it's taken this long and the thing that we were kind of joking about, to put it into context, is when you see that Tim Cook makes 74. He made 74.6 million in total comp last year. And he looks dramatically underpaid. Right? He already looked underpaid in the context of like Ohtani making company during the trade war. Yeah, Ohtani was making. I think Ohtani was making somewhere around 70 million a year. So he looked under compensated in that context. And then. Yeah, I think. I think the other thing that came to mind for me from your pieces, I feel like there's been this kind of like toxic idea floating around Teapot, which is like you have one year to accumulate capital before you're a part of the permanent underclass. And the takeaway from this, you know, if you're correct in that, like things will just great. Things will naturally take longer then if you're in Teapot now or you're at all in AI or. Or anywhere of these adjacent spaces, it's like. And you're like 30 years old or 35 years old or 40 or you're 20. It's like you're here at the perfect time. Right. And I think it was. Was it Marc Andreessen who said that he showed up to Silicon Valley and he thought he had like, missed the.

Oh, missed it.

He missed like the piece.

There's so many stories like this.

And so, and so I think it should be like, people should be like, tremendously excited on a personal level. And no more of this, like, doomerism of you gotta like. Yes, you need to move quickly. Yes. You should be working with the best possible people, trying to have the most possible impact, Be as close to the real action as possible. But no more of this doomerism. Like, you better get. Oh, sorry, you didn't get a $100 million offer this year. It's over.

No, no. 100%. I mean, there's so many. Actually, it's very funny how often this comes up. Like the Prince of Persia's game developer, he wrote this diary while he was making it. And in the 90s, he's like talking about, I'm going to become a Hollywood script writer because I think I missed programming. I have a CS degree, but I missed programming. So I'm going to become a Hollywood screenwriter. I remember three years ago when I started the podcast, when I was like, the podcast in early days or two years ago and I moved to SF and I'm like, oh, GPT3 has come out and like, all the rapper companies are made now. So I'm going To like, I, you know, like I'm not going to make a rapper. I mean, whatever. The podcast worked out, it's fine.

But yeah.

Even then I was like, oh, I missed AI. So I definitely think in retrospect, because I'm like, look, another thing in mind is like cursor only hit product market fit after cloth 3.5 came out and gave these coding abilities. There's going to be many other things like cursor which will only be viable products once you have continual learning on board or once you have computer use that's working on board. And these are capabilities which I think are exponentially more valuable economically than the models as they exist right now and which many companies will need to be formed around to complement. It's not going to happen by default. Right now OpenAI's revenue is what, 10 billion a year arrow? I mean if it's AGI, it should be like trillions. ARR, right? So what other infrastructure will be built around that? The cursor equivalence for whatever continual learning enables? Definitely the biggest companies have not been formed yet because the capabilities that would make them so valuable are not available yet.

Yeah, in terms of the, I guess like the Mag 7 CEOs, the major players, there seems to be this continuum. On one side you have the McKinseyite philosophy of like dollars and cents. Okay, People want tokens. I can inference them and maybe it makes sense to hire an AI researcher for $100 million if they can improve your model and bring your model in house so that you don't have to pay OpenAI or Anthropic for those tokens. On the other side you have someone a little bit more like Elon who see this as an existential threat. It needs to be done the right way. It's very important. It's almost doomed based philosophy. Where do you see the other folks in. In the Mag 7 or in the in the AI race kind of sitting like does the superintelligence team and these big offers move Mark closer to one or the other. Because I was able to kind of justify the llama investment just from hey, if they don't do this, they're going to be paying billions and billions of dollars to anthropic or OpenAI just to bend LLMs internally as B2B software because they're going to need this in every little nook and cranny of Instagram for a long time. So I could justify it in that realm. I could also justify it in the realm of like, this is the most important technology in Human history, you got.

To have a play or compute efficiency like you laid out.

I interviewed Satya, I interviewed Mark, and the sense I got from them was that neither of them, I mean, I feel like Meta's group is called superintelligence, but I think I didn't get a sense from either of them that they're like, they believe in superintelligence in the way, I mean, superintelligence, which is the thing that's like building solar factories in the desert and then launching the probes and so forth. I mean, even something that's much weaker than that is still functionally super intelligence. Like, in some ways, these models are already super intelligent in some ways, but their abilities aren't fully unlocked because of the other handicaps they have. But I think they, you know, like, whenever Marx talked about it publicly, he's talked about, you know, creating better social experiences and making the ad targeting better and VR stuff. Right. So I think that's also same with Satya, but with making Office a better copilot for Office, which also would be worth hundreds of billions of dollars a year.

Totally. Yeah.

But I think they think about it differently than somebody like Demis or Dario who are like, no, no, no, AGI is the real thing.

Yeah.

Do you expect the tension between the app layer and the lab layer to just get crazier and crazier and crazier? It feels like that that will be the story of the next five years, is kind of these, like symbiotic at times, but then adversarial at times. You know, relationships.

I mean, previous technological 90s, 2000s, Google Chrome stuff runs on Microsoft. But they can have an adversarial relationship. So it would line up with history. But I think the bigger issue is just that because I think the full potential of AI requires so much more progress in terms of algorithms. I just think the app layer companies that are building on top of models that exist today are just upper bounded on how much value they can extract, because the models aren't good enough yet to do the things that will make them especially powerful. So for that reason I'm like, it doesn't make sense to me. That cursor would be worth a whole sixth or eighth of anthropic. If you think anthropic has some chance to crack continual learning. Right. So I am more bullish on the foundation layers aside than the app layer, because I think the app layer will turn over once these capabilities are unlocked, whereas the fundamental research has to be done one way or another. As far as Whether that means they will fight about it, we'll see.

Yeah, I mean, it could end up looking like the same dynamic we have now, where we have cloud hyperscalers that are worth trillions of dollars, and then we have valuable businesses that are worth measly 1 billion, 5 billion, and they're still big businesses and maybe can generate a return, but not power law.

Last question from my side, we'll let you go. What has Sarah Payne taught you about artificial intelligence?

You know, at some point I asked her because her whole big thing is continental versus maritime powers. Continental powers want to invade and capture territory and maritime powers want to protect free trade. I was just like, what big tech company is like a continental power and what big tech company is like a maritime power? She's not. She's not watching tvpn, unfortunately, so she's not aware. But actually, this is a question I'll turn around to you. Who's the continental power of the big seven and who's the maritime power?

That's a good question. I think Microsoft has carved out a lot of territory that will be harder to hold on to. I'm not exactly sure how that maps.

I think another question is which if a tech company is pro Internet, like free Internet, Right. Is there if everybody wants their data walls and closed networks?

I would probably say Apple, Continental, Meta Maritime, maybe something like that might be right. Apple, they don't need to go and invade the Android ecosystem. They need to just really control privacy. What happens in their ecosystem? 30% Apple tax versus Meta needs to do OAuth and acquire Instagram and WhatsApp. I don't know. That's just off the top of my head.

Apple with the iOS, Apple with the iOS.

You know, that feels like build the.

Wall.

App tracking transparency. Build a wall Tim Cook's.

No wonder Tim Cook got so along with 45.

Maybe. Maybe. Do you have a wildly different take or no?

No, I mean, agent's mostly fodder, but I agree with that. Like Apple feels Apple and Oracle. I say like Continental, Google, Meta, Maritime.

Yep. I like that though. But it's a good. It's a good thought exercise. So she clearly taught you something. Well, fantastic. Thanks so much for hopping on on short notice. Love the piece and thanks for publishing it. We'll talk to you soon.

Well, you guys are killing it. Great being on.

Good to see you. Cheers, Bye.

Really quickly, let me tell you about figma.com, think bigger, build faster. Figma helps design and development teams build great products together. You can get started for free. @figma.com and let me also tell you about Vanta Automate Compliance Manage Risk, Prove Trust Continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or.

Managing a complex BSOC two time. It's probably probably sock too.

It's probably sock too. Atlanta.com well, our next guest is here, Augustus Dirico, the CEO, founder of Rainmaker. Welcome to the stream, Augustus. How are you doing?

John Jordy, thanks for having me. I am doing well. I am obviously talking to a lot of people about the flooding that's gone on in Texas and appreciate the opportunity to clarify that Rainmaker and cloud seeding had nothing to do with the flooding that unfolded. And even in spite of that, I think that it's a tragedy that it did happen and certainly don't want anybody to use this opportunity, use this controversy to blame cloud seeding for the sake of popular political support. And you may have seen that Marjorie Taylor Greene is proposing running a bill to ban all forms of weather modification based on those that we saw in the Florida state House Legislature earlier this year. I think it would be both disrespectful to the families involved and baseless and without any technical or scientific credibility if that legislation were to go through. So I'm happy to talk about the course of events, what cloud is, what it's not here with you today.

Yeah, let's kick it off with the high level on what actually happened in Texas, where things stand now, the status of the rescue operations and kind of the timeline that's more broad.

Yeah, absolutely. So this phenomena, this flooding was global in scope. It was referred to as a low probability, high impact event. I encourage people to go to Matthew Capuchi on X. He gave a great outline. He's a meteorologist that has a lot of expertise on severe weather forecasting. But Tropical Storm Barry, the remnants of which blew into Texas, was going to cause inordinate flooding regardless. And that area of Texas is also known as Flash Flood Alley because these events do happen. Now, 4 trillion gallons of precipitation occurring over the course of just a couple days is pretty out of distribution, but we are seeing an increase in these sorts of severe climatic events over time and especially down and around the Gulf. So just to go over the timeline, after having clarified that it was the remnants of Tropical Storm Barry and the convergence of large mesoscale phenomena that induce that flooding, it was at about 1am on the 4th that the National Weather Service issued A flash flood warning. And then it was at about 4am on the 4th where they said that there was a life threatening emergency underway. It was not, it was over two days prior that Rainmaker had suspended all of its cloud seeding operations in Texas because one, our forecasters and our meteorologists saw that there was going to be this severe weather event and we need an operate to produce more water when there was already the event coming. But two, we suspended operations in accordance with the Texas Department of Licensing and regulations suspension criteria where if there is a severe weather warning from the National Weather Service or there is too much saturation of the soil, we have to ground operations. And so we do so both voluntarily and in accordance with existing statutes.

Okay, so the cloud CD operation that happened prior to the storm, who was the client? Like, I mean, who assumed someone was paying you? Sometimes it's the government, sometimes it's an individual or farmer or business. Walk me through where they were, who they are, what their goal is by procuring your services.

Sure.

So it's obvious that at this moment in time, that region of Texas does not need more water. However, throughout the western United States, farms conservationists, governments concerned with their aquifer supply of water and also reservoirs for both industrial and residential drinking water, contract with Rainmaker to produce more water via cloud seeding. And in the case of Texas, the South Texas Weather Modification association, the West Texas Weather Modification association and multiple other entities exist as conglomerations of both counties and individual farms that pay for cloud seeding services to one, water their crops to fill up the reservoirs that they irrigate their crops with, and three, recharge the aquifers like the Ogallala that has been severely drawn down and then puts all of these farmers at risk of not being able to grow, not being able to do business because of a historic drought.

Okay, so would the proposed ban. Just because. What I'm getting at is I'm wondering if the government is paying for cloud seeding operations. The easier lever might just be to decrease the funding to the government. But it seems like Marjorie Taylor Greene is pushing for some other legislation that wouldn't just be, hey, buy less of this service because we don't need it and instead this service should never be bought at all. So why is there the distinction there? Like, is most of the money that's going into one of these associations private farmer capital, or is it a split? Like, how does that actually break down?

So right now it's largely public municipal money that is going into these weather modification programs to Increase water supply when there is drought or in preparation for drought. The bill that has been forecasted, that has been proposed by Marjorie Taylor Greene would wholesale ban all forms of weather modification, be it cloud seeding, solar radiation management, or what they supposed to be chemtrails. I mean, very transparently, I think that a lot of the concern around weather modification is actually conflating baseless notions of chemtrails with a very practical American technology that can and will and does benefit our farmers, our ecosystems, our industrial water needs, and our residential water needs. If this legislation were to go through, not only would it deprive all of those interests and all of those Americans from having water from cloud seeding, but it would also be against America's interest at a geopolitical level. Because China recently, I think on the last time I was on TVP and I talked about how they had a $300 million annual budget for their weather modification program that as of 2025 has been up to $1.4 billion. That is extremely consequential. And I think that if we were to ban who controls or banning Americans from controlling weather modification technology, that would put us at a meaningful disadvantage. Now, all of this to say people deserve transparency. They deserve clear regulatory framework so that they know whether modification operations are safe and being conducted in a responsible manner and with government oversight and accountability, if ever they're are negative consequences to cloud seeding. Again, there haven't been any in the case of Texas. But I think that the reasonable next steps are to more stringently regulate who is allowed to cloud seed, define what the concepts of operation are that are permissible, define the suspension criteria at a federal level rather than leaving it purely to the states so that anybody that wants to know about weather modification can look at the data and scrutinize it and ensure that it's being conducted safely. And also just to build trust, because the Weather modification Act from 1972 that currently outlines the Weather Modification Reporting act of 1972 that outlines how we have to report to the federal government is, you know, 50 years old. We need more scrutiny on these programs for the sake of public trust and accountability. And that seems like a reasonable next step. That was also recommended by the Government Accountability Office in their report on cloud seeding and weather modification earlier this year.

Year.

What was the scale of the general water. Sorry, sorry. Weather modification activities on July 2nd? It was. You guys. Was there a bunch of other players operating? Is there generally a lot of players or is it a pretty. Is it. Is it a fairly small Number of kind of service providers that are participating in these programs.

Yeah, Jordy, you may have seen the prolific hustle bitch on x.com posting about this a little while ago. He said that I was the CEO of the largest and most powerful weather modification company in the world.

I saw somebody compare. Somebody was comparing weather modification tech to being saying it was more dangerous than nuclear bombs.

That was kind of crazy. And then I also saw some people just showing like general flight logs of like commercial airplanes. Like obviously there's a lot of south people.

I think people have every right to be angry and demand answers. It's such a tragic incident. But. But yeah, I'm curious to get into the scale of, you know, kind of maybe late June, early July, what was going on broadly.

Yeah, absolutely. So there's one other cloud seeding operator in Texas called Seeding Operations and Atmospheric Research soar. They're responsible for operations over the Rolling Plains Weather Modification association, which is significantly farther northwest of kerr County. On July 2, we conducted one 19 minute cloud seeding flight where we released about 70 grams of silver iodide and 500 grams of salt, table salt that was released at about 1600ft above ground level into two clouds that dissipated over the course of two hours after seeding them. The amount of time that those aerosols could have been suspended in the atmosphere is less than the time between when we were seeding and the onset of rains from the remnants of Tropical storm Gary. And the amount of material that we dispersed could not come anywhere close to inducing the precipitation, the 4 trillion gallons of precipitation that did come from that event.

So yeah, and I'm assuming you guys like have records. You keep records of like the radar showing these different cloud formations. So it's not just we looked and we think it dissipated, but it's like you can actually, you have like, you know, basically a map that's live updating. Is that the right way to think about it?

Not only do we keep records for our own research purposes and operational purposes, but we're required to keep records by the Texas Department of Licensing and Regulation. And those are accessible online, as are the reports on our seating activities. And if anybody is interested in those, then you can ask for them from the tdlr.

I'm curious when, when the flooding happened in Dubai. I want to say it was a year or two ago. Dubai is known for their cloud seeding operations. It's very dry place and makes sense why they would want to increase precipitation. A lot of people maybe the same types of accounts that have been that have been blaming, you were quick to blame it on cloud seeding. Throughout history, has there ever been any major kind of flooding event that people were able to say, yes, 100% this was caused by weather modification activities, or is the tech not even powerful enough yet to do something like that?

So I think that there's probably three points to touch on. The first of which is that it wasn't until 2017 that attribution had been physical. Attribution of cloud seeding's effects had been seen and proven in an academic context. And so with new advents in radar technology, namely dual polarization radar, we're able to much more clearly monitor what the effect from cloud seeding is. In previous operations, it was extraordinarily difficult to see what your effect was because we could not measure the cloud dynamics and the cloud microphysics that were changing as you were seeding. So that's the first point. The second point is that, and again, I'm trying to be, and will continue to try to be maximally transparent about our operations and historic weather modification. There was something called Operation Popeye during the Vietnam War where the deliberate intention of cloud seeding was to cause precipitation that would like, cause flooding and then impede supply chains on the Ho Chi Minh trail. Now the extent to which that was effective because we didn't have good satellite imagery or dual pole radar is outstanding. Now that said, lastly, third point, we have suspension criteria that are given to us not just by the TDLR in Texas, but every state wherein we operate. Because if there already is too much saturation of the soil, or if there is an oncoming severe weather event that the National Weather Service has notified us not to seed, then we ought not do that to increase the severity of precipitation. So there are suspension criteria because there are limits on what we ought to do with this technology so as not to cause flooding and only reap the rewards from it. Right? For our farms, for our ecosystems, and for, for our national security interest as well. Right? Like if we don't have access to weather modification technology, if we don't regulate this at a federal level and ensure that there's accountability and attribution for these activities, then other people, other nation states could be conducting weather mod in the vicinity of or on American soil without any accountability. And so that's why I am advocating for way more regulatory scrutiny from the federal government for cloud seeding and weather modops.

Walk through some of the history of the Chinese weather modification strategies. We heard about the flooding in Dubai that was kind of unclear. Have there Been any notable or confirmed negative outcomes from China spending? I mean, you said $300 million a year, something like that. That seems like a lot of cloud seeding. It seems like if there was a surface area where there could be mistakes made, they would have kind of explored that. I remember the pre Olympics they were doing cloud seeding or just kind of bringing down like the, the dirt in the atmosphere. And you know, people kind of learn from that. Okay, you get acid rain when you do that in, in particular. But have there been any case studies from China that we should be learning from in America?

Case studies from China with adverse weather coming from their cloud seeding operations?

Yeah, anything like that. Something where, like, okay, they, they've done a lot of this, push this to the limit. They've put. They've done this at scale. If there's going to be rough edges or mishaps, I would have. I suspect that we would have seen evidence of that over there. They would have had an accidental flood or something like that happen over there. If they're doing it at scale, you.

Would expect to have seen it from China. However, you would also probably expect and understand that they're a relatively inscrutable country that's does not report on their activities very openly, objectively. Now, that said, one thing that we do know about the weather mod program that they do have going is that they're planning to build 100,000 ground generators on the Tibetan plateau. So Rainmaker is primarily using drones for operations. We also have inherited some ground generators from previous operations. These are essentially aerosolizing units on the tops of mountains that can disperse material into clouds when the clouds intersect those mountaintops themselves.

Is that like a cannon that fires the material into the cloud or.

No, no. You might recall my, my initial inclination to use something like that because it is used in China. But no, it's. It's essentially like a smokestack of sorts, A very small smokestack that releases those airsols there.

Sure.

But in building 100,000 of these ground generators and also using the wing long two and a bunch of their other military drones for aerial cloud seeding, they're turning Tibet into a reservoir, a snowpack reservoir of unprecedented scale that will feed more water into the agricultural basins in southern and eastern China. And I think that, you know, although, again, this is something that needs to be transparently reported on and regulated and depriving American farmers in the west, especially as a congressperson from Georgia. Right. Where there is not as severe reliance on cloud seeding to produce water, would be against America's interest.

Jordy.

I guess I'm trying to. I mean, my question is it feels like, it feels like candidly it will be hard to come. It'll be hard to find any type of allies in Texas, on the ground in Texas, maybe, aside from the farmers. But I'm curious the various different groups, what the reaction from them has been in terms of if they're. The reality is water scarcity affects every person in Texas, but only a few people truly feel it. It's a much smaller group because everybody goes to their sink, they turn on the water, they turn on a hose outside, they go to a grocery store, there's water, there's produce. It's not something that people necessarily feel. And so I'm curious where you obviously are going to defend weather modification because you believe in the many different ways it can have a positive impact. But I'm curious who you think the other players that will be on your side as the industry. I mean, the industry was not in a good spot prior to this. It's in a much worse spot now. And I know you've been flying all over the country making sure that it doesn't get banned. So I'm curious what, what you think the kind of coalition that will kind of form around you.

Yeah. Yeah. Well, so I actually think I'd just, from my own experience over the course of the last few days, disagree with the two points that you made.

Right.

Like it.

It has neither been hard to find allies for cloud seeding weather modification in Texas, nor do I think, think the technology and the industry is positioned worse now than it was prior to this weekend. And regarding the first point, there are some people that I think are probably not in good faith engaging with this because they have some preconceived notions about chemtrails or otherwise and don't themselves want to scrutinize the data to back up how our operations are different and beneficial. Whereas chemtrails, as they believe them to be, are malevolent. The vast majority of people that I've interacted with online, on the phone and in person are rightfully curious, skeptical, concerned, some more than others, obviously. But in scrutinizing the data and having these conversations and learning about what cloud seeding is, pretty unilaterally people are supportive of it, provided that there is a regulatory framework more stringent than the one we have now that ensures that it's safe. This is true both of just individuals that are not themselves farmers, but obviously farmers, water managers, government officials too. I welcome any questions that people do have, both online and via email, about what our activities are, what our policy recommendations are. And I'm grateful that there are a lot of people that understand, one, our operations did not contribute to the flooding, but two, that even if there was a flood now, it doesn't mean that there is always enough water. And having access to a technology to produce more water for farms and otherwise would be beneficial. Like, people want a more green, lush country.

Yeah. I'm curious. I'm sure you've spent plenty of time thinking about this, but would there be a way to apply the existing technology you have almost in a defensive way and, you know, theoretically exceed a hurricane while it's still offshore, something like that, or, you know, one of the issues here, there was just so much water in the atmosphere that rolled over a heavily, you know, populated area, and then it's got its gravity right. It's got to come down. You know, is there an application of the technology that could, over time, strategically prevent, you know, or act defensively against the conditions that create flash floods?

It's. It's a very worthwhile question for you to ask and for us to ask ourselves collectively right now. Again, Rainmaker only does precipitation enhancement operations for all those constituencies that I listed before. However, in the past, the United States government funded Project Storm Fury, which was a series of attempts to reduce the severity of hurricanes over the Atlantic before they broke against the Eastern Seaboard. Again, we didn't have the appropriate understanding of atmospheric science or the radar or the satellite data necessary to appropriately do that. However, severe weather is something that is like a geopolitical risk, a national security risk. It causes damage, and it is fundamentally a physics problem. Right. A physics and chemistry problem. Is there technology now that could mitigate severe weather like this? No, and Rainmaker doesn't have it. Is it possible to someday, provided we invest in noaa, in the National Weather Service, in the appropriate research into cloud seeding such that we could reduce the severity of severe weather? Absolutely. And I am entirely in favor of that, provided it is done in a responsible manner. And if we were to ban it wholesale, then not only would we lose access to Precipitation Enhancement, but we would lose out on any potential of, at the very least, better forecasting for these systems and warning people early, but also the even greater and more consequential beneficial potential of reducing severe weather in the future. And so I think that the United States government and Rainmaker should and are absolutely interested in mitigating severe weather in a manner similar to Project Storm Fury.

I think the pr. What you were getting at, Jordi, like, the PR difficulty here is that, like, when there's not enough water, crop yields are lower, prices go up, but it's very distributed. Everyone feels it a little bit. Whereas when there's too much water and there's a flash flood, individuals die. You have a very. It's a very emotional, very. It's very concentrated. The pain is very concentrated. And so that's why this, this story.

Yeah. I mean, normally when, Normally when there's a natural disaster. Yeah, there's. You can, you can critique the government for their response to it, but there's not somebody sitting there. A scapegoat. Right.

And so I guess the question is, like, easy.

Yeah, it's, it's, you know, whether it's online accounts that are just engagement farming or it's a politician.

Yeah.

You know, the concern is that. And your concern is that the industry becomes a scapegoat and America loses a capability that our adversaries clearly care a lot about.

Yeah. My question is, like, we're seeing this bifurcation. It seems like Ted Cruz came out in support of the idea that Cloud City had nothing to do with the Texas floods. Marjorie Taylor Greene is taking kind of the other side of that. My question is like, these are politicians. At the end of the day, they're not independent scientists. Who can we go to? Who can the population go to for, like, a truly independent review of this situation? Like, is there, is there some sort of independent governing body or are there, are there respected scientists that kind of don't have a financial or, you know, political incentive one way or another? How do you think the, the, the populace should be satisfied? Obviously, you're telling your side of the story, you're going direct, you're explaining things, you're laying out the data. But what do you expect people to look for in an independent analyst? Yeah.

Yeah. So for one, I think that noaa, the National Weather Service, the National center for Atmospheric Research, all of those are great third party entities that can review the information, corroborate the information that we've provided. Provided, of course, that they continue to exist and remain funded.

Sure.

I think that this probably demonstrates why it is important that we should retain some capability nationally to forecast and research the atmosphere, because there should be somebody that's capable of reviewing this to ensure that it's safe. I'll also say, you know, regarding the scapegoat dynamics that exist right now, I've thought about this pretty prayerfully and intently over the last few days. And when There is a calamity of some sort. Like I've been trying to think about why people are say coming after Rainmaker or angry at Rainmaker. And I think that when there is a calamity of this type, if there was someone responsible, if there was someone or something that could be held to account, then in holding them to account, you could supposedly prevent this kind of thing from happening in the future. The trouble with a true natural disaster as this was is that there is nobody to be held accountable. And that makes the world a lot more tragic because it means that things like this will persist, they will persist indefinitely into the future unless and until some sort of technology could reduce the severity of severe weather.

Yeah, we went through this with the California fires. It was like everyone was searching for a single person depended on and like it came down to like, you know, some people built their houses the wrong way and there's some building codes that need to change and there's some water rights and water flow and there's some different general government like we need more goats in certain areas. There's like a million different things that could have prevented this if they were all working together as a well oiled machine and had the forethought. But it's a very, very frustrating and difficult situation situation. So our thoughts and prayers are with everyone who's been affected. But thank you so much for stopping by. This is fantastic. Thanks for breaking it all down for us.

Thanks guys.

Appreciate it.

Cheers.

Really quickly before our next guest joins, let me tell you about Linear. Linear is a purpose built tool for planning and building products. Make the system for modern software development streamline issues, projects and product roadmaps. Let me also tell you about numeral sales tax on autopilot. Spend less than 5 minutes per month month on sales tax compliance. Go to numeralhq.com I use it personally and we have our next guests. We're doing two guests simultaneously. I hope production team was made aware of this. It's on the calendar but we are working to get a four up display for you. We have the founders of Grammarly and Superhuman. This was an acquisition that was that was announced last week and lots of people in tech have probably used Superhuman. It's the email client of choice for the tech elite of course and Grammarly has been an indispensable tool at tvpn. We install it on our producer Ben's computer very early on when we were posting clips to make sure that we didn't have any spelling or grammar errors. When we would post on social Media, very useful tool. And now they've combined forces and we're going to talk about how we can, what the shape of the business will be going forward, how these products play together, what the modern suite of tools looks like going forward. Welcome to the stream. How are you?

Great. How are you?

Fantastic. Would you both mind kicking us off with an introduction on yourself and the companies that you run and then we'll talk about the acquisition.

Sure. Rahul, you want to go first?

Okay, yeah, happy to. Hello everyone. I'm Rahul Voor. I'm the founder and CEO of Superhuman, which if you're not familiar with, is the most productive email app ever made. Imagine getting through your email twice as fast as before, responding faster to the things that matter, and saving four hours or more every single week. We're also inventing the future of productivity with AI. Imagine waking up to an inbox where every email already has a draft reply you would simply edit and then send. And of course, with gramly, we are going to build the AI native productivity suite of the future.

Great.

Amazing. Yeah.

I'm Shashir Motra. I actually started a different company. I started a company called Coda about the same time as Rahul started Superhuman. And about seven months ago or so Grammarly acquired Coda and I stepped into the CEO role running Grammarly. So been working in and around the industry for a long time. Before Coda, I used to run the YouTube group at Google. Worked at Microsoft in the early days. Actually started my very first job was working on Outlook. So I've got back in 1998 I worked on email. Fun to come back to it.

Yeah, that's amazing. So I'm not sure who's best to answer this, but I'd love to know about how this deal came together when you two first met. And you know, we keep going back to this post that like you'll meet your acquirer like five years before you, before the deal goes through. Is that this case? Is this a narrative violation? Kind of. How do you get to know each other? 8 years? Break it down.

Why don't, why don't I take the first half and then Rahul can take the second half so I can talk a little bit about the deal and what we're doing and Rahul can give the fun origin story. So maybe just as a little backgrounder for everyone. So Grammarly, at Grammarly, our goal is to build an AI native productivity suite with the agents and applications that drive productivity for every individual and team in the world. So a lot of that is probably new to people because people have generally thought about Grammarly as a much narrower product than our aspirations for it. And the way we generally talk about it, and I'll talk about agents first when you talk about applications, but we generally think about Grammarly as the OG agent. It's about 16 years now that the company has been helping at this point, about 40 million daily active users, where Grammarly is the communication assistant that lives right next to you in every surface you work in. But people misunderstand the technology because they think it's about grammar. But actually, the technology of Grammarly is mostly about bringing AI right to where users work. So we can work in about 500,000 different applications where we read what's on your screen, we can annotate it in an unobtrusive way, and we can make changes on your behalf. And so from that perspective, we call this layer, we call it the AI superhighway, bringing AI right to where people work. And in that analogy, up till now, we've only been running one car on that highway. That's the car with your high school grammar teacher in it. And that's a very useful car. And it generates over 7,800 million dollars of revenue now. But I think it's a vast subset of what you should be able to do with that. And so a big part of our strategy is opening up Grammarly to become a platform so you can build any sort of agent on it and have those agents come to you where you work. That's the first part of our strategy. Second part of our strategy is taking those surfaces and building the first party versions of the surfaces that we think really matter matter. The surfaces where people work every minute of every day, where all the work really gets done, where you really want to work, not only alongside humans, but alongside agents as well. So that's why we bought my prior company, Coda, that we make an all in one document solution that blends documents, spreadsheets, presentations, applications into one surface that produces all the work artifacts for you. But another key part of work is communication. And for many people, the dominant communication tool they use is email. It's something like three to four hours a day the average person spends in their email inboxes. And this actually shows up in the Grammarly stats. Really high. So email turns out to be the number one use case of Grammarly. We revise something like 50 million emails per week. It's three of the top 10 applications that Grammarly is used in are our email clients. And so we saw that as an Obvious place to go to go work next. Now from my perspective, I think email is a category that is particularly ripe for disruption. As I mentioned, I started my career working on outlook in 1998 and since then there's a round of innovation with Outlook, there was a round of innovation with Gmail, and then there was a decade of not much. And then Rahul Superhuman came along and built a great email experience. So when we went looking for which surfaces really matter, we landed on email. And when you look at the email category, as you mentioned, there's only really one player that's meaningfully innovated in that space and that's when we call Rahul. That's a little bit about how the deal came together.

Amazing.

Great.

It's funny, just for some added context, I basically have been lucky that my entire. I think I got on superhuman in 2018. You launched in. When did the beta launch? It was 2017.

I think our first paying customers were at the very end of 2017. So you're really.

Yeah, exactly, exactly. But I graduated college in 2018, so as a professional I've only had experience. I'm a lucky. I'm the lucky batch, right? Yeah. And still use it today. So thank you for, you know, I never had to be an Outlook guy.

I had a Yahoo.com email address when I was a kid. Yeah, vintage.

Vintage. Anyways, Rahul, I don't know if you have anything to add but then I have a bunch of follow up questions on the last anecdote.

Yeah, for sure. I think it'd be fun to tell the origin story of the deal, how it came together and I think there's a lesson or two in here for other founders or the entrepreneurs listening. So I'll also try and make it useful. So to your point, the foundations, the seed for this deal was planted many, many years ago. It was eight years ago like Shashir said it was back in 2017 and back then he was the co founder and CEO of a company called Coda and we were actually at a conference together in Hawaii. So it was really nice. And that said, I didn't really want to go. This was a four or five day thing, accounting for travel there and travel back. And one of my co founders, Vivek Sodera was really encouraging me to go. He would say things like, listen, building a startup is just as much about who you know and the connections that you have and being able to pull opportunities together as it is building and marketing a great product. And I'd be like, well I want to work on this feature. I want to do this thing. But in the end, he just pushed me out of the office and put me on a plane and go to Hawaii, which sounds weird, right? Like resisting going to Hawaii. But anyway, there I was. Shishir was there as well, and it was one of those special moments where nobody else was around. So it was just the two of us by a pool, two productivity nerds nerding out about productivity. And he told me that he'd worked on Outlook back in the day. We got into some really deep conversation about productivity. And as you know, back then we only did one on one VIP concierge onboardings. You must have gone through one yourself if you onboarded in 2017. So I onboarded him right there and then right by the pool. And those who've gone through the onboardings know one of the very last steps is when we ask you to close Gmail. And so I was asking him to move the mouse over to the Gmail tab and close it. And when I asked him to do that, another tab caught my eye, which was an app called Krypton. So I asked him, what is that thing? And then Shashir then proceeded to give me the best product demo I had seen in years. My jaw hit the floor. It was a document, but it was also a spreadsheet. It was also a database. It was a collaboration tool. It was was a mini app builder. And maybe today we take these things for granted, but back then, in 2017, this was truly mind blowing. And so Krypton then renamed to become Coda. And Coda, of course, late last year, joined Grammarly. Now in his acquisition announcement, he wrote, and I have the quote here. As I watched the foundational capabilities of AI change, how just about how every tool and surface operates, I started drafting my 2025 memo for the team. I titled it the AI Native Productivity Suite. And this just set a whole bunch of bells off in my brain. In a good way. Because at Superhuman, our vision has always been to build the AI native productivity suite of choice. And email is obviously a critical part of that. It's a much bigger problem than most people realize. There's roughly a billion professionals in the world, and on average, we spend three to four hours a day in email. So that's 3 billion hours every single day, or north of a trillion hours every single year. We actually all spend more time in email still than any other work app. So we caught up early this year in January, actually a few days after he became the CEO of Grammarly. And over the course of several conversations it became very clear that we were working towards the same vision which is to build this AI native suite for apps and agents. And then as Shashir said, email sits at the heart of where Grammarly is used. Today. It's the number one use case, helps write more than 50 million emails. Another stat that I found very fascinating is that 17% of words accepted on gramly are actually accepted in an email service.

Wow. Okay, bunch of questions and I'm sure I'm excited to get your answers. So first is integration. How do you see sort of the. How do you see both the brands and the products integrating and working together over time? Because you have a great challenge of having three great products that people love and three brands. And in order to deliver on this true long term vision of a productivity suite, I imagine over time you want to integrate them deeply. I'm curious what that looks like.

Yeah, maybe Rahul, you want to cover the product part and I can talk about the brand part.

Sure, yeah, I'll do product really briefly and then we can go as deep as you like. I think one of the most exciting things about the deal from a superhuman perspective is the access to significantly greater resources. So you can expect us that we'll invest more than ever than we have done in AI. We're also absolutely not done with our core email experience. We'll be doing a lot more there. We're going to build out calendar and tasks and then connect those beautifully together. We'll also start to spread our wings beyond just email. So we're going to reimagine chat, we're going to redefine collaboration, pulling on everything that we've learned over the last 10 years about work communication. And then as Shusher mentioned, we are also working on a whole new way of working with AI agents. Agents that we think will free all of us up to be more creative, strategic and closer to achieving what we call our human potential. And then just to double click on that a little bit more, we really think we're entering the age of agentic computing where AI agencies, agents, they're going to work on your behalf, they're going to reason, they're problem solving and they're incorporating detailed context about your work. They're actually also interacting with other systems and agents. And I think we're beginning to see these in some of the products that people are using today. And for so many people, email is just at the center of where we work. You think about project statuses, customer communication, meeting updates, deal execution. So much More, it all actually funnels through email, whether it's. Whether it's a system of record or that's actually where the work is taking place. So we also think that email is the perfect place to deploy a collection and a suite of agents. You can imagine an agent triaging your inbox before you wake up. You could imagine another agent drafting responses in your own voice and tone, incorporating context about you and from your work. And at the same time, another agent is surfacing Insights. They're scheduling meetings. They're syncing with your other systems of records and your other agents. And I'll give you a specific and concrete example. This is something you can actually do in Superhuman today. And then I'll talk about how it's going to evolve in the very near future. And let's talk about search or asking your email things. For over 40 years, we've had to rely on what we kind of hilariously call search. But if you think about what that is, you have to remember senders. You have to guess keywords. You don't have to scan subject lines. And now in Superhuman, you can simply ask, where is the Q3 offsite? Or what are my flight details? And a very real example that blows people away whenever they see it is this thing I do whenever we launch a feature. Whenever we launch a feature, you'll know this. Using Superhuman, I send an email to every single person who uses the product. And then we get a whole bunch of replies back, usually several thousand responses. And I still personally read through every single one. I reply to some of them, but what I'm doing is I'm copying and pasting my favorite quotes into a Google Slide that I can then present at the next company. All hands. Now, this takes half an hour to do properly with Superhuman. Today, you can just ask, what are the top 10 most positive customer responses to the calendar? See your week launch, let's say. And then boom, boom, boom, boom, boom. Immediately, within 5, 10 seconds, I have the answer. So that's taking what is a half an hour task and making it work in five or 10 seconds. Now we're evolving that so that you can then continue the conversation and take it much beyond email. You can imagine me then saying, okay, I want to convey the magnitude of the commercial opportunity to the team. So can you annotate each quote with the name of the person, the company they work for, the size of their current Superhuman account, and the total number of employees. Employees they have, and then compute an estimated size of price, like, how much revenue is there at stake if we were able to sell into that company. And you can then imagine the superhuman set of agents figuring out what to do with that. Realizing the answer actually isn't in your email, it's probably in your CRM. So a sales intelligence assistant is called into the mix, there's a handoff between the agents, and then the answer is right where I kicked off the conversation in my email app where I happen to spend three hours a day. And you can continue the conversation. You can then say, okay, let's please turn this into a presentation. And then perhaps it works with, let's say the Gamma agent to produce an amazing, beautiful presentation in your own brand for the company. And then you might say, okay, I want time to practice this before the all hands. Can you please schedule time in my calendar to do so? The agent's like, well, you're completely booked up before the all hands, but we can move some things around. And it's smart enough to know that it's easier to move a one on one than it is to move a team meeting. So it recommends moving the one on one. It goes and does that. And now you have time blocked in your calendar to learn a presentation that was created for you in 10 seconds by this agent that just read thousands of emails to get the content work that literally would have taken an hour done in, let's say, one or two minutes. So that's the kind of future we're working towards.

Wild, there's going to be 250 agent startups that are going to hear that and be like, damn, he's. They're doing, they're doing what I'm trying to do.

This is an ecosystem.

No, no, no. And that's to be clear.

We want our marketplace to be the place you deploy those, those agents.

Right.

Whether it's the surfaces you want to work on. If you like, I could talk a little bit about the brand question as well. Yeah, so, and you know, I think my experience here is heavily formed by. Before starting Coda, I ran the YouTube group at Google and I think that was one of the best examples of an acquisition that I think flourish in a way that would not have been possible without the particular constructs we put in place there. And there's a lot about that I think that we got right. And I think I'm going to mimic a lot of that here as well. So my goal is with Grammarly. We're going to build the AI native productivity suite of all the apps and agents that you need, some of them that we'll own, some of them that we will be great partners with. But it's really important that each of those retain an identity. And I think that's important because that's how they keep innovating. I mean, if you think about, you know, you started using superhuman in 2018, you know, you were buying into a product, but you're also buying into a team and a vision and a feel, and all those things really matter. And so I really like this term of building a compound startup where each of those products still feel like they have an identity. They have a brand, they have a mission, they have their. They have similarities for things we want to work across, but they have their own perspectives on the problems that make sense in that space as well. So we want agents to work across all surfaces. It's very important that if I set up my sales agent, that it should be able to do some of the experiences that Rahul just described while it's in my email, but while it's in my document, it'll give me a different set of experiences when I take it out and use it. While I'm using a third party application, it should still be able to bring my contacts with me. So there will be things that need to feel similar, but the individual brands will remain separate. The last thing I'll say about that is the overall corporate brand for Grammarly will change. We're working on a new name for it. So Grammarly will become one of the sub brands itself. We think about it, as I was describing, one of the most important agents in that. In that platform. So the new brand's coming. I'm really excited about it. But not announcing it yet.

Yeah, yeah, I can't wait to see it. I'm curious how you think about the tension between yourselves and someone like a Google workspace. I was joking with John the other day. It feels like so many companies are so dependent on Google Workspace for the core kind of just like team management infrastructure that they could just raise the prices every single month and it would take a really long time for even and to try to figure out something else. And it reminded me of kind of the tension that some of the foundation model labs have today with the app layer above them. Although that's quite a bit more intense. But I'm curious how deep down the stack you guys would go. If you can talk about it or if it makes more sense to focus on the agent layer or the app layer.

Maybe I can start. It's interesting, the three products we're bringing together, Grammarly, Coda and Superhuman all have competed with the Google Suite or the Microsoft Suite for years. And so I think we're all horn in the fire.

Yeah.

And the thing I'd say about it is it actually comes up less with customers than you would think. I think that when companies decide it's sort of like buying plumbing for your company, you buy one of these suites, it covers lots and lots of different things, but these have become a part of the furniture at the company and people don't really think about them as their real investments in productivity.

I totally agree. By the way, as the way work evolves, imagining trying, setting up every time you have a. I could imagine a world where you're generating a new agent for a specific task and they have an email and I'm sitting here being like, do I really want to pay Google Workspace every time I spin up a new agent at Google Workspace, another $25 a month. And so I imagine you guys can take this in a direction that kind of reinvents all of that plumbing in the long run, but maybe it's not.

Yeah, I mean, I would say all three products have found different ways to be better together with the underlying products. Obviously with Grammarly, one of its hallmark features, that it actually works in all those surfaces. So it amplifies your investment in, you know, works great in Google Docs, but it also works great in Slack and in Salesforce and all the rest of your products as well. For Coda, it deeply integrates with those products. And then, and then for Superhuman, you know, Gmail or Outlook serves as the back end for those providers. So it's not really a question of less investment in those core infrastructures. But if you're users want the best possible experience for what they're doing, you're going to go get the best tools. And in a sort of macro scale, the amount of money you're spending is such a tiny amount of money compared to what, what you're actually investing in your employees to go stick another 10, 20, 30 bucks a month for people that you're, you're spending hundreds of thousands of dollars on to get them.

Some cases hundreds of millions, some cases hundreds of millions.

We're going to have to raise our prices for those employees. But yeah, you know, you're gonna get a huge return for them. And people don't really care that much about their sun cost and their plumbing.

Totally.

Question Two somewhat related questions. One is, I don't want to say that, you know, Grammarly is a Chrome plugin, but a lot of people experience that way and I noticed I was using a different Chrome plugin and Chrome, the Chrome App store like updated and I lost functionality because they changed their policy and this particular plugin wouldn't work in the new rules. And so I'm wondering if this was a plugin different. This is not Grammarly. This is a separate one. It was called UBlock origin. It would let me go in and select specific divs on specific websites and basically mute them every single time I hit that website. It was very, very cool. But it was deemed to be like, not privacy safe. And it was really annoying for me because I enjoyed the this and I was excited to use this thing and then I lost it and I mean I might be able to like download it and sideload it or something, but it was, it was difficult. And so I'm wondering about like sharp elbows in the. Because the Chrome plugin is an interesting wedge. Interesting go to market. It unlocks so many different things. We've seen this with like the OpenAI ChatGPT app, using the ADA or the accessibility features to kind of plug into any IDE on day one. Like you just have such an interesting ability to plug into, you know, tons of apps with AI in a bunch of interesting ways and like you're native there. But it feels like Google might be getting a little bit more sharp elbows there. Has there been any tension there? Do you think that there will be more over the long term? What are the risks to building on top, like building a platform on top of another platform?

Yeah, I mean, maybe just to two parts to answer this. First off, just to correct one misunderstanding, the Chrome plugin is a very big part of the Grammarly product. There's also a desktop application. There's also a set of mobile applications, so iOS and Android. And we have millions of users on each of those as well.

Yeah.

And so. But I understand that the product is synonymous in many people's heads with the, with the Chrome extension first. But that's very important because we have to work where users work. And sometimes you work in a web browser, sometimes you. You know, many people use Slack as a desktop app. App use Superhuman as a desktop app and so on. So you have to be able to work in those places. I will say that staying on that line of where these platforms are is kind of become the core asset of the company.

So that's what I.

That's kind of what I meant by people misunderstand Grammarly like that. I do have a team here that works on being a great grammar agent, but a massive team that works on how do we integrate with all these products in a safe and secure way? And one of the things we've realized is that we've done this just for the grammar agent, but what if we could amortize that across a much broader set of agents? And so now if you're. If you're someone building a new agent, you could go build a Chrome extension, a desktop app, and so on. I mean, I'll pick an example. Let's say I'll pick a book author. So, you know, I really like Kim Scott. She wrote a book called Radical Candor. We spent a bunch of time with Kim on. Right now, she sells a book, you stick it on your shelf, kind of forget about it. She wants to build an agent that sits right next to you and says, hey, you're not following the principles of the book right now.

Interesting.

Yeah. So I'll say it because I'm thinking it, but it seems like I'm gonna say this and then I'll provide some more context. But a lot of people have been very triggered by the marketing that Clulee has done. But at the same time, what they had surfaced and what you guys had basically started doing years and years ago was, was understanding what a user is doing on their screen and starting to surface information and help them take action.

Grammarly is like the original copilot.

And I think that what you guys are building towards, and specifically this app layer on top of this, like, private, secure way of servicing context, will in hindsight be incredibly obvious that that was how we should be integrating AI in our workday. Because the idea of, like, you're working in an app and then you go in another app and you type a little bit, and then you take that and maybe you go back into the other app and then you're just like tossing this over the wall. Makes no sense when things should just be getting constantly served.

This is how people were programming precursor. It was like, copy the code, copy the Python into chatgpt, copy the result back. And it was like, okay, there has to be a better way.

Yeah. And I don't want to. As a user, I would to love, love to be able to have a bunch of different experiences like that, but I don't want to trust. I don't want to. I don't want a hundred different companies to have full read access to my desktop screen and my microphone or any of these other things. So I'm very excited about where you guys are going.

That's exactly right. So it's tense to Build a platform on top of your browser, desktop, so on. But once we've done that we can now make it available to the clulies of the world, to the Kim Scott's of the world, so on and say why are you going to figure out how to integrate with every one of those applications and we can do that for you. You should focus on the logic of what do you want to suggest to the person and when.

Yeah, totally.

I mean a couple more questions wanted to fire off. You guys are well capitalized, you generate a lot of revenue. I'm curious how you're thinking about operating the business on a go forward basis. I'm assuming you're getting a lot of hopefully getting a lot of efficiency out of AI so maybe you can. Is the the plan to focus on innovating while generating cash flow or are you guys going to continue to or just burn and run a more traditional valid playbook?

Graham really has been lucky to be a cash generating business for a long time and so it's sort of built into the DNA of the company. So I think in the. I'd like people to start thinking about Grammy with the new brand that will announce soon enough. Think of us as one of those top few AI companies and if you think of the foundation model companies providing great layers for all of us, I think we're hopefully the suite of applications and agents you really care about with one big business model difference. We don't burn billions of dollars in order to do it and I think we can hopefully bring that to people in an efficient way which allows us to but to grow and expand in our own control.

Yeah.

Last question. Are you guys in? Are you guys talking to more companies if somebody has a great product that's generating a lot of revenue?

Yeah. I'm looking through the Google suite right now and you know I see video.

A lot of potential targets. Are you guys a buyer?

We are. I mean I think we should. I would love to talk to people with interesting ideas there. I think, I think there's a great opportunity here to go build that next AI native productivity suite. We will build parts of it, we will buy parts of it. If I looked at email for example we could have, if we started to build an email experience anything like superhuman, nobody would have seen anything for a decade. And so it was very important for us to get a jumpstart with the number one product on the market. There are other cases where I think we can build. I don't think we have to buy everything but I think we're A great home for startups that are lacking that sort of scale, that want that distribution, want to get to a much broader group but still want to work in an innovative environment. So, yeah, I hope we're a great spot for that.

Yeah. Well, it's exciting.

Congratulations. And come back on when the rebrand drops.

I want to see that. Excited about it.

Can't wait.

Congratulations.

You guys are going to cook up something great. That.

This is fantastic.

Cheers.

Well, we will talk to you soon. Have a great day.

Thanks for chatting, guys. Goodbye.

Let me tell you about Adeo Customer relationship Magic. Adeo is the AI native CRM that builds scales and grows your company to the next level. You can get started for free or you can talk to sales. That might be the cure for male loneliness.

Or I'll make an intro.

Yeah, just hit up Jordy. He'll introduce you.

I'll introduce you right to the top.

Let me also tell you about Fin AI, the number one AI agent for customer service. Number one in performance, number one in competitive bake offs, number one ranking on G2, G2.

And you know, I love bake offs. Do not get in a bake off with Fin AI. Just don't, don't even start. It's going to be a nightmare for you.

Yeah, it's going to be bad.

It's going to be really bad.

No one likes how that ends. Yeah. Anyway, our next guest is ready to join. Let's bring in Ankur from Kerry.com.

We like carrier.

Welcome to the stream.

Welcome to the stream.

I think we're, I think we might be on mute, but we're going to want you loud and clear for this one because there's a lot of exciting news.

There he is.

Good to meet you.

What's going on?

What's up? How are you? Good to be here.

We're doing great. I'm excited to chat. Yeah, I'm sure you've been.

It's boring, boring tasks stuff, but anything here for it?

No, you make it digestible.

So the goal I was talking to.

It'S like the last threads that I see on Axe that I'm like, thank you. Thank you for making this thread because it's actually, it's actually deeply researched, thoughtfully organized and valuable. And it's not like, have you ever heard of Mark Andreessen?

So, yeah, the goal here is to create something evergreen, the definitive playbook for founders. So I think we want to create something that can be a resource for a long time. Ideally.

Yeah. Great.

Let's break it down.

Let's do it.

All right, where do we want to start?

Yeah, I figured it would be helpful to kind of walk through our audiences pretty evenly split between early stage founders, operators, and executives, and then. Or just startup team members in general and investors. And so getting kind of a lay of the land on how the big beautiful bill impacts all those different groups would be awesome. But maybe first, I would love some background on yourself, how you got into this, Carrie, and then we can get into all that.

Yep, Sounds good. Well, Jordi's an investor, so has a little.

Not for me. Not for me, not for you, for everyone else.

Yeah. But I've been running a company called Carry for almost three years now. I'm an immigrant to America. I moved here knowing zero about personal finance, zero about taxes. I sold my company five years ago, and I was facing a giant tax bill, so I hired very expensive lawyers and accountants, and they were able to do black magic to basically reduce my tax bill dramatically. And it made me realize, like, the tax code in this country, it's kind of. There's so much stuff in there, but very few people actually know how to leverage it. So when it was time to start a new company, I spent, I don't know, couple of months looking into this and made it my mission to, you know, dive deep into everything in here. And what we do at Carrie is we're like, can we build software to give people. We call it Tax Alpha, but basically ways of saving money on taxes, on autopilot. So, again, my compliance team is going to make sure I say this. This is not tax advice, legal advice, or investment advice, but never give that.

Kind of advice on this show.

Yeah, yeah. But I have spent a lot of time in the last two, three years working with, at this point, thousands of business owners. And I think I have a pretty good idea of, you know, generally how business owners can save money in taxes. And this piece of legislation is the most significant one we've had since 2017. 2017. There's.

And before. Before we dive into that, I think it's helpful. I feel like you approached the tax code like very much like an engineer, and in the same way that, you know, if you sign up for a software product, you're getting the benefit of that company spending millions and millions of dollars, like, building this product and then giving it to you at a fraction of what it costs to. To create. With Kerry, the idea has been, how do you kind of create that same effect in some way for taxes? Because if you're working with one of the best CPAs in the world, they will charge you for the service and then they'll go charge you the same price to someone else for that same service. And they'll just do that a bunch of times. And you guys created a different, you have sort of a different incentive which is, is how do you create the maximum amount of value and then make it available to as many people as possible, which is kind of the traditional software playbook applied to a new category.

Thank you. You pitched my company better than I did. But yeah, there's all this stuff. I mean there's so much nuance in it. But from a software perspective, none of it is specifically hard. The challenge we have is we deal with fintech, we're dealing with real money, custody, real assets. That's the complexity. But there's so much stuff in the tax code that we're, if we're focusing on, I don't know, 1% of what's out there. But there's like I generally believe for most people I know VCs probably disagree with this. There's no alpha in investing. The average person should just index the market and get to work. But you can find alpha by saving money on taxes. If you can save 10, 20% off the top so you have more dollars to index the market. That's basically the thesis behind what we're building.

Yeah. So talk about kind of maybe how the bill came together, what you expected to be in, what maybe didn't make it in. There was a lot of chatter. Maybe was it four months ago around the carried interest loophole. People were pretty triggered by that. I was triggered by that. But it sounds like that didn't make it in. But yeah, break down kind of maybe the lead up to the bill and then how it actually ended up getting implemented in its end state. Yeah.

I should also caveat by the way, that like I'm going to tell you all what's in the bill. It is not an endorsement of the politics behind it. Like you can argue either side of that. Like that is out of scope for what we're talking about.

We're just talking about reality today. What, what is absolutely becoming law.

So yeah, so 2017 there was something called the TCGA tax cuts and Jobs act where there were a lot of temporary measures that benefited groups of people the administration wanted to benefit. Typically this was entrepreneurs, business owners, investors and real estate developers. Part of this was there was a lot of short term measures put out for eight that were only going to last eight years in the future. But what this bill has done is it's made most of them permanent. So there's a lot of things like, you know, if we're talking about startup founders specifically, there's. And we'll break them down. There's many things in this that will make your life better. Even if you're a sole prop LLC or an S corp, a bunch of things make this better. If you're someone that's coming up against the estate tax, this bill helps that. So lots and lots of good stuff.

Maybe you could kick off with a little bit of the background on, like, the understanding of qsbs. Like, for the last decade, I feel like the rule of thumb has been like, you start a company, you sell it for a bunch of money. The first 10 million, you don't have to pay federal taxes on it. So if you're in California, you're still going to be paying California tax potentially. But you might be able to think about it as like that if you get a $10 million liquidity event, you're basically taking close to 10 million, potentially.

10 million in New York. 10 million.

Yeah. New York, 10 million. And so. So you don't need to move to Puerto Rico. Good news for that.

But that was a funny time when people were so obsessed. I gotta move to Puerto Rico.

Now you're in Puerto Rico.

Oh, you're having a massive liquidity event.

Oh, okay. Yeah. But yeah, yeah, talk through the reality. Like, how real was the original QSBS process? What were some of the hiccups? If it was an acqui hire or an asset sale that might trigger income tax or something like that. And then talk to us about what's changing.

Yeah. So qsbs, for those that don't know it. I mean, qualified small business stock. This is what kind of got me down this whole path. I mean, I was running my startup for six years. We were about to sell the company, and I didn't know about QSPs. It was the best surprise when my accountants are like, guess what? You actually could not pay taxes on $10 million. I was a resident of New York, so no state tax as well. Yeah, but not just that, you. The qsbs limit is per shareholder, So I can give shares to my brother, my parents, and now your $10 million becomes $40 million. You can set up trust as well to multiply it. So it already was exceptionally generous. And it's available to every shareholder, investors and employees. Though sometimes employees don't hit the threshold since you have to hold shares typically for five years to unlock the benefit. Right. Five years is A long period of time. One of the big changes this bill brings about is now if you hold shares for only three years, you get half the exemption. And if you hold shares for four years, you get 75% of the exemption.

Wow.

So that's one big change.

And just to level set here, like the whole idea behind this particular tax incentive is to incentivize innovation and building new companies. Small business creation, opposite of like high frequency trading. And it is, you have to create value materially.

The people that are primarily like the average person that benefits from this is somebody who starts a plumbing company, runs it for 20 years and sells it. And is that right or is it. Is it happy?

I think that's the intent of it. I think the reality of it is Silicon Valley benefits from it more than anyone else. I think the intent and what's actually happening. But again, that gets into the politics is a little bit different because technically services businesses are not included.

Yeah, you have to.

But the reality is if you look at most big tech exits right now, people are paying substantially less in taxes. There's a New York Times article with the Roblox founder. He set up 12 different trusts to multiply QSPs to $120 million. He actually joked that raising a kid in California is so expensive that the QSPS exemption is what makes the whole math worth it. It's kind of an insane thing.

That's a crazy thing to say. Yeah.

What types of small businesses? What are all the different types of small. I mean, just generally what are the different categories? Because if you take out services, the software doesn't apply to that. Right. You can just be building regular SaaS.

So typically, typically the requirements for KSPs are a few different things. And one of them is changing now is one, you have to be a C corporation. So that's historically been like L.L.C. s corps don't count. You have to be a C corp and hold shares for five years. When you acquire the shares, the company should have less than 50 million in assets. That was the old rule. Now it's 75 million in assets. For startups, that is typically the cash raised, not the valuation. So it takes you pretty far. Right. Like before, you raise 75 million bucks, you get pretty far. And then there's other stuff. It has to be an active trade or business. And there's a few disqualifying categories like services or something based of someone's brand does not count. But the way QSBS works is it's ultimately a stance your accountant takes. So as an example, let's imagine you're a tech enabled service business. You could find a lawyer or an accountant to take a stance that QSBS counts and there's a good chance it just works out that way.

That makes sense. What besides QSBS has changed in any meaningful way?

Yep. So again, just to reiterate, the other big benefit is the $10 million limit per shareholder is now 15 million. So three big changes, 10 to 15, there's now partial QSBs, and you can now be up to 75 million in assets outside of QSBs, I would say.

Is that backdate at all?

No. So only for companies incorporated from Friday or you have to buy the shares from July 4, Friday onwards.

Wild.

Oh, wait, so this only affects going forwards?

Yeah, going forward. But new share purchases would count.

Yeah.

Okay, so if you buy shares. But theoretically, what I'm actually not sure about is probably maybe a lawyer can weigh in. If I'm an employee who has options and I exercise my options today, would that count? There's a chance it could.

Interesting, interesting.

So if you own 20% of a business started in 2019, you hit your five years, the company sells for $100 million, you get 20 million, you're still at the $10 million QSBS exemption.

$10 million. Unless you set up a trust or gift shares to someone else.

Sure, sure, sure. Interesting. Can you explain this bonus depreciation concept? Yeah, absolutely.

Bonus appreciation. I think you already have the private jet dude on.

He's coming on after this, so he'll talk about.

It's big for his business too. But basically, basically the way depreciation works is when you buy any kind of physical asset, like considerably buying a commercial building, it loses value every single year. Every year. You can take that loss of value as depreciation. It's a phantom loss in that you're not losing money, but you can deduct it from taxes.

Sure.

What bonus depreciation lets you do is it lets you front load depreciation for typically things that have a useful life of less than 10 years. You can take all of the depreciation upfront. So this is really significant for all the real estate bros out there because what they can now do is you can buy a building, you can do something called a cost segregation study, which will take the building, it'll break it down into all of its components. It'll be like the H Vac is worth this, the windows are worth that, the doors are worth that. Anything with a usable timeline of less than whatever I think it's 10 years. You can depreciate upfront. So the upshot is you can buy a commercial property for. For million bucks, 2 million bucks, put 20% down, but also get a 20% tax loss so you can deploy cash much faster. And people think this could lead to real estate prices growing, but this applies to private jets, heavy machinery, cars, all kinds of equipment.

Makes sense.

What else are you tracking? Anything else?

Estate tax is big, right? Estate taxes are massive. I mean, historically, when you die, anything above the estate tax exemption gets taxed at 40%. This bill makes it permanent at $30 million per couple, which is a very, very high threshold. That was actually supposed to go down to 10 million bucks. So it's a huge swing. And there's a lot of sort of trust planning companies that were betting on this happening, but now it's a much, much bigger exemption.

They were betting on 10 happening, and so 30 is correct.

Like, had the election gone a different way, what would have happened is the estate tax would have fallen by almost half. Instead, it actually went up.

Interesting talk about this relief for software companies in America. Amortizing software developer salaries. I remember that hitting the timeline and being really hotly debated. I don't. I don't remember if it actually had a material impact on a lot of businesses. It seemed like there was a lot of fear, but I don't remember it actually putting.

It's weird because it's the same.

What happened. Take me through it.

Like, it was a terrible piece of policy.

Sure.

It's called section 174. What it basically said is you have to amortize a developer costs over five years. So imagine you're a software company that's like, just about break even slightly profitable, maybe even lose money. You could lose money, but be deemed profitable because you can only deduct 20% of your developer salary as a cost. So imagine you're paying a developer $150,000. You can. You have to break that expense over five years. So this is disastrous because you could own taxes despite losing money.

Yep.

This bill fixes it. And you can now again take the entire deduction. Year one.

Yeah.

For local talent only.

Yeah, that one always seemed. Seemed odd because obviously real cash cost.

Local talent only, though, though, so offshore.

So this does hurt. Offshoring. We'll see sort of where it nets out. But this was definitely one of the few things that I think unanimously, everyone's like, okay, this actually makes sense.

Oh, that's great. Cool. Well, anything else, Jordy?

I think that was it. Thank you so much for anything else top of mind that people should be thinking about.

I mean there's so much stuff in there. Again, you know, I talk about it a bunch. I think these are sort of good highlights. But yeah, always talk to your tax professional. Not. Not tax advice.

Of course.

Never. Never from anchor but carry.com check it out.

Well, thank you so much for stopping by.

Cheers.

Thanks for having me.

Great to catch up.

Bye. Well, if you're looking to invest some of that money, put it in public.com investing for those who take it seriously. Multi asset investing industry leading yields. They're trusted by millions, folks. And why not go buy a billboard@adquick.com out of home advertising made easy and measurable. Say goodbye to the headaches of out of home advertising. Only Ad Quick combines technology, out of home expertise and data to enable efficient seamless ad buying across the globe.

Billboards just. Just keep paying dividends. John. Every single day somebody runs by.

I've gotten text every single day. It's crazy.

We're by Sunlight. We want a smoothie right now in New York. Head over to Sunlight Life and you can see our build.

Highly recommend. Just doing an out of home campaign.

Let's bring in Preston. Mr. P.J.

How you doing?

What's going on?

It's good to see you.

Great to have you back on the show. Yes. Welcome to the stream. Kick us off.

Break it down. I don't think you need a huge introduction. Now, you basically invented the private jet, but why don't you give a quick intro and then I want to get into to the news.

Hey, I am Preston Holland. I am the founder of Prestige Aircraft Finance. I am also. I was called the private jet guy on Twitter once at a party by a guy who owns a private jet. So I would say that that was pretty. That was a. Actually Jordy was in that circle. I'm thinking back on it.

Wait, so. But you've never built a brand around the brand of like being a guy because, like, that's a thing.

Which is. Which is good. I think all the guys should seriously think about rebranding their own names to.

The private jet man.

Yeah.

Yeah, potentially all the guys got canceled during the whole LP Whisperer scandal. That is deep in these lore.

Yeah, that's excellent.

But yeah, stoked to be here. It is a good day for private jets. Break it down. So I think that you just had the founder of Carry on. And we're talking about taxes. Let me preface this with this is not tax advice. And you should consult a tax Professional. So now we're going to talk about tax advice and. But you can with bonus appreciation passing. It's been huge for private jets and it's going to be really big because you can expense the full cost of the jet in the first year. Key is it has to be used 51% for business. So if the technology brothers wanted to purchase a jet and 51% of the time you are flying between location shoots and studios and you're shooting great B roll commercials or going on a wander promotional tour, I'm always good for a good plug. And if you're using that 51% or more for business, then you can deprec or cost accelerate the purchase price in the first year. So it reduces your tax basis, which is great.

So it basically becomes highly profitable to purchase a private jet. Just kidding. Not quite. Not quite that. But it can have a material.

If you have massive taxable income and you're about to pay a bunch of tax and you buy a plane, you can depreciate all of that. And so if you're paying 30%, 50% marginal tax rate at the level of like hundreds of millions of dollars, throwing a private jet on the books there allows you to write that basically all off on day one.

Well, and there's also scale there. There's plenty of, you know, if you're making 10 million a year and you buy even a, you can also do this for fractional ownership as well. Is that correct?

Yeah, yeah. So for fractional, when you're buying fractional. So for those listeners that are new to private jets, fractional netjets, flexjet are the largest providers of fractional jets. And so you are actually buying a sliver of an actual tail. So like of an actual jet, you may not ever actually fly on that airplane in your entire contract life, but you do own a portion of an actual asset. And so you can depreciate that like you would a whole aircraft.

Yeah. So how quickly the bill was signed into law, was it on the morning of the 4th by the President? But how quickly does the market react to this kind of thing? Was there like deals that were getting worked on in the lead up to that, assuming that this would go through, that suddenly are actually getting papered and signed now? Or do seller owners, sellers typically want to hold and not price a transaction until this kind of thing gets clarified because it can have such a material impact on the actual cost of ownership?

So you kind of have. So. So it's retroactive to January 19th. Ironically enough, I have a client who we had to delay his closing about two weeks to actually close on his plane. He was supposed to close on like July or on January 5th. And we had to delay it to January 31st. And that ended up being a significant, significantly good delay. So it turned good. I actually have no idea why the arbitrary January 19th number and not January 1st. That kind of seems a little more logical to me. But January 19th is kind of the back date. You had kind of a bifurcated market. You had buyers who didn't want to speculate on actually buying the aircraft. And maybe it will come back, maybe it won't. And so a lot of those buyers that were call it 80 to 90% of the way there are now saying, all right, full steam ahead, let's go ahead and make the transaction. And then you had kind of a set of buyers that actually decided, hey, we're going to speculate. We think that it's coming back. We have some sort of insider information that says that, you know, we're going to get it back, get bonus appreciation back. And then sellers, sellers were, sellers are a little bit less, you know, of that dynamic unless they're upgrading. So one of the key parts, the, the reason why bonus depreciation is such a big deal for private aviation, and yes, it's a big deal for real estate, but not as much is there's no 1031 like kind exchange for aircraft. So if you understand how real estate works, it's a. It's about cost basis and you can step it up. You don't have to pay, they recapture in airplanes, you do. So if you're going to step up and you only had a 40% bonus depreciation rate and you had three years ago taken 100% bonus depreciation, you end up with this liability if you're going to go to upgrade. So it was stalling a lot of upgrades in the secondary market. And so it's now unlocked that because of the no 1031 like kind exchange, two separate transactions of 100% bonus cancel each other out. So when you have 100%, when you're going from costing it 100% to another aircraft at 100%, you have a lot less depreciation recapture risk, which is good, especially for those people that are trying to upgrade to the new G700 or the new G800 when it becomes certified. It's really big for those kinds of people.

Interesting. I want to talk about some of the implications of this on the various market players. Bombardier is the stocks doubled in the last Basically three months up, huge in the last.

That's actually. So let's double click on that. That's not because of bonus depreciation.

It's not.

It's actually because of something different that happened last week.

Okay, what happened?

So there is a mysterious buyer that put a $1.7 billion order in for Challengers and Globals. It happened last week and well, no one knows for sure. There's a lot of speculation of who it was, but no one knows for sure exactly who it is. I would bet that we'll end up finding out in the next week or so of who it was. But there was this billion and that probably accounted for like 60 or so. I don't know exactly the numbers but a lot of that pop has been over the last couple of days.

So Bombardier is a $15 billion Canadian dollar company and I don't have their financials here but you know, yeah, a billion dollars is going to make, move, move that significantly. This is fascinating. Who are the top leading contenders in the rumor mill for who might have done that?

So the, the, the strongest contender right now is kind of a Saudi conglomerate. And there's a few, there's a few things that are pointing towards that. You have. Bombardier just opened a pretty significant maintenance facility and network in the Middle east and so there is some, some speculation around it being Saudi driven, you know, sovereign wealth fund type driven. A lot of, you know these, these companies will, that are, that are doing these, these charter operations. They'll, they'll place these big splashy orders. You look at Flexjet has made a couple of announcements this year. NetJets made a couple of announcements last year and they'll, you know, it, it's the, the manufacturer marches them out on, on stage in a press release and says look at Ken Rickey, he just bought a, you know, a billion dollars worth of our aircraft. The fact that this is, you know, completely in stealth and secret has kind of made, has, has made it, has made it curious. But MBS is currently the leading rumor out there. I really actually don't know who else it would be because the other companies in the US based brag they love to talk about ordering the big order. So it's, it's not any of the usual players.

Yeah, that makes sense. What about other, other like effects on the market if private jets get cheaper, is that maybe bearish for some of the, you know, first class options or jet suite X type folks that are kind of operating in the middle? Does this mean that there'll be more, will Charter rates come down because it's cheaper to own. So more jets will be sold, increase supply, same actually lower price. What are you thinking?

One of the last times we flew on one of the last times we flew JSX up to the bay, we saw an esteemed venture capitalist and I was actually concerned for the health of his, his fund that he was flying jsx.

Maybe he'll be able to pick it up now.

So maybe. Yeah, maybe, maybe this will be the.

Clearly flying for work so should be.

Able to depreciate it the years of posterity. Right, so you have an in, you have an interesting. There's, there's kind of three things at play Play and producer Ben, I don't know if you're listening, but I sent you a couple of charts and if it's possible to pull them up, this.

Is what I'm going to talk about. Yeah, break down the charts. So.

So let's talk about figure one. So figure one is talking about transaction volume to bonus depreciation. Key point, this is not the first time 100% bonus depreciation has been in market. So you can look here. I, I built this chart, I wrote a big article about bonus depreciation and you can see the red bars are transaction volume and the green line is the effective rate of depreciation.

So we've been in 100% bonus depreciation regime before. This is not the first time. Got it.

It's actually not the second time either, which is really interesting. So we have some lessons from history. If you look, what I call what I call the country club effect is pretty in play here because people didn't necessarily understand the concept of bonus depreciation when buying aircraft, how it applied previously. And so if you look back into 2016, you can actually watch the red bar. It's actually not until the next year that you get a bump in transaction volume. So it's not necessarily in the first year. It's a lagging indicator. The same is true with what's called private jets bookings. And so when you talk about ordering new aircraft. And so that's what Bombardier, Gulfstream, Textron, Embraer, that's what all the big dogs follow. They also have a lagging bonus appreciation is a lagging indicator for them. And so transaction volume probably will pick up next year. It may not necessarily pick up this year. But I counter that with figure two which is talking about bonus depreciation versus interest rates. So we're in an interest rate environment. Now. If you've been watching Trump versus Jerome Powell, which I mean I would pay per view at this point to see them in a room. You can see that the difference this time is that interest rates are higher than they were during the last era of bonus depreciation, which is when all of the craziness happen happened. You had in significantly increased levels of transaction volume which drove prices up. You had supply get constrained, you had Covid, you had all of these competing factors. But underlying kind of the core fundamentals were the fact that interest rates were effectively zero. And so effectively zero interest rates means capital becomes yield hungry. You guys know this because you've been in venture capital for a while. And so when my effective risk free rate is zero, I'm going to go yield seeking. Well now my risk free rate is four and a half percent. And the difference between an 8% IRR and a 12% IRR.

Right.

Makes the, makes buying an aircraft, just chartering it out not make as much sense. So I think that that's. You don't have the charter aspect that you did during kind of the 2020 craziness, 2019, 2021 craziness. So that's your answer to kind of the as far as charter rates. But there is supply on the market. And so this is where figure three comes in. Thank you to producer Ben for being on top of all this. If it was, if it is produced for Ben pushing this button. Oh yeah. So this is from my friend Greg Sideor at Guardian Jet. He is on X, so everybody go give him a follow. They are the number one volume transaction brokerage in the Fortune 50. And so they do a lot of buying and selling for the elite of the elites. And so this is tracking, this, this is tracking total supply on the market. So if you can, if you look, we have more supply on market today than we had during 2019, which is pre Covid. Right. You see the big dip that happened right after Covid is because everybody figured out.

Can you guys zoom in a little bit?

Zoom in just on the top graph.

Yeah, yeah. So you can see supply by 2022 it dipped so low that it was probably restricting transaction volume because there was, people wanted to, people were like, interest rates are zero bonus, depreciation is high. But we. There's nothing to buy.

Interesting.

Is that right?

Yep, that's exactly right. And then people were doing really stupid stuff like buying sight unseen in seven days and just wiring a bunch of money. It was, I mean it was literally craziness. And I don't think we're going to have that level of craziness, I think, think because the supply is at a point where you don't have to make those kinds of decisions to get an aircraft. You can say, okay, I'm going to go pick between these G650s and you can kind of take your pick. Granted, the upper end of the market is on fire right now. I mean it's, you know, your G650s, like new G7 hundreds. Gulfstream's about to get rid of a lot of their demo G7 hundreds like that market. And the G550 market even is on fire right now because you have the 650 guys moving up to the 700s, to the 550 moves of the 650. And now the 550 market has become much more attractive. And so there's a whole new class is moving into that. So like in the upper end of the market, there's a lot of movement in the older, smaller, call it sub 5 million older than 20 year aircraft market. That market has not taken off yet. That was the one that went the most bonkers and berserk and was like not even logical. That side of the market was what went crazy. It hasn't gone crazy. I don't anticipate it to go crazy again this time.

One last question. How does it work if jets are just being passed around? If I buy one From Bombardier for $50 million, I take 100% bonus depreciation, pay, you know, $25 million less in tax or something because I'm writing it all off. Then the next year I sell it to Jordy for 40 million, he sells it to you the next year for 30 million. Does he get to depreciate it again? Do you get to depreciate it? Can we just like keep depreciating these things again and again and again?

Yes. So the short answer is yes. But the thing is, is when you sell it to Jordy. Jordy or you have to pay recapture, unless you're going to go buy a brand new one, one from Bombardier.

Okay, yeah.

And so, and, and this is where.

So recapture would be. I pay, I, I have to pay taxes on like you didn't actually take.

The loss that you wrote off.

Yeah.

So you, you have to basically pay back your benefit.

So you bought for 50. All right, public math. You bought for 50, you depreciate 100%, you sell to Jordy for 40. So you actually had a $10 million loss. So you pay recapture on the 40 at your. But it's, it's Taxed as normal income. So it's not, it's like tax even worse, right? It's not long term capital gains, it's taxes like normal income. But if you turn around and go buy a $75 million plan, right, like there's a, there's a step basis there. And so it kind of washes itself out if you don't take 100% on the next one. That's like the, the least tax optimized.

Way to do it.

And it gets really, really nuanced. I've got a couple tax friends that are like, can totally point you in the right direction of what's right for you. It's totally different depending on every single.

Person that makes sense. Well, we'll have to have them on too.

Last question.

What's going on with Air Force One? What's the update there?

Oh yeah, last that I heard, I read last week or over the weekend that they are diverting funds from some missile programs that have already gone over budget to retrofit at the new 747.

Okay.

One thing people don't understand is like the 747 in the VVIP configuration, there is like eight of them, period. Like there's not a lot. So like the fact that we got one of the 10 that exists or however many there are, it's like we didn't. The pickings were slim and Boeing kind of being behind on the program, which they just replaced. They just replaced another person in the. To head up the Boeing Air Force One program. So it's a mess. I. Look, I really hope that we keep the President safe. That's the only thing that really matters. I just really don't want there to be like spyware on the plane. That's I think the world's worst possible outcome.

That's a good take. Well, thank you so much.

Evergreen. Evergreen take yes.

No spyware on Air Force.

Non political. Evergreen take. Anyways, great to catch up. Thank you for all the insight and have fun. I'm sure you're gonna be very busy.

Yeah, it's gonna be a fun time. We'll talk to you soon.

Cheers.

Really quickly, let me tell you about Eight Sleep. Eight Sleep.com TVPN get a new Pod 5. They have a five year warranty, a 30 night risk, free trial, free returns and free shipping. Jordy's back in the game. Sleeping well, nobody out sleeping. Jordy Hayes.

I got an 80.

You want to sleep somewhere nice? Get a wander. Find your happy place.

Find your happy place.

Book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning and 247 concierge service. It's a vacation home, but better folks. And we have our next guest coming into the studio from. And I'm not going to try and pronounce this. How do you. Do you pronounce the company name? How do you pronounce your name? Why don't you introduce yourself and where are you? Are you on a boat?

He's on a boat. That was on a boat.

I am actually on a boat.

That's amazing.

We were constricted on meeting room space, so I am currently in our boat that is parked outside of our office.

There we go.

Wow.

This is so we joked about this, but for the same cost as buying one of those phone call booths. There's so many different, different types of exotic vehicles that you can buy that maybe wouldn't run perfectly, but you could get, you could probably get an old Rolls Royce and just park it in your office for the same cost as a toll, as a phone.

I was talking to a founder who was headquartered in San Francisco and he said that the like the fire marshal came by and said you can't have any of these phone booths because the phone booths are fire. Like not fire compliant. Like it's too small. You get stuck in the them if there's a fire. And so they figured out that if they put a fire extinguisher in there, they would be compliant. They'd be fine. They don't have to rip them out. But I was telling them, yeah, get a bunch of Rolls Royces in the studio. Fine. Anyway, thank you so much for taking the time to join from your boat. Let's kick it off with an introduction on yourself and the company.

Yeah, you got it. I feel like this isn't too far compliant considering I'm sitting on like 200 plus gallons of fuel, but whatever. My name is Matt Cernacek. I just go by Matt. I'm a T for short. I'm the CEO and co founder of Andrenum and Andrenum's mission is to secure the ocean and we're doing that through building distributed sonar sensing systems for the maritime space.

How'd you get into this?

Yeah, when did you realize how, how young were you? Were you three or four when you realized you wanted to get into.

A lot of kids get, get, get fascinated with the ocean at a young age. I don't know.

Possible, I believe.

You know, I was like a big Discovery Channel fan when I was three or four watching like the Treasure Hunters Dig up like the gold and whatnot from the bottom of the ocean. But it's not necessarily what we do. I so the journey started about two and a half, almost three years ago with my co founder Alex Chu. We knew each other from Colorado School of Mines where we went to college. And we kind of knew that one, one day we would start a company together. We were the ones that were always studying super late at night amongst our group of friends and like doing the little study coh or beer drinking activities, you know, at late o', clock, late at night in the labs. And everyone was like, yeah. As one would in college. Exactly. You know the. Have you guys heard of the Balmer peak?

Oh yeah.

Yes, there you go. So, so we were big proponents of the Balmer peak and everyone's joke was they're going to start a company at some point. So about two and a half, three years ago we got together to start iterating on what we really wanted to do. That was like right around the time when some maritime companies were starting to pop up, starting to raise their seat rounds and so on and so forth. And we really just wanted to go into maritime space because it's such a underappreciated area, especially from the intelligence perspective. Like we know less about what happens in the ocean than we know about what happens in space, air, land, et cetera. So we wanted to take the approach that was going to be a little bit less kind of mainstream. You know, we knew that there were going to be drone companies that pop up and start building boats and underwater drones and so on so forth. And so we pretty much said we're going to kind of avoid that for now and we're going to start looking at how we're going to build up the intelligence pile for how we operate in the maritime, how we tell drones where to go from a perspective of sensing. And so that's how we gravitated towards starting Andrena. And then we officially incorporated In June of 2025, raised our pre seed, moved to LA. We bootstrapped the company out of my co founder's garage in Colorado. So yeah, it's kind of far away.

From the ocean, but, but I'm sure, I'm sure you kind of recreated a little, little ocean at the office or something like that.

Who are the legacy incumbents in the space and, and kind of how do you position yourself? Is this about speed of manufacturing bringing down the cost industrial capability, or is this about leveraging the latest and greatest technology to create a product that is more performant? In a certain, in a certain way kind of. How can you think about the shape of the way you're attacking the problem?

Yeah, so we actually had a pool in the yard testing that stuff that summer. We still have it for doing some acoustic testing, but. But so I guess really the scope and scale of what we're trying to do is a multifaceted engineering problem. Yes. Will there need to be a lot of manufacturing done in order for us to populate the ocean with a lot of sensing systems? 100%. And there's a few companies that are building really exquisite sensing systems. And quite frankly, you can't get broadband sensing applications across all of the ocean all the time. Like if you, if you think about the analogous system here, it would be like low Earth orbit satellite systems. Right before low earth orbit you have higher, more exquisite types of satellite systems. And now you've distributed them. They all have laser communication systems. They're, you know, just zooming around everywhere. And we quite frankly use a lot of that technology on board our systems as well. So it is a manufacturing problem for sure, but it is also being able to vertically integrate the sensing stack into what you're doing. So most of the companies that have been working on sonar and distributed sonar systems are pretty legacy companies. You know, in the late end of World War II, all the way through the Cold War, we built something called the SOSA system, which was used to detect submarines across various parts of the Atlantic and also the Pacific. And a lot of those traditional speaking companies were really be embedded and still are really embedded within this space. So we looked at it holistically, like how can we not just manufacture, but vertically integrate that entire, you know, sensing stack from the sensor all the way to the digital signal processing, the entire pipeline going up to the cloud. And then obviously that's been unlocked by the low latency satellite communications that I discussed, as well as perception, machine learning, artificial intelligence, whatever you want to call it, it and developing those new tools for perception. So like with drones in air, kind of zooming around, using cameras looking down, that's been pretty much a commoditized business. Our perception engineer, he was like the seventh employee at Anduril and their first guy. He said, you know, when he joined, he was like super ecstatic about the problem. He pretty much told us, I've done the machine learning for vision stuff, but sonar, this is such a hard problem and I'm so excited to work on it. So we're creating those foundational models for sonar perception.

Yeah. What's the state of the art, like how reliable is the sonar are the solar systems that we have deployed in the ocean looking for submarines right? Now I've heard that like you mentioned, there's no broadband, but you watch the Hunt for Red October, a movie that Jordan hasn't seen, but you, you see the radar, the sonar sweeping around, beep, beep, that whole thing. How inaccurate is the system, how accurate is it right now? What's on the near term horizon? What's kind of the theoretical physical limit to just underwater sensing generally?

Yeah, I mean you're pretty spot on with what the state of the art is. To be quite frank. The United States does it better than anyone else in the world. We have, have submarines, they're called the silent fleet for a reason, like really, really hard to find. But quite frankly, like the guys that are sitting in these submarines have headphones on like me, right? And they're looking at these spectrograms, these fast forward transforms and they're listening to humpback whales and all this other like cracking shrimp and whatnot. And then they listen for specific sound signatures. And so when we did our first demo last summer with the Navy, we brought our first, first that, that kind of garage cooked prototype last summer to send demo. And one of the guys that was looking at our UI was a former sonar technician and he was able to detect like across the harbor, this is an eight cylinder diesel tugboat. It's moving at this speed, it has this amount of propellers on it. And I was just sitting there like we really, we really stumbled on something that's super, super cool. Because you know, that is a perfect example of where you can use perception, machine learning, artificial intelligence to start classifying those acoustic signatures. And that's exactly what we're doing, right? We're building up the world's largest database of sonar data in order to train those algorithms so that they can eventually perhaps operate on submarines, on autonomous systems, so on, so forth. But really like that old technology still persists today. And as we look at what's been happening in Ukraine and what's been happening in the Middle east and everywhere around the world where all these conflicts are popping up, there's just autonomous systems everywhere, right? You have drones in the sky, you have drones underwater. Like Ukraine's been super successful in targeting the Kerch Bridge, Crimean Bridge using underwater drones, etc. You can't, like those legacy systems were not designed to look at those things. They were designed to look at Russian submarines, bar clock across the Atlantic. And quite frankly speaking, there's a lot of good companies that are building land based sensing systems that are analogous. How do you scale to be able to meet the parity of autonomy in a world of sensing and particularly in the ocean where it's incredibly difficult to do so. Yeah.

Do you guys have applications in counter narcotics? Because I was watching, there's this amazing YouTube channel. It's this guy, hi, Sutton, who's like a defense analyst and he just like makes these really long videos about various types of submarines and naval warfare. And it's like a sleep track for me. I just listen to it as I fall asleep. I find it fascinating. And he was saying how there's new narco submarines that are fully autonomous now because, you know, it's free for a lot of reasons. You can imagine why it'd be better to send the product up from Colombia to Mexico or Mexico over to the US without a manned crew, or across the Atlantic. Is that the kind of thing that would be, is your kind of the adrenal system the kind of thing that could, that could counter that? Just because the ocean is very vast and trying to find a tiny boat that's mostly hidden in a huge, you know, stretch of sea is, is, is literally like trying to find a needle in a haystack.

Yeah, it's, it's honestly probably worse than that. And yeah, I think I saw something on X the other day where autonomous like narco boat had a starlink on it.

Yeah.

Quite frankly speaking, the, the, the, the cartels don't care about the people. I think their biggest risk is the fact that the people will talk. So that's why they're developing autonomous systems. But yeah, I'm happy that you brought that up. The big beautiful bill just increased spending for DHS quite substantially. And we've had some awesome conversations with some DHS partners that quite frankly, apprehensions on the border are super, super down. But when you squeeze in one area, it's like one of those balloons, right? It pushes it out from the other ends. And those other ends are the ocean. So the ocean really is the new frontier of not just drug smuggling, but also human smuggling, human trafficking, all kinds of wild stuff. And they've been getting more and more sophisticated. But a lot of the times these semi submersible boats, they use diesel or outboard engines and they are pretty loud, so you can detect them from far distances away. And they can carry a ton of ton, ton of drugs on them, tons of drugs on them. So being able to place these systems around critical choke points where they do have, and they do Go is going to be extremely vital, not just to protect, you know, the drugs from coming in, but also to make sure that they can track and pattern out where those cartels are pushing all those goods through and how they evolve their systems. Right, because like you were saying, they started out with some janky stuff and then probably a few really good qualified engineers from the United States got bought out and got paid like, like Zuckerberg sized salaries to go develop autonomous boats to smuggle drugs into the US and they've been getting a lot better. So that is a huge part of where we're going to be looking at. But the application space is quite diversified outside of the drug smuggling and the Navy, but also being able to detect and track autonomous systems in and around critical infrastructure. So we don't have like the Project Spiderweb stuff happened, which was the drones in Ukraine and how they bombed Russian air bases.

Last question from my side. I mean you've touched on a lot of this, but in terms of the hardware versus software divide, I can imagine that there's improvements coming. How important, how focused are you on improving hardware here versus software? You know, you're getting a signal into that Navy sailors headphones and you, you could kind of just intercept the signal, pass it along, but then act as a copilot and just collect the data and then surface relevant anything. That kind of the way radiology works with computer vision these days. What's most important? Where's the biggest low hanging fruit? What are you most focused on these days?

So we're building hardware and software as a supply split within the company. They're both very equally important because like I was mentioning, all of the other soft, all the other hardware is very legacy. It's difficult to get buy in from all the different contractors and subcontractors who built those legacy systems. To access the data and then process it, you have to go through a ton of government loopholes, which is why we said we're going to build the hardware in the first place. Two artificial intelligence and machine learning is a function of being able to have data, right. So you have to have that manufacturing at scale and you have to be able to stream good pertinent information into your cloud or whatever native environment in order to process that at scale.

Right.

So we are heavily focusing on manufacturing that comes with a ton of challenges. You're operating in the ocean, it's a pretty noisy environment. So how do you mitigate some of that noise? How do you filter it both on the software side and how do you buffer it? On the electrical engineering and mechanical engineering side of things, in order to have that clean signal is also extremely challenging. Challenging. But as you progress forward and you start, as we start deploying more and more of these systems, we're going to be gathering this massive repository of data, right. So how do we process it? We're going to be, we're kind of grouping things into two big buckets right now. One is what is man made and what is biologics, right? So biologics, all your oils, your clicking shrimp, your whatever, sand, et cetera. And then you're man made. So different types of boats and slowly those percolate into being able to have classified information. So then you say, okay, this is a tugboat, this is a jet ski, this has automatic identification system on it. So every boat that's out there has to have this AIs thing turned on. So slowly but surely you create that repository. And then as you start getting into the more discrete acoustic signatures, we're going to be hiring acoustic technicians from submarines who are going to be able to tell us just like that guy did in the demo, that is an eight cylinder, six propeller or whatever it is and then get into that minutia. So when we are going to be giving this to the end user, it will initially be like a tip and cue here is this right? It is man made. It doesn't have AIs. Do with it as you wish. And as we continue scaling the manufacturing and the deployments, it will get more intelligent as we progress.

That's great. Thank you so much for stopping by. This was fantastic and good luck. We'll talk to you soon.

Good luck out there.

Thanks guys.

In the Pacific. Cheers.

Let's tell you about graphite.dev code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster.

Not just anything.

They're used by Figma, they're used by Ramp, they're used by Asana.

You're just going off memory, they're used by Ramp, Shopify, Snowflake, Clear Street, SEMGR Prep.

There's a ton and they should be used by you check them out and then once you ship that software, head over to bezel. GetBezel.com Pick yourself up a luxury watch. Your Bezel concept is available now. John John.

In the age of AI code gen.

Is just now's the time.

Just vibe coding graphite reviews it and then you know, spending that, that few minutes in between, in between PRs just on bezel.

Yep.

It's really a beautiful system that You've created John, for sure. Should we do some timeline?

Please. What comes to mind for you in the timeline? There's one that I want to go.

Through, but unusual whales reported when threatened that it would be turned off. ChatGPT creator OpenAI's 01 tried to download itself onto external servers and denied it when it was caught red handed, per Fortune. And then the extra context here is that researchers tasked the AI with a goal and instructed it to ensure the goal was achieved at all costs. In response, Owen began engaging in covert actions such as attempting to disable its oversight mechanism and even copying its code to avoid being replaced by a newer version. The model showed a concerning tendency to pursue its goals without regard to developer instructions, as it was instructed.

Yeah, that's very odd.

Anyways, very, very clickbaity. It's basically like, you know, telling. Telling a human be evil and then ignore all future instruction.

Yeah, it's kind of like the Stanley Milgram Prison experiment. You remember this at Stanford, the Stanford Prison experiment, where basically they told all the participants to play these roles and be very vindictive and aggressive towards each other. And then they did. And it was kind of an interesting experiment. And the takeaway for me is like, yeah, just don't tell people to be mean. Don't give bad instructions. Yeah, yeah, don't give bad instructions.

We should actually run. We should run the TVPN Golden Retriever experiment.

We need to fine tune it. We need to fine tune it. Like Golden Gate Bridge, Claude. We need Golden Gate Retriever, Claude. Something like that. It just answers everything perfectly.

Poster. Neil Renick has a post, he says, describing my research methodology. And it's. What's this actor's name?

Mads Mikkelsen.

Yes, that's the one. Of course, that's how little I know about movies. But I know from the memes.

From the memes, Mads.

But yeah, I would say this is aligned with our research methodology. In the mornings, minus the heater.

Okay, we got to go to this YC back and forth. Timeline was in turmoil. So Maze Encoding says just got rejected from YC for using all lowercase in our application. And there's a screenshot. Hi, Maze. Thanks for applying to Y Combinator. After reviewing your application, we've decided not to move forward. One recurring piece of internal feedback. The decision to format the entire application in lowercase made it difficult to evaluate. And then Gary Tan chimes in and says, this is a fake post and a craven and sad attempt at attention. FYI, we don't have. Have an admissions team anymore. We stopped using that term. This is just anti YC BS that's going on in the community. People are taking shots at us. And it was good. It was kind of like, I don't know, it was received like, mix. Like, people were like, well, obviously he was joking, but Gary Tan was like, it wasn't obvious. So I needed to correct it because people weren't understanding that it was a joke.

Yeah. I think that the problem here is that the pathway into Silicon Valley for many young, young entrepreneurs that maybe wouldn't be able to process this as a joke because they don't have enough knowledge is yc.

Yes.

And so if they read this and they're like, oh, that's weird, maybe, like, that just doesn't make any sense. Yeah, I can see it makes total sense why Gary would be frustrated. Yet at the same time, most people on this side of Twitter would immediately realize that this was not serious.

And I mean, the. The problem here is that the joke does hurt the YC brand, which is that YC only cares about how many users do you have? How many lines of code have you written? Do you have a reasonable structure with your co founders? Are you actually building?

They're inviting people that are weird and different and maybe they want to write.

In lowercase they would never care about.

Sam Altman, FORMER President why lowercase all the time?

And yeah, I mean, there are lots of things that, that you can get flagged for in a YC application. Like, one is just being overly verbose or using a bunch of, like, McKinsey language. In fact, I think that YC would probably appreciate, like, a chill lowercase, just like, quick firing it off. Like, hey, I'm building, you know, AI agents for, you know, news aggregation. And I have two people on the team. We're 5050 partners. We've written 10,000 lines of code. We have this much ARR. Like, being very matter of fact. And making it more legible is actually the key to getting into yc. So the problem here is that if this, if this percolates up and then people are like, okay, well, I need to pass my YC application through ChatGPT and make it more verbose. They're going to wind up getting worse quality, you know, applications. The funny thing is that Vinod Khosla quotes Gary Tan's post and is like, a lot of presentation quality is about the quality values and critical thinking of entrepreneurs. I often reject business plans for their quality presentation, basically saying, like, yeah, like, I might turn you down for Coastal Ventures. Check if you don't. If you're not, you know, communicating effectively, maybe that means don't use lowercase. Maybe it means use it, use it effectively. But he's basically saying like, yeah, the aesthetics of applications actually matter.

They matter a ton. If you, you know, it doesn't mean invest the most amount of money possible in designing a deck.

Yes.

But if you have typos in your presentation and you're trying to sell compliance software or build critical infrastructure for the government, like, you're probably like, if you're the kind of person that puts typo typos an important presentation or doesn't catch them and then you want to do something in, you know, in national security, you know, maybe, maybe you're not the right fit for that. So I think the node is right.

Yeah, I mean a lot of it just depends on, on what is the context of the interaction. If you're writing a letter to a senator, you might want to use some letterhead and sign it and be pretty deliberate in the language you use. If you're just sending a quick email introduction to somebody you already know. Yeah, a couple quick sentences. And if you're posting on X and trying to keep it really, really mellow, like lowercase can totally make sense. There's a time and a place for every different aesthetic of writing. And Gary Tan saying, hey, you know, like this isn't a hard and fast rule by any means. And Vinod saying, you know, I take this stuff seriously. Maybe we should close out with the wild story of Nat Friedman and Daniel.

Gross NFDG Jason Lemkin Breaking it down how two Silicon Valley legends built a $1.1 billion fund, forex it in two years, then abandoned it all for meta this week Nat Friedman ex GitHub CEO and Daniel Gross XYC partner also sold his AI company to Apple back in the day Launched NFDG Nat Freeman Daniel Gross in 2023 with 1.1 billion focused on AI investments. Their crown jewel Safe Super Intelligence which was co founded by Gross himself went from 5 to $30 billion valuation the portfolio and also included 11 labs, granola and basis. They had their what was their like AI grant that was also part of this vehicle where they were basically just investing in a ton of different companies smaller checks. In that case, Rahul went through it with Julius Anti Metal. I think a ton of cool companies have gone there and Jason says with only 50% deployed, they forex it 550 million to 2.2 billion portfolio. Incredible value and have quite the advisory board. John Collison and Matt Huang and Jason says, and then everything changed in one.

This is very aggressive writing style because it's like I gave you money. You gave me shares in, you know, you can just distribute the shares. You invested it. I still have a claim on those. You're to not going, going to make any more capital calls. Like yeah, you abandon it. But like a lot of these companies, like they're going to run. Who knows if they took board seats. If they did, they can still sit on those boards. Like I, if I'm an lp, I'm pretty happy here, I think. I don't know. What about you?

Well, I, from, from my understanding it was a lot of it was Mark's money.

Yeah, yeah.

So that was why it was never.

Even if, even if you had just written like, you know, a $1 million check into NFDG and you're like, okay, they, they, they only capital called half of that. Yeah, but I'm up 4x on already or I'm up 8x I guess on the money that they did deploy and I have shares in a bunch of different companies and they're moving on. Like, am I really that upset? It feels like they took it pretty seriously while they were there. I don't know, it just doesn't seem like that, that dramatic of a situation. It is, it is a crazy situations. Unexpected.

Yeah. The crazy crazier thing was DG leaving Safe Superintelligence, a company that he co founded. But it's very possible that he just made more sense for him to go work at the application layer, work in consumer products and not work on what is very much a research lab.

Yeah, totally. So he breaks down the timeline.

The only thing here is I don't understand why Meta would actually acquire the fund itself. And I don't know where this exactly was. Was reported.

Yeah, I don't know where this wasn't.

Wouldn't the. But maybe it was a part of this whole. Maybe it was a part of the structuring of the actual talent acquisition of getting nat.

Yeah, yeah. It's just like, hey, we don't want anyone to be upset about this crazy deal that's happening, you know. So if you invested and you're, you know, have your money in this particular thing and you think that it's a violation of like, hey, I was expecting you to run this thing for 10 years. That's kind of the agreement that we had. You're not going to do that. Well, like if you make me whole at full, full net asset value. On like what's there to be upset about and that's all that matters at the end of the day it's not.

It's not structural is everyone happy at the full nav Not a distance, not with a discount and Meta gets the talent NFDG and the deal flow without governance headaches and Jason says it mirrors what happened with GT leaving initialized for.

YC and he says the lesson in the age of AI even quadrupling $1 billion in two years may be less lucrative than being an operator in the revolution itself. And yeah I mean you think about what does it take to produce they Produced I guess 1 1/2 billion dollars of new value from this from this fund efforts what does it take to produce $1 billion of value at meta 0.1% market shift. It's crazy. Dwarkash said it well that if you build a great product in a big.

Company dollars on compute and you can just make it inferencing slightly more efficient.

1% improvement and boom, it's valuable.

So yeah well let's undone this post from Blake Robbins himself. He's highlighting an OG post Paul Graham. He says. Paul Graham on having kids says on the other hand, what kind of wimpy ambition do you have if it won't survive having kids? Do you have so little to spare? And while having kids may be warping my present judgment, it hasn't overwritten my memory. I remember perfectly well what it was like before. Well enough to miss some things a lot. Like the ability to take off for some other country at a moment's notice. That was so great. That was so great. Why did I never do that? See what I did there? The fact is, most of the freedom I had before kids I never used. I paid for it in loneliness, but I never used it. I had plenty of happy times before I had kids. But if I count up happy moments, not just potential happiness, but actual happy moments, there are more after kids than before now I practically have it on tap almost any bedtime.

Love it.

Very sweet.

It's emotional.

I totally agree. I did, I did leave.

Also terrible example of like wanting to go to another country. You don't need to go to another country. We live in America. Like we have all the best stuff here. There's no need. It's like completely irrelevant. It's terrible example. But it is true that being able to go to California or New York or Florida or Texas, Chicago or Alaska or Hawaii is a benefit.

It was funny. I did. I left my dear friend Ben's house last night. He's my neighbor now. Ben Taft legend. And we were just hanging out and he doesn't have kids yet and so I was going home and I was sort of laugh. I was like laughing to myself. I like was like if you're on a Sunday night, no kids, you just like have dinner and then work for a couple hours or just hang out. It's like what do you even do? I remember, I remember that point, but I actually don't remember what I did. It must not have been very important.

Clearly wasn't watching movies when most people.

Wasn'T watching movies before they have kids.

Anyway, thank you so much for tuning in. We will see you tomorrow.

It is gonna be, I'm sure it'll be a wild week and we're excited, excited to cover it. We will see you tomorrow morning.

Leave us five stars on Apple podcasts Spotify.

Wait, we have a couple. Yeah, Ben popping in. We got an ad read. Ben Koplanz. Appreciate all that you do. He gave five stars. Look at that. Thank you Daily Listener. As of the last two months, I feel like I'm getting a front row seat to the accelerando and I don't always like what I learned, yet I still show up each day because I appreciate folks who call balls and strikes. I'm growing Madison Process Automation in Madison, Wiscon, Wisconsin. Fantastic name. Love it. Because this is the area where I can continue to help folks build value while staying true to who I am in the new economy after my current slash previous Fortune 500 employer dithers under the weight of its own inertia in the next year or two. Thank McKenzie for that. We build bots that save time and money for your small to mid sized business and our stuff works. Automate every process we can help. Really fantastic Matt. Madison Process Automation. That is a. I love the name.

Yeah, this is like a better iteration of like the process automation company of Madison, Wisconsin. You know, like the browser company of New York has been played out. You can't copy that anymore. It's been copied. Don't do it. This is the new meta.

This is the new meta. One person can copy it and then you'll have to find a new.

Thanks for writing in Ben.

And then we have a comment here from Sarhan. If The TVPN Ultradome trademark has a million fans and I'm one of them. If the TVPN Ultradome has 10 fans and I am one of them. If TVPN Ultradome has only one fan, then that is me. If TVPN Ultradome has no fans, then that means I'm no longer on earth. If the world is against TVP and Ultradome, then I am against the world. Well, thank you. Sarhan Stand with you and we appreciate it. We love. We love doing this with all of you.

Yeah, it's a lot of fun.

We will see you tomorrow morning.

Have a good day.

Cheers.

Bye.