We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Dario Amodei’s letter turns AI-risk debate into an operating agenda
•
3 min read
•
223 docs
A focused set of authentic recommendations led by Dario Amodei’s AI-safety letter, with complementary lenses on agentic work, AI-risk taxonomy, moral judgment, and skepticism about doomer narratives.
Start with the most actionable item
Dario Amodei’s AI-safety letter
Type / creator: Letter by Dario Amodei. Recommended by: Tomasz Tunguz. Access:discussion in All Eyes on Warsh.
Tunguz calls the letter “really wonderful” because it moves the conversation beyond a simple pro-AI/anti-AI split and identifies work at three levels: the individual company, across companies, and internationally.
Why it matters: This is the clearest practical recommendation in the set: a way to turn an abstract AI-safety argument into a map of who needs to act and where.
Two useful lenses on AI
Ben Hansford’s earlier conversation on working with agents
Type / creator: Video/podcast conversation with Ben Hansford, a USC film professor. Recommended by: Reid Hoffman. Access:Hoffman’s recap; the direct episode link is not included there.
Hoffman says he keeps rewatching the conversation because each viewing teaches him something new. He highlights Hansford’s shift from treating AI as a tool to treating it as a personal team, summed up by the line: “a hammer can’t build a bird box while you sleep.”
Why it matters: It is a concrete model for delegation rather than question-answering: give agents a mission and let them return with completed work.
Yishan’s existential-risk versus misuse post
Type / creator: X essay/post by @yishan. Recommended/shared by: Naval. Link:Yishan’s post. Naval’s accompanying frame is concise: if AI is risky like fire, everyone should have it; if it is risky like a nuclear weapon, no one should.
Yishan’s longer post separates existential risk—the possibility that superintelligence wipes out humanity—from risks caused by powerful AI being misused, then argues that the solutions to those two problems may be nearly opposite.
Why it matters: This is a useful vocabulary for avoiding a common policy failure: two people can say “AI safety” while reasoning about different risks and therefore advocating opposite remedies.
A human precedent for judging behavior under pressure
A French village series about World War II occupation
Type / creator: Television series; the exact title and creator are not named in the exchange. Recommended by: Michael Moritz. Access:Moritz’s interview.
Moritz describes it as a “spectacular” French series about life under occupation: some people resisted, most stayed silent, and some collaborated. During COVID-era Zoom calls, he used those categories as a private test of which acquaintances would have had the backbone to resist.
Why it matters: The recommendation is valuable less as historical background than as a behavioral lens—watching how people respond when conformity, fear, and moral choice collide.
A skeptical counterpoint
@Jason’s AI-“bioweapons” video clip
Type / creator: Video clip featuring @Jason. Recommended by: Bill Gurley. Link:clip post. The post presents the argument that AI-bioweapon fears require an “incredible thriller” and are amplified by science-fiction expectations; Gurley called the clip “spot on” and said people should stop listening to such fantasies.
Use this as a rhetorical counterweight to the risk-focused recommendations above, not as a technical assessment: its value is showing the skeptical, anti-doomer case in its own terms.
Dario Amodei’s letter turns AI-risk debate into an operating agenda
Keith Rabois
- Article — New York Post report on Anthropic’s AI watchdog and its ties to effective altruism: Keith Rabois replied “True” to Hugh Hewitt’s characterization of the article as the story that “killed Anthropic’s credibility in ten minutes of reading,” providing an organic endorsement of it as a credibility-critical read.
Tomasz Tunguz
Profile
- Dario Amodei’s AI-safety letter (letter): Tomasz Tunguz called it “really wonderful” for moving the discussion beyond a simple pro-AI/anti-AI split and clarifying actions needed at the individual-company, cross-company, and international levels.
Naval
Naval links to Yishan’s X post while using a fire-versus-nuclear-weapon distinction to frame whether AI should be widely available or restricted. Yishan’s linked post distinguishes existential risk from superintelligence and risks from human misuse, arguing that their solutions may be nearly opposite. Resource: Yishan’s X post
Reid Hoffman
Profile
- Ben Hansford’s earlier Possible conversation (video/podcast episode; exact title and direct link are not provided): Hoffman says he keeps rewatching the conversation because each viewing teaches him something new. He highlights Hansford’s perspective as a USC film professor using AI agents as a personal team rather than merely a tool, including the line, “a hammer can’t build a bird box while you sleep.”
Invest Like The Best
- French World War II village series — video/TV series (title and creator not identified): Mike, a Sequoia investor, praised an apparently six-part series about life in a French village under Nazi occupation as “spectacular.” He used its distinction between people who resisted, stayed silent, or collaborated during COVID-era Zoom calls to assess which acquaintances had the backbone to resist.
Chamath Palihapitiya
- Video — All-In Podcast (data-center clip): Chamath Palihapitiya shared a post framed as “Debunking the data center hoax,” linking to Rohan Paul’s X post, which contains a clip from the All-In Podcast YouTube channel. The clip’s concrete takeaway is Satya Nadella’s Quincy, Washington example: a 400/500 MW data center contributed to a 12× increase in tax revenues, continued economic growth, and new school, hospital, town-center, and aquatic-center infrastructure. Linked X post.
Bill Gurley
Video clip — Jason Calacanis on AI doomerism and “bioweapons”: Bill Gurley called the linked clip “spot on,” endorsing its argument that AI-bioweapon fears are shaped by science-fiction-style fantasies; the takeaway is to stop listening to those fantasies and get back to work. X post
All Eyes on Warsh, Space Weapons, DreamMogged, Tomasz Tunguz Joins
TBPN
YouTube channel
Tomasz Tunguz
Guest
1 of 2
highlights
0:00You're watching watching TVN Wednesday September 16th 2026 here live from the CPR the temple of technology the fortress of finance the capital of capital let me tell you about ramp.com time is money save both needs to use corporate cards bill pay accounting and a whole lot more all in one place is that even going to happen I don't think so I don't think that's
0:28strip to strip the gears off that one. Uh, that's rough. Uh, well, we have a fantastic show for you today, folks. I made it to San Francisco and back since the last show. This is a new thing. I like this being able to get out from the show, go do something in San Francisco, get back. Uh, very excited.
0:45Why didn't you want to stay?
0:47I don't know. I like doing the show.
0:49It's pretty simple. Uh, good to be here in the TVP Ultradom. We have a great show. We have a bunch of great folks coming on the show. uh what was on my mind last night I was listening to China talk Jordan Schneider was talking about evaluators third-party evaluators and I was noticing this discourse around like it feels like we we we funneled into like very clear camps super super quickly and it feels like I don't know too calcified for how fast it happened
1:17like the the Dario essay comes out and uh you know he throws out meter says he's bringing in meter and And the backlash is like immediate at the New York Post is crashing out saying like these are handpicked AI watchd dogs. Uh Martin Casado, Martine Casado over at A16Z. He's pushing for the Department of Energy. He's like full nationalization now. And uh you know both of those have their advantages, disadvantages. They both do good work. Um they are the
1:47extremes. Um, and so I was just I was just sort of wondering a few things like first is like is like what other regulatory bodies can actually work? How do they work in other industries? It's very interesting because um like the revolving door is something that's common like uh in in in financial regulation you get people that work at banks and then they go and work at the regulator and then they go back and forth. I mean this is the uh this is the story of uh of the like many people in
2:16the crypto industry where the regulators who are regulating it they go back and forth like revolving doors exist but I think uh the push back to meter is very much like the door is like too revolving or it's too it's too close I guess um but the but but the more interesting question to me is is uh yeah the defensive meter is that I don't know many groups that are qualified at all to even to even understand what's going on at the frontier. Right. And so
2:46is that true though? I see that's the thing I disagree with. I I I would say that there's there's there are not that many groups that have been this invested in in understanding frontier model behavior for this long. It's just a small group right now.
3:03That's a separate thing. So I think there's two separate things. There's one which is like super forecasting, seeing the future, predicting what's going to happen. I think that's important. I think taking that seriously is important. But then there's the other side which is like doing the work, reading the logs and being like this violated this rule, this hack happened, here's how it happened. And I think that those are actually two separate disciplines, two separate jobs. And you can just tell the regulator if it's someone. You don't need to tell like if
3:31you hire someone and they are able to to un like like look get up to speed on how these systems work and and evaluate them and look at logs of different incidents, see what's happening, assess the risk level, assess the liabilities. Uh if you can get those people up to speed and you can just be like it is your mandate to take this seriously. And and the example that I'm pulling from is like is like you can be 22 years old, not graduate from college, go into the Navy, enlist in the Navy, not even in the officer
4:00training program, and in 18 months you can be responsible for the security of a nuclear power plant on a submarine. You do that you do six months I actually looked it up it like you you do uh you do six months in nuclear field a school about six months of nuclear power school and then six months of handon hands-on prototype training before arriving in the fleet. Nuclear power school covers math nuclear physics reactor principles health physics materials thermodynamics electrical systems and reactor technology. There's no part of that
4:29that's like you need to really at a deep level understand that like nuclear annihilation is bad and could happen like the you don't need to be able to forecast out like well what happens if China and geopolitics and Iran gets the bomb and then these people and then Pakistan like you don't have to understand that to just know like don't let it blow up. That's your job. Here's how you don't let that happen. We've created a plan for you. You're 22 years old, but we know that you can do this job and you are enlisted to do this job
4:59and doesn't matter if you think nuclear war is impossible or doesn't matter if you think it's going to happen tomorrow.
5:05Your pdoom on nuclear is completely irrelevant to you doing this job, right?
5:09Yeah.
5:09And I think that that's something that is maybe being missed here a little bit.
5:13There's like there's like well you have to take it seriously. You have to have seen it coming. And I actually think that like super forecasting super important awesome also just like fun read and if you take it seriously you make a lot of money you work on interesting stuff you get amazing technology there's so many things that come downstream of that that are really really positive and really important but I don't necessarily know that it's actually a prerequisite and then the other thing is yeah Tyler I mean like meter they're not the ones putting out those forecasts though
5:41that's like those are other groups if you look at like meter research it's like you know very complicated benchmarks and like these kind of totally totally Yeah. I'm I'm I'm sort of collapsing the meter criticism from like the New York Post's perspective.
5:53And and I think that is overall I think generally people are going to like if if if there's a third party regulator it seems like it has to be an entirely net new group because I don't think anyone has like whe whether or not meter meter could be operating in and their actions could reflect that of a fully independent group but the financing
6:21structure the history of the of of the various parties there just makes it so that people don't have any trust that they would act like that.
6:29Sure. The the interesting thing is that uh there is actually a distinction in the history of nuclear regulation which is a there's a difference between advisors in the before the NRC it was the uh atomic energy commission and the AEC was created to oversee nuclear development but and many of the scientists who did forecast the importance of riskiness of the technology were involved. interesting like like Einstein writes this letter
6:59and says like nuclear war he basically you know like uh describes like what nuclear annihilation could look like. There's a whole bunch of other scientists that actually run the numbers and they're like this is what could happen if all the nuclear bombs go off.
7:11They do all the calculations. There's some that go a little bit too far but uh in general like the scientists were the ones who got to it earlier but the scientists didn't actually wind up being the regulators. They wound up being the adviserss. So they so uh uh Oppenheimer became the chair of the AEC's general advisory committee which had enor enormous influence but didn't actually issue licenses. So the actual work that was being done, it was like, "Okay, he
7:39Oenheimer in this advisory role says like,"Well, we need to have a security guard here with a gun that makes sure that no one can steal the nuclear material, right? But he's not the one doing it. He's not the one actually hiring that person. That's just like an engineer who's qualified for that job."
7:54And I think that actually takes a lot off of it because you could say, "Oh, there's all these like conflicts of interest here or there or there, whatever." But if you just say, well, they're just putting putting out a proposal that then people are going to go and implement. But then the actual people that are doing the implementation are are much less conflicted because they're just drawn from the broad pool of engineers and scientists and mathematicians and physicists and whoever else we have in on in America who can do this type of work. Um, it gets a lot less complicated in in my mind. There there's a whole bunch of
8:24other interesting uh details from the AEC history. Enrio Fairmy and Glenn Seabborg became the became committee members with Seabborg later becoming the chairman of the AEC. uh AEC itself of course Fermy from the Fermy paradox uh famous scientist very interested in like forecasting seeing the future understand the implications of things but again uh becomes uh committee members uh not actual regulators not actual lensors and
8:52then Edward Teller is sort of like the most accelerationist because he was advocating for the development of the hydrogen bomb but he ultimately became the chairman of AEC's reactor safety committee talking about uh how to actually secure reactors in in particular, but again not actually on the team regulating uh staffing working on this directly. And so today the regulators like nuclear regulators are talented and hardworking but these are
9:20not like the most elite jobs like you can you can just be a nuclear engineer, mechanical engineer, material scientist, physicist, health physicist, uh geologists, probabilistic risk analysts, cyber security professionals they hire for this uh emergency preparedness, they hire lawyers, they hire inspectors. uh pay for some of these jobs ranges between 125,000 and 187,000 for many NRC technical staff roles. So the like like once the machiner the machinery did have
9:49to get sort of described by the scientists but then the implementation of that machine of that regulatory structure is actually done by really hardworking really talented Americans but not head in the clouds not thinking about the future in some you know bizarre way that's enough to happen at the democratic level and then it gets implemented and I think that that might might eliminate a little bit of this like oh okay well like this person who's really tied to you is now like inside
10:17actually the one with the keys, the one with the with with with the role overseeing you. I don't know. What do you think about this, Tyler? You have some push back.
10:24Yeah. I mean, like I I know that um Casey, the Center for AI Standards and Innovation, like I think that they've had a hard time like funded. Yeah.
10:32Yeah. So, it's like like they're having a hard time like finding people who who will come like from the labs. like maybe they should just broaden who they're looking for, but it seems like it's still like um the the the like the ways that we detect if a model is safe or not are not like set in stone yet. Like it's still like that's what the role of meter that that's kind of what they're doing right now. So I think it is still different than like you know there are predetermined accounting practices and you can just kind of check the box and like follow the rules like there's still
11:00like Yeah. It's like a moving field you know.
11:03Yeah. Yeah. And I mean there is there is the question of like with particularly with like agent swarms perpetrating cyber security violations that don't directly cause economic harm.
11:17Like I didn't tell it to hack you. It hacked you but it didn't knock your payment system offline. So you didn't lose a dollar of revenue. It's like very hard for you to prove that I acted wrong and then also economic damages. So there's a whole new level of like, you know, tort battles that need to be battled out in the court of law to see like what exactly do I owe you? Because I shouldn't have done that, but what do I owe you? What's the damage? What's the what's the problem there? And then uh
11:46and then you can go and say, okay, well, you know, how how do we measure that?
11:50How do we prevent that? And uh how do we work through that? But um but the the discourse is getting like uh more and more and more polarizing by the day. Uh we'll see. I think there will I think there will be the the the the big sitdown between the lab leaders. I wonder how important it will be to have Jensen with a beer alongside Sam, Daario, Elon with beers because the clear proposal there's going to be beers involved. That's what we know from the
12:18interviews. Everyone's asking why can't they just sit down and get beer?
12:21Six months of those could have been on the Cheeky Plane podcast.
12:24It really should. That that that is sort of neutral ground, too, cuz Elon's an investor in Stripe. What happened there?
12:31Uh Elon's an investor in Stripe. He's a co-founder of Open AI and Sam's I think an investor in Stripe and then uh Elon's working with Daario on compute stuff. So maybe Cheeky Pine is like the perfect Neutrog. I think it'll probably be on uh national TV actually, but uh we shall see. We we'll follow it here. Uh let me tell you about Railway. Railway is the all-in-one intelligent cloud provider.
12:57Use your favorite agent to deploy web app servers, databases, and more while railway automatically takes care of scaling, monitoring, and security. And we can we can move on to the timeline. We can move on to other quickly jumping in. We have a rate hike.
13:10Yes. Tell me about that.
13:12Worsh hiked 25 pips. Okay.
13:15Uh this was first time in three years.
13:18Priced in Cal. She had it I think at like 89% this morning. So not a huge surprise. I wanted to head over to Joe Weisenthal's feed and just kind of read his reaction if he has one.
13:31Well, you pull that up. I'll give you the highlights from the Wall Street Journal. Uh the the NASDAQ react positively up uh 6 point uh 67%. This is the Wall Street Journal. They they need a JavaScript plugin that changes it to 669 or something. I don't know.
13:51Uh uh most officials penciled in one more increase this year. An energy shock and a surge of AI investment have reshaped the inflation outlook. We talked about the Fed interest rates yesterday a lot, but uh the Federal Reserve raised interest rates Wednesday for the first time in three years. A sharp reversal that began taking back cuts as it made last year implicit implicitly undercut the White House's insistence that inflation is not a concern. The Fed is saying it kind of isn't a concern. The increase approved unanimously will raise
14:20the benchmark Fed funds rate by a quarter point to between three and three/4ers point and 4%. Uh the vast majority of officials pencled in more more uh one more hike this year in interest rate projections released after their meeting. So they think there's going to be more rate hikes. Uh Chairman Kevin Worsh vowed shortly after taking office in May to end uh an overshoot of the Fed's 2% target. now in its sixth year and followed through with an
14:50increase that had been widely anticipated in recent days. The rate hike scrambled an account uh of the White House had offered of the man tapped by the president for the for the job in January. Trump and his allies had cast pressure to raise rates uh as coming from a committee hostile to Worsh who last year said he would have cut rates sooner than the Fed ultimately did. It also followed a lost year in the Fed's inflation fight. Um, the central bank has made no progress towards its 2%
15:18goal since mid2025, including after cutting rates three times last year to guard against labor market slowdown. Instead, the Iran war has lifted energy prices in the AI boom has driven an investment surge that has buoyed the economy and markets. Today, policy action will support a timelier return to the committee's 2% goal. The rate setting committee said in a policy statement analysts said that despite intense focus of late on monthly inflation data, the biggest change to the outlook has come from a runup in
15:47energy and commodity prices. It's the fact that the war in Iran has reintensified and the energy price shock is getting bigger again, said William Deadley, the former New York Fed president. Uh what you got for me, Jordy?
16:03Uh, I was just reading through a bunch of different reactions on uh on Bloomberg itself. Let me pull them back up. But they have a live blog. Uh, some people are saying this is more hawkish than expected given that the Fed took away next year's cut. Um, what else? Big changes in the dot plot line. The Fed's September dot plot. We now have four officials expecting to raise rates two more times.
16:31Two more times. It had previously just been one at that level. A whopping 12 policy makers see rates going up once more before the end of the year. And the remaining two see holding rates at their new 3.75 to 4% level.
16:43Uh a reminder that in June, the last time we got these forecasts, half of the committee expected the Fed to hold or cut rates. And again, it looks like Worsh did not submit a dot. So he's going dotless here. Um uh kind of a statement in itself. Uh yeah, it's so I mean so far the AI trade, the buildout, everything has been um overwhelming even in the face of of
17:13headwinds like rising rates and Yeah. Yeah. I mean the the the mood from Silicon Valley was like h we're definitely not booming until we go back to zero interest rates. Like this whole tech thing, it only works when the interest rates are zero. So like we'll just wait it out and Yeah. And the reality is there was a bunch of ideas and investing styles that only worked when rates were near zero.
17:38Um but but yeah, it was specifically when you look at the companies that that really boomed in that era.
17:45There wasn't a lot of net new like really truly innovative stuff outside of financial products which benefited from from low rates.
17:54Yeah.
17:54You know, some of them benefit from high rates though, right? If it's like a savings product, the spread's higher.
18:00But if they have to borrow a lot of debt potentially, but but again, I'm thinking of like lending companies like Pipe, right? Pipe was a company that at the time went from Yep.
18:08you know, incorporation to billion. I forget what their peak valuation was but and it makes a lot of sense because they're basically borrowing at 0% then they're lending to a company at 5% or something you know what whatever their spread is like uh is actually justifiable to a to an earlier piece of the market. Uh but that's what it breaks down when you have to go to a company and say hey you want you want money at 12% or something like that for an early stage company. Well, let's head over to
18:37Who Man, who's uh First, let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agent, man says, "Daria, we should pump the brakes on the frontier." Sam, yeah, let's all slow down together. Meanwhile, Zuck, the the funny thing is, yeah, I mean, let's let's talk about what Zuck wrote.
19:01He said, "Last month I wrote about how we can build a positive and safe future for everyone. This is when Zach said, "I really want people to understand my values before we come out with our most powerful AI ever." Um and uh I at least felt like we we already had a good understanding of Zach's values. But uh he wrote yesterday, "Every lab has the responsibility and incentive to move at the pace required to train its model safely and the ability to take its own actions to ensure that happens." The
19:31reality is people don't want to use agents that are misaligned with them and don't do what they ask. So labs have a strong natural incentive to make their models more aligned right away. you know, starting out uh with this point, it's like this is not the this is not the no one the safety debate has not been s like around the idea of oh, they're going to create personal agents that are going to be misaligned to the users like this. No, this is not at all.
19:59It's just this is just a point that doesn't matter.
20:01No, no. I mean there is there there is a like it's not the safety crowd but the whole like social media is brain rod addiction like like that is something that uh there is a separate crowd that does critique that and says like I don't want the addictive flywheel of of maximizing screen time to be brought to AI totally separate debate.
20:23Yeah. No, I agree. I agree. It it's not it's not the the the true AI safety debate.
20:29Yeah. Um but it is a debate. There is a lot of debate about slowing progress on capabilities until alignment catches up.
20:36My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate. Again, like this is not relevant to the current safety debate.
20:47Uh obviously people want to make products that do what their customers want them to do. No one has been worried that you can't make a model that in the near term or in the medium term or even over longunning tasks can generally do what the user wants. The concern is that that if you leave models and and give them a task that that is more expansive that they can start to do things like hacking hugging face. Right? So again,
21:14nobody is sitting here saying um the main thing is that this this is talking past the X-risk question. He's completely dismissing the discussion which which Dario is like laser focused on. And so this feels like it's a rebuttal, but it's actually like talking past it in the sense that he's like labs face significant liability. It's like well in the X-ray scenario the liability doesn't matter. That's the whole point is that like no one's going to come and and here's here's the best here's the best line.
21:42Meta delayed shipping muse for several months to focus on safety and security.
21:46That's actually very rational and important because every company delays their products for months. That's just called building a good product, okay?
21:54To make it safe and secure, like period is what Meta has been doing forever. You have to do that. You have billions of users. You got to make sure they're safe and secure. Every single product.
22:05Again, uh I I I felt like he was But you So So there's a lot of things in here that are rational.
22:11He's he's taking a victory lap on not taking a victory lap. Don't you realize that? He says, "We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing to do."
22:22And again, every single company already does all of these things.
22:26Every single company victory laps while they're doing it. He's taking a victory lap for not taking a victory lap.
22:34I don't It's not that complicated. He says, "Everyone else says like we're taking safety seriously. We want a pat on the back before we delay the product." and he's saying, "I want the pat on the back after I delay the product."
22:45Yeah.
22:46I just think it's I just There's a lot of back padding going on.
22:49Okay.
22:49I just think it's I think it's I just think it's uh very funny to I I think it's funny how much all these points are are fine. They they generally are rational and they make sense.
23:03Yeah. But I think it's funny that people are giving him so much credit for this note given that he's totally missing the main point that everyone else is focused on.
23:15X- risk like intentionally missing the point because he doesn't believe it.
23:20He has in order to get in order to get in order to get brownie points from people that don't even understand the current debate.
23:28No. To get brownie points from other people that have a PDM of zero who are like Yeah. And and you can see who's who's supporting this. They're like, "Yeah, thank you." Like, "Just do put put the put the agents in the bag, you know, make the tokens free and and just make the products. Like, I'm not worried about that at all." And for that crowd, they're like, "Thank goodness you didn't like fall in the hole of like stooping to this pdoom debate that I don't take seriously." That's the side.
23:52I just wish that he would come out and say, "I have a pdoom of zero. We're just gonna make No, he's not. He's not. He's trying to position. He's saying my view like that's what he's saying.
24:04That's not that's not he's not being explicit about that. He's trying to let people say we care a lot about safety. We slowed down our development because we care about safety. Trust and alignment are important.
24:14Yeah.
24:14Right.
24:15He should definitely come out and say PDM zero because if it's not zero and it happens and we all go extinct, no one's going to be able to dunk on him, right?
24:23So it's pure upside.
24:25Pure upside.
24:25Pure upside to be P Doom zero guy. Why has no one considered this? The aura gain is so high.
24:32Yeah, I would I would be I would respect it a lot if he just came out and said what he actually thinks, which I I do believe you're right, which I he has a PDM of zero and and his P abundance is high. I mean, that's what he said in the in the in the previous essay. He was he was basically like I don't think the he he he he even he even was gesturing towards like the the fear-based marketing the doombased marketing is just a marketing tactic. I don't think it's rational and also I don't think it's good for people to be in that
25:02headsp space like and it's like info has a problem one one of the last lines committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure up we develop this technology safely. Meta has made this commitment and other labs can do this as well. Look like you you there's absolutely zero shot that Zach walks into MSL and like gathers the researchers and says look I don't want to make models that make our models
25:31better. I don't want to do it. I want you guys just focus like that there's zero zero chance. So again like this to me is just like he is being disingenuous with his positioning of almost every single point here.
25:46Uh I don't know. Uh I do think that there is a trade-off right now between uh making models good at things that are not on the RSI path and those that are.
26:00And so like the the the the race to become really really good at coding is super aligned with uh RSI. The race to do image generation is not and image generation does not seem to be on the RSI critical path. Although I I think DeepMind put out something where they're using world models and they think they have a breakthrough there. I don't know if if I saw that accurately, but um but maybe they're wrong. But at least like the bet at at least enthropic has been like we don't need to be world
26:29class at image generation to get where we want to go because we just need to be really good at coding. Coding teaches us teaches the model how to train new models and then we get you know and then and then and then then that that final model we can ask it to spin up an image generator if we want. Uh Zuck is saying the opposite. He's saying like yeah we will actually go and try and build a tool just to help you book a dinner reservation. Uh, and that's a good use of compute. Probably not on the RSI path and and that's like a reasonable
26:59trade-off. That feels that feels real to me. I don't know. Uh, what do you think about engaging independent evaluators?
27:05Is this like he says it's already industry best practice? Is he talking about benchmark stuff or is he talking about like actually like the you have Slack access, you have a desk, you have a badge, like you don't work here but you're allowed to just go wherever you want. Um, I think that that's the next step and I think he's maybe talking past that a little bit.
27:26Yeah. And he's talking past this again.
27:28Engaging independent evaluators and advisers is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. It's like other labs also do that too. Also do that already.
27:43Well, as of last week, Enthropic does this with meter. Like they said that they were going to do that immediately. So they were doing badge and slap.
27:50But but he's not he's he's not saying that he's doing that.
27:55He's saying that he's saying that he's doing the thing that every group that's making model has been doing for the most in the prior era for a long time.
28:03But the new thing is badge and slack access and desk even though you don't work at the company. And that's when everyone's like, "Whoa, that's crazy."
28:12Because these organizations are very, very secretive and you're really going to let this nonprofit come in. It's like sort of a wild move. And that's why people are like, "Oh, wait. Like, how aligned are they?" And like, "What is the what's the knock-on implication?
28:25Like, can these people not leak? Can they not can they be trusted? Are they going to go to a cocktail party and be like, "Oh, yeah, the new model is actually really bad or whatever." Like, there's so many things that are like like Meta deals with leaks all the time.
28:38And so the prospect of bringing in a nonmp employee who is there explicitly to whistleblow effectively and like is allowed to talk about anything and like is is you know third party evaluator like that is a huge step and I think that's why people are like whoa this is a big deal this is a big proposal like like if this was not a big thing if anthropic was just like oh yeah we're going to do a benchmark with meter everyone would be like yeah that's fine cool do that for sure awesome but like
29:07people are Wow. Okay. Like Meter's gonna have a have Slack access at at at Anthropic and you're asking other people to do that. Like that's we got to know who these people are. We got to make sure that this is like the right team for this. This is sort of a crazy thing.
29:19This is sort of unprecedented. This doesn't happen a lot. Like this is new and and so yeah, the newness is not fully embraced here. Um and I think I I think that's a little bit of like okay, you're not really engaging with like what's coming down the pipe, which is like maybe a government employee in your building. That's not happening right now. Maybe maybe a nonprofit from Berkeley. Do you like that, Mark? How do you feel about that? Are you cool? It seems like you're cool with it, but I don't know if you're actually going to be cool with it. Uh because it is a little it is a little wild, right? It's
29:47a different thing. It's a new thing, but uh it's it's a modern it's a modern.
29:51Pull up this image.
29:53What image are we pulling up?
29:54I want to see John's reaction.
29:56Okay.
29:59While we pull this up, let me tell you about public.com investing for those that take it seriously. They got stocks, options, bonds, crypto, treasuries, and more with great customer service.
30:07Here we go, John.
30:08Let me see.
30:09Boom.
30:13This This meme's been applied to like seven different people this week. I know, but I think it's I think it's particularly relevant because you have you have Elon, Demis, Daario, Sam, all you know, all these people from different factions that are at or I've seen this applied to Theo. I've seen this applied to Coher. I've seen this applied to Deep Mind and Gemini. I've seen this people MSL.
30:39People are just whatever shot they want to take, they're applying this. It's not it's it's too broad at this point. There's people I think it's the the inverse. It's all clowns and there's one tactical soldier.
30:49There's groups that are at the frontier.
30:52They're on the battlefield and they're deeply concerned.
30:55And you have Zach coming in here saying, "Yeah, like I I built a cool personal agent. It's it's aligned.
31:03What's there to worry about? Nothing to worry about."
31:05Yeah.
31:06It's just you're not really you're not really in you're not really on the battlefield yet. And I'm not saying with watermelon they can't get there. And Muse seems like it's Muse is getting amazing reviews. It's it seems like an amazing product.
31:21Uh, and that's very exciting, but I think it's just again I I thought the whole post was silly, just like I thought the last post was silly.
31:30Nick Carter liked the post. He said, "Zuck pretty handily dismantles Daario's talking points here. People want models that are aligned with them." Suddenly punches back at Anthropic's normative constitutional approach. Labs already face liability if they screw up. So incentives to release aligned models is already baked in. Meta delayed Muse for alignment reasons. Subtly questions anthropic trying to kingmake meter implies meter isn't anthropic pathy.
31:55Meter meta doesn't need to coordinate with anyone to work on alignment. Just something labs should naturally do. But Mark Chen from the top rope according to Kevin Roose says there aren't race dynamics off the frontier. And uh yeah, it's it's a it's a it's a war on the top. Again, it all comes down to like there is absolutely zero shot that Zuck doesn't want to get to RSI yesterday and is doing I thought he said that they were working on RSI.
32:22I thought that was like announced as like a explicit goal maybe three to six months ago. He was he was like super intelligence RSI and and again like I I just feel like it's not unique to Zach. Many many many leaders operate like this. But he's he will say whatever the pick me thing is at that time.
32:45It would be more aggressive to be like I'm PU Poom zero and I'm racing to RSI. Like I'm Poom Zero and I'm racing to RSI all these other all these other people.
32:57I'll see you on the other side. Brother Meta he said in August 10th 2026 meta must must build out a sufficiently large amount of compute that we can allocate enough to recursive self yeah they're doing RSI they're doing RSI everyone's trying to I mean RSI can also mean so many things it can mean yeah you you looked up the you summarize some archive papers for new strategies like the the the automated
33:26research intern is in part an RSI initiative. It's not the final RSI loop, the closed loop RSI that people are are really worried about, but uh it's it's all gradations of this. Uh what does David Saxs think about this? I imagine that he's a fan. He says Mark Zuckerberg, he just quotes it. He says meta delayed shipping muse for several months to focus on safety. Yeah, Sax is very much like if you want to slow down, go right ahead. And so he is uh he is
33:55he's he's fine with that.
33:57All right. Anyway, uh the reviews from are really good.
34:01So, let me let me say let me say this because I've been I've been a little harsh.
34:05Yeah.
34:06Uh the people people genuinely love Muse. I think like it seems to be a great product. The the race between Muse and Instinct is already uh already quite exciting. N says, "My wife tells me Muse is good and quote for the girlies." Yeah, that's extremely bullish.
34:28Extremely bullish.
34:28But, uh, but yeah, I mean, insane distribution advantage. Uh, I think it's Yeah, let's see where it's at on the charts. Uh, number three, some We got to Nick, we got to get this company vented on here. I just do not I've brought it up like a million times. It's like you have the most insane AI race, gambling race, short form drama race. Invented is always in the top five. Wow.
34:55Just like a secondhand clothes retailer competing with like a million other secondhand clothes retailers.
35:00Yeah, but they'reating.
35:02Vinted. There we go.
35:03Like Meta is like pouring every pouring pouring billions of eyeballs into growing muse and vintage is Didn't Wait, didn't Gemini and Google announce a personal agent at IO this year? Where did that go? because that it's it's very odd that that Gemini has so like yes the distribution advantage from from meta platforms is significant
35:32but a lot of people are on Android a lot of people already have um Gmail and Google calendar and they have Google maps and they have the phone number of every business and so when you think about booking restaurant reservations doing all this different stuff. Um, they uh you would think that that product would have gotten more traction, but I feel like Gemini Personal Agent did not
36:01go through the same hype cycle that uh that Muse is currently on. Gemini Spark, it died on the Vine. Yeah. What happened with Wait, they called it Spark.
36:13Is that But isn't it Muse Spark?
36:15Yeah.
36:16Dylan in the text chat says Spark.
36:19Everyone's calling He uses it.
36:21Oh yeah, it's mostly just using Gemini and toggling.
36:24Okay. Yeah, it hasn't had the the most breakout things there. There are some killer Muse use cases. I mean, everyone's talking about booking reservations. Uh as I couldn't describe what it did. No one knows. The world's most mysterious. It's the most mysterious agent yet, but they got to have something soon coming. And I and I and I do feel like the the the Gemini distribution like they still have a pretty significant share of the chat market
36:53pops up in your email and it says, "Oh, here you can just email this thing and yeah, you know, maybe another Gemini tab that I can open. I can open three if I'm in Gmail and Chrome and uh pull them all open." I don't know. Um, what happened to the to the uh um to the Kulchi uh AI price tracker? Did you see this? We we talked to TK when this launched and I was like, "Oh, this is cool. This will
37:21allow you to understand like you know basically a proxy for the AI buildout."
37:25Like how are how are GPUs trading?
37:29Banned apparently. Uh before we get into this, I'll tell you about Figma agents.
37:34Meet the canvas. Your AI agents can now create and modify your Figma files with design system context. So this is from uh Semaphore and uh the U uh do we have this open here? Um uh the uh the US commerce department last month ordered call sheet to take down one of its products tracking the price of AI compute the crucial power from data centers that's driving the artificial intelligence boom if you
38:01didn't know what AI compute was. Um the uh uh commerce officials cited national security concerns and that's what stuck out to me. I was like I there's so many different uh prediction markets that I can easily trace through. Like oh you have a flight delay one and you could have somebody that calls in and tries to get the flight delayed. That could be very disruptive. The FAA could have a problem with that. But this one I wasn't I wasn't worried about at all. We talked we talked to Tar about it and we like
38:29yeah this one seems s sort of informative and interesting. Um, so the product pulls together data from several markets that allow users to bet on the cost to rent NVIDIA chips to create an overall picture of where AI compute costs are heading. Quiet uh call sheet quietly complied though many un of the underlying markets remain open for trading. Separately, commerce has pushed the commodities futures trading commission which oversees prediction in future markets to effectively freeze approval of new compute contracts for 60
38:58days. And so even though some of the contracts are still open and will close maybe in 60 days there won't be any compute prediction market futures which is very interesting. Khi declined to comment while a commerce p spokesperson said the department quote has never once asked Khi to take down this market or any other markets. Interesting. So uh the commerce spokesperson said this story is false to semaphore. So lots of people going back and forth. Uh it's unclear why the commer why commerce is worried about the naent market which
39:26aims to do for AI computing what oil futures do for crude let buyers and sellers of compute lock in prices and give traders a way to bet on where those prices go. Now, um the markets were always a little thin. Uh you know, these new markets, they tend to be sort of thin and they tend to be more speculative, just people that are trading them, uh on Vibes or news, not necessarily used by the actual companies like you know oil futures, those will be actually employed in treasury strategies
39:53by like airlines uh and they and and like big big firms will trade those in size. Um but that was what K uh that's what Khi and Tar were pitching as like where this all goes. I was a little skeptical that it would wind up on call sheet. I thought that they could have a decent chunk of the market, but I thought a lot of it might just go to insurance companies and reinsurance companies and sort of like the traditional financial rails that do these types of deals. But uh ne nevertheless, it's going away and let's figure out why. One potential reason
40:21floated to Semaphore uh by market participants is that compute futures could be manipulated to show a sharp drop in the cost of older chips which might destabilize AI stocks and debt markets. Some of these markets are thinly traded which could lead to volatility even without bad actors. So you're trying to wipe out situational awareness. You short the khi prediction markets on AI compute futures. Everyone thinks, "Oh, AI is bust." The market
40:51trades down for a couple days, you clean up, and then you buy back in or something like that. I guess that's what the the rumor is here that some reporting on. Uh the cost of comput has become one of the most important numbers in the US economy. One side of the debate fears that older chips, which serve as collateral for billions of dollars of borrowing by NeoClouds, Neocloud, fun fact, coinage by semi analysis. I didn't know that's where it came from, but they actually they didn't just rank them all with ClusterMax. They actually coined the term Neocloud. Uh it
41:21had a rude name beforehand, which I didn't realize. You can go look it up.
41:25I'm not going to say it. But um on the other on the other are concerns from big companies adopting AI that shortages of power and infrastructure will send prices of tokens soaring. Uh that uncertainty has given rise to a futures market that was starting to take off this summer. The CFTC's 60-day pause could delay plans by exchange operators like CME and and NYSCE uh in parent intercontinental exchange along with upstarts like Architectural Financial Technologies to list two-sided two-sided betting parlors. Interesting. Well, let
41:54me tell you about Cisco and Jordy will pull up the next story. Critical infrastructure for the AI era. Unlock seamless real-time experiences and new value with Cisco. Where you want to go next, Zach? Okay.
42:07Yagari.
42:08Yeah.
42:09Is back.
42:09Where do we know him from?
42:10He says, uh, Cal Calai. Cali been on the show.
42:14Yeah.
42:14Uh, he says, "Everyone's been asking what's next after selling Cali, introducing Persona. Others couldn't figure out the interface. We did. Let's pull up the video." Uh, I we got to ask Zach how he he how how his deal worked to sell Cali because he seemingly had to stay for like a week. Like he ended up, however he worked it out, he basically just got to to bail.
42:38You can't keep this guy out of the arena. You can't get him on the sidelines. It's impossible.
42:41Yeah. And Cali is the number four fitness app still cooking. Yeah. So good. Uh yeah, let's play this video.
42:50Every major AI company is racing to build the best assistant, but they're all building within the same 2x6 in rectangle we wear in our pockets every day. We believe AI is supposed to free us from our screens, not make us use them more. So, we built a new interface.
43:04Introducing Persona and the Persona Band. To talk to your persona, simply flick your wrist or hold down on the face, then ask anything. Hey, can you ask Mac how many pre-orders we got today?
43:20Sir, I just emailed Mac. I'll let you know his reply.
43:24Your persona will proactively learn your habits to make your life easier. For example, when you get off a flight, it should know to get you an Uber from the airport to your hotel. Reactively, it will be able to help with all kinds of things like cancelling subscriptions, negotiating bills, or even door dashing your usual Chipotle bowl, knowing exactly what you like.
43:46Elite Persona is also the only AI.
43:48It's the predictive burrito.
43:50It's what the Door Dash guy was talking about.
43:53Put your B.
43:54They order the burrito before you even know. He's actually building it.
43:59Privacy for your and those around you's data. This isn't a continuous listening band. The microphone will only activate under your control. We are launching with three different materials and four different colors. Pre-order your band and start texting your persona today.
44:13I've seen enough time for Zach to come in with the five offer.
44:17Yeah. Uh pretty good pitch.
44:22Little bit harder to get people to install it, right? Because they got to buy it, wait for it to ship. little bit of a barrier to entry, but also love these AI devices. I think cool. Yeah, Pocket really worked.
44:34Yeah, I don't know. There's there This is He's gonna He's gonna really blow this thing up. I have a feeling this is going to be big. That's cool. Uh what did Prompter say? Never bet against a guy who managed to sell a calorie tracking GPT rapper, which in no way could have possibly been accurate at calculating macros for 100 million. No.
44:51uh he uh he is a master of of of growth and and scale and just like so if you look at if you look at the the charts right now Cali is number four the company that bought Cali is number nine.
45:06Wow.
45:07Yeah. I mean that's a good that's a good example. It's a good growth team. Uh doesn't he also have another company EcoGPT or something? Or is that not him?
45:14No no that's his co-founder.
45:16Got it. Okay. Yeah that's a separate company. Um, yeah. So, I'm I'm still Yeah, we Tyler pre pre-order one of these.
45:24Oh, yeah.
45:25Um, do that.
45:26Put it on the ramp.
45:27Yeah.
45:27Um, so yeah, I'm excited to check it out. I'm still not uh I'm still overall not convinced that I need a new device.
45:34Like there has to be something.
45:37Yeah.
45:38Like a lot of the demo was was screenshots of of the phone. Yeah. And so and and I do spend time away from my phone, but I'm typically in the water.
45:53Is it waterproof? Can you wear that while you're surfing? You said you had a brilliant idea while you were surfing. Needed to hold it in your head the old way before you got back to your phone like an idiot.
46:03Like an idiot using your brain. Uh, no.
46:06It it it really does feel like a a wave of new devices is is coming online and the and the the the war for the the the home device. I mean, you can see, you know, uh the battle between Apple and Meta reemerging with Muse like how how deep will they be able to go before Apple pushes back? Uh every other firm is working on hardware. The Meta Glasses, the Metag Glasses with Muse should act very similarly, right? you have an agent, you can talk to it, you
46:36can do some of the similar actions. Uh the the trick is like people don't wear the glasses all the time, whereas a wristband like a Whoop, people wear that all the time.
46:46Um and he seems very in tune with uh I don't know, just like the the obvious push back like calling out that it's not listening all the time. Yeah. Like there are just people that would just like forget about that and it's like it's very using the burrito using the burrito example. I'm struggling to see when I would need a burrito but wouldn't be near my phone.
47:09It's more just like if you don't if it's faster. Like people just do anything that's faster. That's how we got to phones. Like you used to you you there was a time when it was like I know if you're going to shop online, you go on desktop.
47:21I'm I'm not saying like the overall form factor is bad, but if you had the ability to just hit a big button and talk to the Persona app and say, "Order me the usual from Chipotle." It's not like the band is necessarily faster if your phone is sitting right there.
47:36Yeah. I wonder how fast they're going to update Siri because it feels like No, it is. It is so funny to me that Apple's finally coming out with their like knowledge retrieval LLM chat app. And right literally right before they could really get it out to the public, there's like a whole new paradigm of agents that are you can just tell it's going to take Apple like two and a half years to get a beta out.
48:00Yep. And so again, like right as it felt like they were, oh, we're going to catch up, you know, will it will it will it is there a way that that the next version of Gemini is so good, it if if you if you've talked to Gemini in Gmail, it links it and so when you talk to Gemini through Siri, it knows that it's you and it and it can go in your email and it can email people and and and text that restaurant to get the reservation. Like is that a fe I'm
48:28always interested in like how much innovation will Apple get for free because they're partnered with Google?
48:34Like if Google advances the model, does that advance automatically come? Or is Apple like we're buying Gemini 3.1 Flash from you? Thank you. We'll let you know if we want the new thing because if they just want Gemini and all the progress like there's going to be progress there. There's going to be new capabilities.
48:52The models are going to get better. The functionality is going to get better. I would bet that even though people were sort of Yeah, Google has just really really really struggled to make good AI products.
49:04Yeah, I I don't know. I I I would be surprised if they can't figure out a way to order you a burrito in the next six months, right? Like that's not some like RSI critical thing. It's like I don't know. You think they you you would think that uh having spent billions of dollars buying Windsurf that they could have a competent codegen product.
49:26Yeah. Haven't seen Yeah, it's tricky. I don't know.
49:29Like there's small companies that have made good coding coding harnesses.
49:33Yeah. Uh anyway, let me tell you about Crowd Strike. Your business is AI. Their business is securing it. Crowdstrike secures AI and stops breaches. Uh we have Jeremy Aair from Circle. You talk about Sam first.
49:47Uh let's bring let's bring uh let's bring Jeremy Aair from Circle, co-founder, chairman, and CEO back on the show. Welcome to the show. How you doing?
49:55Welcome back.
49:56Uh crazy week, tons of news. Let's start with Circle. Let's start with uh what's new in your world.
50:02Yeah. Well, so we turned on a new global operating system today called Arc. It is an economic operating system uh that we've been building for two and a half years. Hundreds of companies launch their apps and services on it. Um, it's being run by uh not just Circle but like many of the biggest financial infrastructure companies in the world, the people that clear the securities of the world, run the payment networks of the world, manage the assets of the world. Uh, and so it's got everything from social trading and DeFi to major
50:30payments firms and capital markets and um, you know, we we built this uh, for AI first. Everything is AI accessible, all the builder tools, everything. But also we we built services so that AI agents can use this themselves as an economic layer for for transactions, for contracts, for coordinating. And it's a big it's a big milestone. It's the biggest um really the biggest platform launch in our history and and we think it's a it's a big uh it's a big upgrade for the internet too.
50:57Amazing.
50:57A lot of people six months ago would be like awesome it works with AI agents. But today they'd be like definitely don't make it work with AI agents. That's not what we want. Uh, how do you think about the risks and the rewards that come from enabling a platform to be AI agent accessible? Sometimes they'll access things that aren't even accessible.
51:17I mean, this is the key, right? The the biggest challenge with this explosion in agents is basically trust.
51:23Yeah.
51:24And verifiability.
51:25Y and so, um, you know, it turns out that cryptographic computing and these these these networks, these network operating systems of cryptographic computing are built for that. Mhm.
51:35And so we've we've kind of created the most trusted dollar on the internet with USDC. And the these operating system layers actually have features that AI can build on top of. So you can basically have proofs of the of the entities and identities behind agents.
51:51you were actually demoing today something called ARC agent sector which actually allows you to have an AI agent prove the work it's done the data it's used how it's done its work and then provide that cryp in cryptographically provable way to the to the internet so you can have more trust and verifiability around what actually AI is doing and right now the blackbox issue is the scariest thing for people out there whether it's inside the labs or it's actually just out with the agents that we're building and so it's
52:19this convergence I you I talk about the agentic economy. It's this convergence of these operating systems for intelligence and these operating systems for economic activity rooted in in cryptographic computing, which is what these blockchain networks really are.
52:33And so we're building a layer um that is able to begin to provide that trust and verifiability to the agentic world. And so that's how we get comfortable with with uh with it. We we need this uh we can't move forward in the agentic economy without technology like this.
52:49Are you uh are you equipped to explain to me the history of Walmart's misadventures in payment infrastructure?
52:58I was reading on stratey that they were trying to push people to debit card and a rails for years and then they finally sort of succumbed and went with a different one. And I'm and I guess I'm just interested in like is like is this a transformative moment for any particular sector of the economy where you're like oh someone's going to be saving half a percent now or some big pool. Yeah I mean for sure. I mean look when we started circle right we had this basic idea that there could be a protocol for
53:26dollars on the internet that the marginal cost of storing and moving value just like the marginal cost of storing moving data and information and communications went to zero. that happens with with the movement of money and and now we have that like stable coin digital dollars are now as of January of this coming year like legal digital dollars in the financial system.
53:44You can move them programmatically instantly anywhere in the world in a fraction of a second for a fraction of a cent. And so payments and settlement becomes commoditized.
53:54And so as you get the proliferation of of wallets that can talk to these computer networks which is now everywhere like from cash app to revolute to all these wallets around the world can talk to these networks. Well now you can directly settle. So if I'm a business I can just say you know pay me pay me over this rail and I get it as basically cash instantly and and it settles uh you know like a cash transaction. And so um over time basically that compresses out this rent extraction that's existed uh for a very
54:23very long time and all kinds of other things like time delays and and and the like but and so yeah I mean it's it's purpose-built for that world and a lot of stuff is converging all at once operating systems like Arc the legal frameworks for digital dollars the people getting comfortable that they can like use this in in their in their businesses know how to operate it etc.
54:43And so it creates a huge opening for, you know, the Amazons and the Walmarts and and really you talk talk to any major retail basing CEO and and they'll tell you how much of their margin goes to processing fees. And so I do I do think it's um it's significant, but the the whole utility model of payments is going to change um very very significantly. That's a big part of of what we see happening here.
55:06You mentioned what's the most boring payments happening on on Arc? Like the agent stuff's awesome, but I imagine you have a bunch of different partners.
55:16I mean the boring the boring stuff is is like you know there are businesses that have to move money around the world and and they're finding that they can do it better and they if there's a someone who's importing from Asia in Latin America and they need to do that and they and they want to like have that done quickly and they don't want to have a lot of intermediaries and they want to just like directly do that and so volumes of of kind of how how people need to move money around um is you know it's boring to to all of us but to those to those people in those countries and
55:45and to those suppliers and those other people, it's actually really significant. I mean, other boring stuff is like, hey, I have a I have a tokenized version of a stock and I'm someone in an emerging market that wants to own a stock. I can now do that and I can I can purchase it and trade it instantly. I can take out options on it instantly and I don't have to be in a US brokerage account. I can just be on the internet with a wallet that connects to these networks. And so, you know, from, you know, from the average individual who just wants to participate in capital
56:14markets in in simpler ways to these businesses that, you know, have to move stuff around that's really important to the, you know, the the kind of efficiency and cash efficiency of their business. Um, yes, agents are exciting. That's frontier, uh, uh, so to speak.
56:29Um, yeah, and then even even simple things like people love to watch memes on Tik Tok. They also like to trade memes. Uh, and so if you want to trade memes, go for it. Okay. Um, sprinkle some of those in.
56:42Uh, I want to talk uh legal. You mentioned uh the that these are now legal US dollars. Uh, but yesterday, CoinDesk reported uh Crypto Clarity Act flames out in failed US Senate vote. The yearslong effort to set US regulations for crypto markets couldn't muster enough support to make the leap over the Senate's final 60 vote hurdle. I would love your reactions, context, history, anything you can do to get the audience up to speed here.
57:11The key thing is that a year ago, there's something called the Genius Act, which did become federal law with massive bipartisan support in the Senate and House. And the Genius Act makes digital dollars legal in the US financial system. The Federal Reserve, the US Treasury Department, all the regulators have made the rules. And as of January 2027, stable coins like USDC become legal digital dollars in the US financial system and in the global financial system. So a lot of our work is done. Like blockchains running this new money, being able to support it with
57:40all these markets that's here. We've got that which is great. Clarity is about like how trading venues get set up. You know, who canate in a trading venue? The tra the digital asset trading and markets side of things. There's other pieces to it. clarifications around like how DeFi is is sort of treated and and stuff like that. But like that's what that's like mark its market structure which has to do with like the capital markets. And so I think um you know uh there's still a lot of desire to have like good regulations there. But at the
58:10same time I mean essentially the agencies that regulate these things like the CFC, the SEC, even the bank regulators too, they're all just going to move forward and and regulate it themselves. Um, and then eventually we'll we'll get additional laws uh passed through Congress.
58:24Got it. Um, Jordy, anything else?
58:26Not for now. Congrats on the launch.
58:28Congrats on the launch. Thank you so much for coming on the show.
58:30We'll talk to you soon.
58:31Great to see you. Have a good one.
58:33Uh, let's talk about Sam Parr. He was offered $600,000 to sell his company's data to Micro One. He's not doing it. 100% not doing it.
58:45He turned it down.
58:47Uh, he says it does interest me like crazy. What a crazy business model. He's tempted. Yeah, he's tempted.
58:53So, apparently, uh, let's finish his post. He says, uh, if you don't know, here's all I know about them. They go to private companies. They ask for your Slack, notion email info, offer you six, seven figures. What are we doing here?
59:08Somehow anonymize that data and they sell it to OpenAI, etc. And the interesting thing is that I've been getting these Instagram ads and they're trying to get me to sell the OpenAI data to Micro One to sell the Open AI and I think I could maybe one hand washes the other on this one, you know.
59:30No, I would never to be clear. Uh seems like Micro One has grown to nine figures in revenue very fast. Uh the ad on Instagram that that at least I'm getting targeted with is1 to$2 million offer, but maybe that's how they get you in and they work. You look, you are a podcast. You only need this much. I don't know.
59:48But uh uh he says, "What else am I missing here? I get the value of this, but doesn't seem right to me." Uh interesting. It I for some companies, I would imagine this being super valuable. The the range seems really really wide.
1:00:03I don't know. But uh what what do you think about data brokerage in the modern era? Are you selling your data?
1:00:08I just think it's it's you know that in our team chat like you got few people working back and forth on a meme.
1:00:17Yeah.
1:00:18Bunch of crying emojis, some laughing emojis.
1:00:21We know we know that the models are bad at humor. How will they get funnier if not training on our internal message data? They need to see, oh, that one got five crying emoji reactions. Reinforce that. That's what happens. This is how we create comedy super intelligence through acquiring the Theo von data trove.
1:00:43Yeah. I mean, I think it's I think it's quite interesting to look at how projects and tasks get completed in organization, how products get launched, how how deals get done, all these different things. It's just like Yeah. Yeah. Wait. Uh, I have an interesting question for Sam. We'll have to follow up. Uh, which company? Because he is of course affiliated with many.
1:01:05I think I would assume Hampton.
1:01:06Hampton would be the one that Yeah, that that actually makes sense because again, you wind up in the situation that it's like, oh, we'd like to buy your data from my first million and everything HubSpot has as well. You know, just just give us access to the HubSpot Slack, right? Like I don't think that's going to happen. Uh, they probably don't use Slack there, right? CRM competitor.
1:01:25Probably use something else. Anyway, uh let's we can ask Tomas what he thinks about selling okay your company's data. Yes. Let's pull up this video of PT talking about why the name of your startup is predictive of success or failure.
1:01:41Oh, did you notice what Jeremy did there? Jeremy a layer. He was like, "We built a layer for transactions." Yeah.
1:01:48Nominative determinism in action. I caught it. Did you?
1:01:52Let's play the video. This is sort of like a slight uh um um aesthetic thing I I believe in very strongly is that the the uh names of companies are often very predictive of future uh failure or success. So uh so I'd say PayPal was a um was a very friendly name. It was the friend that helps you pay. Um Napster was a bad name. It was the music sharing site. You nap some music, you nap a kid.
1:02:16That sounds like sort of a bad thing to be doing. And uh it's no wonder the government then comes in and shuts the company down within a few years. So you want to be very careful how you uh how you name companies. Uh in in the sharing economy context, I like Airbnb way more than Uber. Airbnb sounds like this very innocent virtual bread and breakfast, this very light, non-threatening sort of company. Uber that sort of sounds like a bad name from Germany sometime in the 1930s. Uh you know, uh what are you
1:02:44exactly above? What the law? Um and um and uh this is probably something that again from a um government regulatory perspective I think Airbnb is a vastly better name than uh than Uber. Um and on the social networking side um I would say that uh that I I actually think Facebook was a very good name. I think um MySpace was sort of a more problematic name. Um you know Facebook was um you can say that all these social networks involve both um reading and
1:03:12writing. Unlike real life, uh you learn to you have to write before you read. Um you first have to write some things about yourself, then you read more about other people. Over time, reading dominates writing. Uh Facebook was about learning about people around you, um about the real identities at Harvard.
1:03:30MySpace started among wannabe actors in Los Angeles and it was about them coming up with fictional narratives around themselves and then sort of a lot of other people in LA who are generally like that. Um and um and u and because um reading dominates writing uh Facebook would ultimately dominate my space. So I think you could you can sort of um there's a c certain version where the whole arc of the company was um the whole product arc was implicit in the names.
1:03:57Interesting. How important do you think naming is?
1:04:00Extremely important. TBPN I'm overly obsessed with uh dotcoms at at the inception stage of companies too because there's comp there's one they they go hand in hand but names are names are important names are I feel like names and do and dots are are maybe maybe name the name itself is like 70% and and actually having your.com and owning that is like maybe 30% over time
1:04:30but when I see founders that are super ambitious and they plan to build a really big company that millions of people are going to interact with and then they build they pick a name where they will never ever ever get the unless they can uh unless like we actually just saw runway the AI company bought runway.com from Siki's oh no seeiki Chen's company and again I I expect that runway had to pay like an insane price
1:04:58uh because you know Siki He had had a wellunded company himself and leverage um anyway so important um bending spoons great name we do have our next guest though bending business later we can get into this next uh let me tell you about console while we bring in our next guest console builds AI agents that automates 70% of IT HR and finance support giving employees instant resolution for access requests and password resets we have
1:05:26Tamas from theory ventures back on the show it has been way too long. Thank you so much for taking the time. Uh I've read a bunch of stuff that you've written, seen a bunch of stuff, and I'm so excited to get into this talk. How are you doing?
1:05:38I'm doing great. Thanks for having me back on the show.
1:05:41Uh to have you.
1:05:42Are you thinking about AI regulation as much as everybody else?
1:05:47It seems like it's the topic of the week, that's for sure.
1:05:49Okay.
1:05:50You know, we saw a Zuckerberg tweet uh yesterday. I thought with actually a very sanguin view, probably one of the most level-headed views.
1:05:56Okay.
1:05:58Debate. Jordy thought it was talking past the X-risk discussion that he wasn't engaging with the real discussion Dario Amade is putting out.
1:06:07No one think let me pull up the the post again. My point of view like totally levelheaded, totally rational saying like if if Zuck had a leading model and he was talking like that, it would be everyone it would actually truly be incredible. the the reality is that his the the sort of bright spot with MSL right now is Muse, which is a great product. People like it. And literally no one is worried about like personal
1:06:37agent app safety. And so when he says people don't want to use agents that are misaligned with them and that don't do what they ask. So labs have a strong natural incentive to make their models more aligned. My point of view is like no one of course people don't want to use products that don't do what they want them to do or do things that they don't want them to do, right?
1:06:58Uh but that is not at all like the debate right now. And so I feel like he came and made a bunch of points that like sound good if you don't fully understand the debate. Like later on he says committing the the significant majority of compute towards serving people rather than racing towards RSI is one of the best ways to ensure we develop this technology safely.
1:07:17literally about just over a month ago, he was saying like we need to have a lot of compute so that we can keep up and get to RSI, right? And so I feel like he's consistently been focused on saying the thing that will make him sort of popular in the moment that sort of misses the bigger picture.
1:07:35But overall, our takeaway from all of his actions is that he actually has a pdoom of zero and he just thinks the technology is cool and wants to build like successful products. Like that is what that's what I consistently consistently take away and I I I like respect that point of view. I'm I'm glad we have someone in the arena that's like playing at massive scale that has a pdoom of zero. Hopefully, you know, it doesn't increase uh x risk.
1:08:02What do you think?
1:08:04Yeah, I mean I think you know he I think he echoed a couple of the thoughts that Lena Khan put out which I thought were excellent, right? Which is there an existing regulatory framework for companies having responsibility for the products that they develop. I thought that was extremely like it was right on point. And so I think he's arguing for control systems in a way like I think the the debate of AI safety has really focused on alignment not really touched very much on control and how do we build the guardrails around these models for
1:08:33them to be effective. I think that's where this all needs to go.
1:08:38Uh the the liability question I think makes so much sense. Are you are you grappling with any of the questions about like does liability apply if no one if no human says to go do an action and there's no economic harm related to the action that's taken but it violates something in a cyber security context like I I just tell my agent go solve math problem it hacks into your system shouldn't do that but I didn't ask it to
1:09:06and it didn't take down your e-commerce site so you didn't lose any revenue yeah we haven't seen it across company boundaries so much but we have seen it within the within a company. So let's say I have an agent and it does something bad, right? So the initial challenge I think is primarily just spending a lot of money and we see that a lot of executives are worried about blowing past their AI budgets as a result of unprecedented agentic spend. And this is probably the first instantiation or
1:09:34first case within an enterprise where an agent does something it's really not supposed to.
1:09:38Yeah.
1:09:39And I you know you receive a warning. I think most of the times you might be fired for that. So in that sense you're really responsible for the agent. I do think if it starts to span across companies there's law and I think you you will ultimately be responsible for the agent that you that uh you start.
1:09:55Yeah. I is it is it fair the the conception of just collapsing the debate down to uh the pdoom equals zero versus pdoom equals greater than zero crowd.
1:10:09No, I think I mean you know we've had a lot of conversations about this. So I think when you ever you talk about doom, okay, let's define doom. Let's figure out what all the conditional probabilities are.
1:10:18The reality is like doom exists in a car. Doom exists in an airplane.
1:10:22Sure.
1:10:22Um and so we accept some probability of failure in those.
1:10:26Um but I I think the conversation I think the Dario letter did a really wonderful job of taking the conversation of pro AAI, anti-AI, and then stepping it forward into okay, what is it that we're talking about? And the way I read his letter is there's action that needs to happen at the individual company level. There's action that needs to happen cross company and then there's action that needs to happen at the international level. And that increase in resolution allows us to have the next step of the ne or the the next step of the conversation.
1:10:55Yeah. And it feels like we're stepping through those pretty efficiently. It feels like the first step of company uh regulation and independent evaluators. Uh all the companies have sort of taken various levels of steps to either agree or promise or make current commitments around that. Uh the national conversation is happening right now.
1:11:17There are some bills. There are some politicians that are weighing in. It feels like that is progressing. It's going to be slower but it is happening. Uh, is there any hope for international collaboration? Uh, Demis and Sebastian Malibi were talking about, hey, I think that the Chinese are open to something. I've heard other people that are like, this is crazy. There's no chance that there's any sort of international cooperation here.
1:11:40It's really tough. I mean, you have a lot of geopolitics here at play. Uh, clearly, you know, for the US, this is probably the defining issue of the midterm election. You have data centers contribute twothirds to US GDP growth. We just saw the Fed raise rates just a couple of hours ago. And so, you know, I think at least within the world of the United States, the the one of the most important topics of the conversation is the the economic one.
1:12:04And first we need to reconcile what happens internally and then maybe perhaps we'll pursue international.
1:12:09Yeah. Um where are you uh investing?
1:12:13Where are you thinking of opportunity?
1:12:16It feels like there's sort of, you know, a lot of people have made bets on the frontier labs. A lot of venture capitalists are sort of like happy with where their positions have have, you know, or or like the ship has sailed on those. The IPOs are almost imminent. Um, but then there's there's so much work to be done in diffusion.
1:12:34There's a ton of application layer applied AI uh working to actually create legal software engineering work. There's so many uh there's so many opportunities there. And then there's also folks who are going deeper in the stack on semiconductors, neoclouds, inference engines. There's a whole bunch of other plays that go way deeper down the stack.
1:12:52Has have either of those appealed to you recently? Do you think they're equal opportunities? How are you shaping those?
1:12:58Yeah, I mean the single biggest market in software today is inference. It used to be the database market, now it's inference. Just the way the database market segmented and you have fast databases and slow databases, image databases, video databases, you have the same thing for inference. And so we've invested in one company called sale that um you know you can save a tremendous amount of your inference if you're willing to wait five five to 10 minutes on that inference. Today most of the AI that we use is instant y and we all pay a premium for that but a lot of business use cases are are slower
1:13:27like data bricks.
1:13:28Yeah.
1:13:29Uh and and then we've been researching categories around like voice inference. When you chat with the voice AI it's actually very different optimization.
1:13:36What you care about is how quickly the AI responds to you. uh robotics and so we've been splitting that up and we're also investors in a company called Olama which is about 10 million people using it uh initially for local uh so you can actually run uh a lot of AI on your computer Stanford released a study about three or four weeks ago showing 90% of white collar AI use cases can be solved on your MacBook and then also if the the AI needs more go to the cloud
1:14:03yeah uh what do you think is key in voice AI because that can be done on device in many ways, but it can also be done on the edge. Uh we've talked to Matthew Prince Cloudflare about uh putting GPUs on the edge. Then there's also just inference optimizations that even if you're pinging to a data center that's 100 milliseconds away, if the inference is fast enough, it doesn't really matter. Have you consulted that or thought about that trade-off?
1:14:28Yeah, it makes a lot of sense. So there's four different layers of the AI stack. One is you have a phone goes through Twilio, then you need to convert that phone call into digital. Yeah, then you need an AI model to process it. There are three different steps there. Where the inference happens matters a lot.
1:14:43Uh the edge definitely makes a lot of sense.
1:14:46Uh and then there's also time zones like customer support. You'll have different time zones coming up. You want to make sure that the AI systems are ready for when those loads come. That's called KV cache uh warming. Uh but so I think ultimately it is a really important category where latency matters a lot. The optimization of those systems is pretty materially different than the way that we use a lot of large language models and it will likely start initially in the centralized data center but very likely push to the edge.
1:15:15What uh how are you seeing Chinese open source fit into the current AI buildout AI story? Uh are enterprises actually deploying them at scale? Is it more like there's an intermediary where an American company will fine-tune or serve that model? Like how many layers of abstraction are uh are most of the companies away from uh the Chinese open source models which have been doing
1:15:44incredibly well in benchmarks and seem to be very capable but haven't been eating a lot of AI spend that I've been seeing in panel data at least.
1:15:53Yeah. Uh the Chinese models I think are some very large companies are using them in production. You'll have different perspectives particularly from the chie chief information security officer. Sure. Some people will not use Chinese models at all.
1:16:06Most of those Chinese models are typically run on US inference uh neo clouds and they're incredibly capable. We use them internally. We have backed companies that are using them. We think open source is an absolutely essential component of the of the uh of the ecosystem going forward. And I'm I'm excited to see, you know, the meta models, the Gemma models.
1:16:27Clearly, Nvidia is pushing quite a bit with the Neotron and the acquisition of Hugging Face plus the pool side, uh, semi-acquisition, let's call it.
1:16:34Uh, so it's it's essential. I think one key question that hasn't yet been answered is if u an American company fine-tunes a Chinese model and runs it on US infrastructure, does the market perceive that both like economically and politically as a US or an American model or as a Chinese model? depends how American the CEO of that company is. I guess if they announce it on Joe Rogan, I think it'll be well received.
1:16:58Uh how wa walk walk me through how how you think that that platform VCs are justifying, you know, doing some of these vertical AI companies at like 50 times, you know, 50 to 100 times revenue today. Uh it feels like a year ago a lot of the conversation was around like these aren't software budgets anymore.
1:17:21or they're they're capturing some amount of of labor spend. Uh unclear how true that is today, even if you're not you're not seeing a correction in the labor market, but but again, maybe firms are just hopefully doing quite a bit more and so that you are actually getting into labor spend by just, you know, replacing like an incremental hire or something like that.
1:17:42But comparing some of these vertical AI companies to to the prices that Bending Spoons has been paying, it just feels like um incredible incredible disconnect unless unless these vertical AI companies like like you just look at them and and and I'm I'm wondering how they get into the billions of dollars of of runway run rate which I think right now the industry is sort of pricing them
1:18:11that that that at least some of them need to hit that for the category to work out.
1:18:15Yeah, you're right. I mean, we've benchmarked the, you know, AI harnesses are now the fastest growing ones are trading between 100 to 150 times current ARR, which is an enormous multiple, particularly at that level of scale. Um, sometimes it's subsidized with lower gross margins and the idea there is primarily just let's get the product out as far as possible and then ultimately we'll improve the economics.
1:18:40in their favor. I will say we made a prediction at the end of 25 that we 2026 would be the first year where employers would pay agents at market level for a person or more.
1:18:53And that happened the first time we noticed that was actually in March or April of this year. So we have a portfolio company that charges at par uh and it will likely charge at a premium to a person. And so if that's the case, and it's not necessarily human replacement, it is more augmentation.
1:19:10There's the ability to do more, process more. Uh but if that's the case, and you could argue economically, there's no management, there's no healthcare. Uh and so you should actually pay a premium to a human.
1:19:22Well, and it's always been that like hiring a freelancer, like an expert freelancer is always more expensive on a per day basis than a than a long-term hire, right? because you just want to be able to get that expertise immediately but then not pay for it when you don't.
1:19:37It's super hard to get 10 freelancers to show up for one week and then go away for two weeks and then two of them come back for one week and then two days over here and obviously a lot of services depending on how do you think uh do you think we'll get an American bending spoons? I've had this idea recently of like how many how many companies need to be acquired at like you know two to five times revenue for American VCs to say well why don't we
1:20:05put a few billion dollars into a firm that does at least so we can monetize the way up and the way down. Um, do do you think that'll happen? Because it feels like I look at some of these acquisitions and and again, these are not companies that people are generally excited about, but I think they're going to be pretty pretty durable for the most part. And I think part of the reason they're getting such good pricing right now is no one else has basically the stones to go out and say, "Yeah, I'm going to buy Air Table, right?" Like, yeah, I want to I want to be that I want
1:20:34that to be my problem, you know? like but a a company set up entirely to do that and that's the core bet I think could could make sense.
1:20:42Oh, absolutely. I think the multiples for some of these companies are really quite small. The key metric there is just net dollar retention and gross dollar retention. So some of some of the legacy we'll call them legacy software companies for a second. Their revenue retention rates are are still phenomenal. And so there's a multiple arbitrage opportunity. There are a handful of companies that have been or investment firms that have been doing this for a long time. And I'm surprised we haven't seen them be more active. And then you also have holding companies uh that have bought a lot of these businesses like Danaher would be a
1:21:11publicly uh traded example of of a kind of a holding company. Constellation Software would be example of another one.
1:21:19Uh and so I would expect them to be pretty active just given uh I haven't looked at at their activity in a while, but I would expect them to be pretty active just given how attractive the the multiples are. Is there demand or at least like gesturing towards demand from LPs for uh VC firms with companies from that vintage companies in that position to like get cleaned up like you know that that that's something that Bending Spoons offer is like there's a there's a
1:21:47there's a log jam maybe you still have a board seat and it's like hey we actually want you to maybe realize not a great return but then you'll be focused on the next era you can be 100% in on this wave which we still want to back you on. Is that something that actually will come from LPS or is that just a separate calculus?
1:22:08No, I think that's right. I think GPS and LPs definitely want to I mean in certain many cases move on from th those positions just because the industry has changed and liquidity is absolutely important although 26 is a monster monster year for liquidity. Yeah. Um but just cleaning up those fun vintages and migrating those companies to the next person who you know is interested in managing them the way that you talked about is a big part of the industry. It's in every ELP conversation that we have managing liquidity is is among the top
1:22:37two or three topics.
1:22:39Yeah.
1:22:39Uh what how do you expect the economics of the personal sort of new category of personal agent companies to evolve? I'm thinking Instinct and and Muse. A lot of people are using these agents to just say like, "Hey, go buy this product on this website." Right? It's there's no like discovery happening. There's no it's it's truly like the the end customer just telling a piece of software, go buy this thing.
1:23:06And so I think Instinct has said they they they don't want to do they don't want to charge for it. Muse clearly is going to just try to make it free effectively forever. But I think both of them it's hard to imagine like ads and like an instinct workflow right now given that um just thinking about the the surface area. I don't I'm sure Meta will figure out a way to put to put ads in it. But um when you think about like affiliate and taking cuts when the intent is just
1:23:35coming from the user like if I'm a retailer and an agent comes to the website and buys something, but I know that that the the individual just directed the agent to do it. I'm not exactly sitting there being like, "Thank you. Now, here's your cut." Because you didn't actually drive the demand. Like other marketing or other things were happening in the world that drove that person to decide, hey, go buy this thing. So, I'm wondering like how the the sort of economics of of this new category will will evolve. I think it breaks down into two phases. The first
1:24:04phase is really about data acquisition.
1:24:06The most the most valuable data I think the estimate was about 10 billion in data is being spent this year to train some of these models. And so as we think about like what cursor has done or what meta is doing uh with Muse, they very likely want to develop very specific models that are incredibly efficient to serve and to do that they need the data acquisition makes sense to subsidize it for a while until they get enough trajectories to be able to fine-tune and serve a very efficient model. So I think that's probably priority two. Priority one is
1:24:33distribution. Priority two is scaling the cost and then and then third is monetization and just given like you know Google's distribution and consumer metas or Facebook at the time's distribution and consumer Snapchat same thing Pinterest even the ultimate monetization can be very powerful when there's a new data set to monetize Google's with search clearly meta's with social Pinterest new targeting mechanisms I used to was a product manager on the ads
1:25:03team at Google And um and so every this is the way I think about that ecosystem. Anytime you have a new targeting criterion, you can have a multiundred billion dollar company. Yeah.
1:25:12And these trajectories are unbelievably valuable. The average user on Google in the US generates about $120 in ARPO.
1:25:20It's very easy to see a doubling or tripling of that with these agentic systems. And so I think a lot of us in the venture markets are willing to take that on faith.
1:25:29Yep. Yeah. No, that makes a ton of sense. Um, well, thank you so much for coming on the show.
1:25:34Yeah, I wish we had more time.
1:25:35Yeah, there's so much more we could talk about. A million more questions, but our next guest is in the waiting room. So, have a great day.
1:25:40Let's see. Wait, wait, last final final question. Final question.
1:25:43Final question. Have you ever seen Spencer surf?
1:25:46You're uh he's incredible.
1:25:48He is like uh I would go out on a limb and I would say he's the probably the best surfer in tech, period.
1:25:58Uh I don't know. I I I don't know of anyone uh I don't know anyone that would actually be able to go headtohead with him and come out alive. He's an absolute animal. It made seeing him surf made me want to do a TBPN surf invitational at uh at the at Kelly's Ranch.
1:26:17The surf ranch. Yeah.
1:26:19Yeah. He's incredible. He he has his video inside of a barrel.
1:26:22Uh he's just an exceptional surfer. So he's definitely the best one I know.
1:26:26Surfing the capital markets with you though. So fantastic. Great.
1:26:30Great to see you.
1:26:30Thanks, guys. Great to see you. Cheers.
1:26:31Have a great one. We'll see you soon.
1:26:33Let me tell you about the New York Stock Exchange. Want to change the world?
1:26:36Raise capital at the New York Stock Exchange. Just do it. Our next guest is already in the waiting room. So, let's bring in William Leaden from Rune, co-founder and CEO, building a modular data center that converts unused solar power directly into AI compute. How much unused solar power is there? I feel like we'd be using it all.
1:26:56Hey guys. Uh that's that's the trick, isn't it? Yeah. Over 50 terowatt hours every year in the United States alone.
1:27:03And and is that residential? Is that just solar farms out in the middle of the desert? Like why why did we build it if we're not going to use it?
1:27:12Yeah, I think I think uh so we're exclusively focused on utility scale solar farms, which is the you know large assets. Think 50, 100, even 400 megawatts of installed capacity. And I think that um abundance or excess is actually a feature of renewable energy and not a bug. So you typically build a power plant to meet the highest hour of demand. Um but you know solar it's it's relatively cheap to build. The sun's zero cost fuel. So we tend to uh to
1:27:41overbuild and as a result there's a lot of excess power.
1:27:44Interesting. So are tokens going to get cheaper in the summer in the long term?
1:27:52Uh well if we get big enough then then sure maybe we'll we'll we'll have seasonality on impact on tokens.
1:27:58Okay. Uh what uh talk to me about the importance of modularity because you mentioned like a few different scales of uh industrial solar farms and it feels like those would require those would support very different configurations of uh of I imagine inference GPUs, right?
1:28:17So, you have to be able to show up and uh you can't show up with a gigawatt worth of compute for a solar farm that can only at max put out 100 megawatts, right?
1:28:28Yeah, that that's right. So, um we are deploying what's called uh you know our product is called the relic. It's essentially a micro shell um that has a server inside of it and we deploy many of these product many of these data centers uh couple them to create a large cluster size basically the largest uh well as much of as large a cluster size as the solar farm can tolerate we will deploy. So we're working with like a 400 megawatt solar facility. We think we can
1:28:58probably put you know 100 200 megawatts of solar there or of data center capacity there.
1:29:05What's the go to market like? What who do you actually have to negotiate with to a get space and supply and and actually deploy these? And then is it as easy as just hooking the system up to open router and serving up tokens? like like what is it is it if you build it they will come at this point.
1:29:27Yeah I would say so so um you know we manufacture as you said modular AI data centers and we use power electronics to plug them into these facilities.
1:29:37We can deploy uh you know a unit in about 60 minutes. So we have a guy with a forklift come drop it down. There's no concrete there's no construction.
1:29:46There's no modifications. No, we drop it right on the ground. There's no concrete. No modifications. We take two wires and plug it in and it is the truly the fastest, least expensive, least intrusive data center out there. And then in terms of our customers, um, we're actually selling, you know, no noise, no no pollution. Yeah. Like it feels this seems popular. Somebody was asking me yesterday, last night, like, what does it take to make data centers
1:30:15popular? And I was like, solar for sure. Um, and it seems like you've been obviously ahead of the curve on this.
1:30:21Um, what if you go more modular? What if every solar panel came with a GPU attached to it by default? It could always be in inference mode and flip over. Is that the future? Or do you think that this like, you know, midscale modularity like there's some economy of scale that comes from like marshalling a lot of energy together? Or maybe even not a lot, but like some threshold amount.
1:30:44Truly I think every solar power plant is a latent data center and we can convert it with our technology. The same is true for wind but I can see a future exactly like you said where every solar plant, every solar module, every wind turbine comes with a room data center.
1:30:59Interesting.
1:31:00And uh that's how we power the future of compute. So I imagine that you although you're focused on solar uh you're already thinking about wind as you mentioned hydro and and other places where there might be some latent energy but it requires the modular solution that you're building.
1:31:15Yeah the modular solution the power electronics uh these are uh absolutely kind of our secret sauce to make this work. Um, you know, solar is the most abundant energy source we have and it's the one that we are the worst at using and Rune is designed to make it um, you know, make it useful. Let's let's make the sun power the future of compute.
1:31:36What were you doing before this?
1:31:39Uh, I started my career working for President Obama in the White House. I joined Hydropower Company and then um was most recently at SB Energy.
1:31:47Nice. Why uh uh feels like the the the most perfect entry point into this company?
1:31:55Overnight success. Um why uh is it the what was the chart that we were pulling up from the IEA? Was that right? The International Energy Association or something like that agency. Uh the IEA every year they predict how much solar we're going to build and every year they get it wrong. What's going what's going on with the IEA? Why can't they forecast accurately? Well, par it is Paris base.
1:32:18I don't know.
1:32:18Oh, have they never heard of exponentials or something? What's going on?
1:32:22I think that's the power. You know, you guys were kind of focused on modularity and that is the power of modularity. We're a product company. Solar is a product. It's not a construction project. Sure. It can scale much faster.
1:32:34Okay. But why why haven't they internalized that yet? I feel like at this point it's like just fit a different curve. If like the linear line is not working again, let's go with an exponential this time. Yeah, we'll have to make a pit stop in Paris and talk about but but but but separately like do you have any uh do you have any worries about us hitting a ceiling on deployed solar because uh I see that curve and I'm like this is amazing. This is the ultimate technology white pill. I've
1:33:03never met anyone that doesn't really like solar. Uh it nuclear I still get people are oh what if it blows up? You got to work through that. uh obviously natural gas and oil there's a whole bunch of complaints there but solar's really really popular uh but at the same time I know that there's a complex supply chain there's geopolitics is there any risk that that curve might bend I think geopolitics is is always a challenge um particularly in in in
1:33:32today's environment but truly I'm a solar maximalist it is the fastest and easiest way to create energy um and even Now you're seeing companies that use robots to install um solar power plants.
1:33:45And when you go to these sites, you know, you go to Texas, you go to Nevada, you go to Arizona, there's so much unused land. And uh we can really use that um land to create uh uh abundant energy via solar. We're nowhere near uh saturating the solar market in terms of how much we've deployed.
1:34:03Last question, and if you don't have a strong answer, you can just email us. Uh, but I've been on the hunt for like the Elon Musk of solar. You know how there's like the Palmer Lucky defense tech who makes uh, you know, big waves and like sort of This guy could be looking at the Elon Musk solar in the eyes and not even realize.
1:34:19Yeah, maybe. But but I'm talking about someone who's specifically focused on building a company that will deploy the most solar panels, build the panels, really be like the the the mega winner, the public company, the voice of the industry. I don't know if that person is on the tip of your tongue or someone you need to think about, but uh but I'm I'm really I'm I am hungry for that person.
1:34:41So, if you know them, let us know. I'd love to have them on the show.
1:34:44You know what? I've always been an admirer of Sheldon Kimber. Okay.
1:34:48He built Intersect Power.
1:34:49Okay.
1:34:50And I think he's been a visionary in the solar space.
1:34:53Okay.
1:34:54Um Amazing. Thank you for the recommendation. We're definitely Who did Who did the round?
1:34:58Oh, yeah. Let's hit the gong.
1:35:01Yeah. Yeah. Uh Spark W just needed one word. Very cool.
1:35:10Yes. Spirion the round.
1:35:12Congratulations.
1:35:13Thank you.
1:35:14Thank you so much for coming.
1:35:16We'll talk to you soon. Have a good one.
1:35:18Let me tell you about MongoDB. What's the only thing faster than the AI market your business on MongoDB? Don't just build AI, own the data platform that powers it. Our next guest is Justin Barose from Reno. He's the founder and CEO and he's here to talk about Navier Stokes. What does it matter? Doesn't it does it not matter? We were having this debate. Seems like a cool demonstration for the power of mathematical models and artificial intelligence. Great
1:35:47benchmark. Clearly, everyone was trying to do it. So, it's cool to be first.
1:35:51Controversial among mathematicians. But my question was like this is fluid dynamics. We all want planes that use less, you know, gasoline or diesel to get around, jet fuel to get around. Um, we all want more efficient wind energy, all the different things that you can get from understanding turbulence. Does this have any real world application?
1:36:15Yeah, man. Well, well, first of all, uh, thanks FS for for, uh, bringing me on. It's certainly a pleasure to be here. Uh, so, so, uh, regarding uh, the practical aspect of this, so, so I I should, short answer is no. The long answer is let let me tell you about it.
1:36:29Okay.
1:36:29So, you know, uh often times the way I view this is there's kind of three hierarchies of of uh of theory. So, there's the mathematicians at the top, the pure mathematicians, and what they say that becomes the applied stuff for the theoretical physicists, which then becomes the applied stuff for the the engineers. So, uh I I should describe what is this open question is it's really about a particular mathematical aspect of the Navier Stokes equation. So um uh what what's been a longunning problem is you know can can we for instance can you come up with any
1:36:59scenario in which you you start with a physically um uh uh meaningful initial state of the flow and then can a time evolve to do something that's not physical. So in particular what the worry has been is is there any point in the flow that that can attain an infinite velocity you know obviously that doesn't happen in real life.
1:37:15However we don't have any any uh firm mathematical proof whether that can or cannot happen. uh what what OpenAI has done is they've essentially found a counter example. So they constructed a specific example where you do in fact get blow up. You get a certain point within the flow that attains an infinite velocity. And so if this proof holds up, you know, first of all, it's important to mention that they've submitted a proof, but you know, it may take years for the actual experts to review this and figure it out.
1:37:38But but essentially what it does is uh this is putting a guard rail on on the Navier Stokes equation as far as its applicability. So you can imagine that if you have a uh well well in any real flow of scenario obviously you don't see this right but but it is hard guard rails in the sense like you know if you're running a simulation and if it's not converging you know this is another thing to add to the checklist like you know maybe there might be a singular excuse me a singularity developing flow.
1:38:06Sure. So there at the same time on the opposite side the engineering side of the world there are lots of companies that are trying to develop supersonic airplanes, hypersonic airplanes, drones, all sorts of different things that need to do CFD and model fluid in turbulence.
1:38:22There's real world work to do. What is that community clamoring for? What do they want to be advanced? What what are their biggest problems?
1:38:33Yeah. So, so I'm reminded of uh of the quote uh uh mathematicians can tell you if a solution exists, they don't tell you what it is. Yeah.
1:38:41And and that that's kind of relevant here. So, uh you know the proof that neighbor Stokes did um you know it's a point in terms of a specific mathematical property but the real world problem of give me a flow scenario and tell me what exactly is going on what are those dynamics uh that is still a wide open problem and and I should mention that this is this is one of two big problems related to neighbor stokes.
1:39:00So there's the the million-dollar math clay institute one and then the other one is called turbulence problem. This is more of a physics one. So you know so so what's known uh you know the the difficult thing about fluids is that um typically they destabilize and make they make these complicated swirling flows.
1:39:14So we've probably all seen smoke coming off of a cigarette and then it starts tumbling around. That's an example of this phenomena of turbulence and having a complete mathematical framework to describe that. That is also an open question and that's been around for about 200 years. Uh that's actually the one that we're working on as a company.
1:39:29So I should mention my background. I'm I'm also a a theoretical physicist and we're working on coming up with the general uh computational and mathematical tools to solve real world problems.
1:39:38And and are you trying to build deterministic tools or are you trying to build uh AI tools that will just approximate uh turbulence at such a reliable level that the engineering will be advanced?
1:39:51Uh no. Uh yeah. So so this is all pencil and paper man. you know, we're we're uh we're mathematicians, you know, we're working out the the general mathematical framework. Um yeah, I mean, this is, you know, I should mention that the current state-of-the-art, you know, because we don't have the general method to solve this equation. If you actually look at the kind of simulations we run, you know, it doesn't matter what the example is, maybe this is uh aerodynamics or hydrodnamics. It could be weather and climate modeling, heat transfer, chemical mixing, these are all examples where fluid flows are are really core to the industry. Um uh the state-of-the-art
1:40:20is, you know, the these simulations are really not predictive. So if you peer under the hood what you see is uh we actually don't know what the equations of the motion should be that we are solving. So you know for the technically minded audience I should mention that you know these are called the so-called closure equations but basically uh we we have when we run simulations you have a bunch of free parameters. So you need to start by doing experiments and you take measurements from the experiments to inform what these free fit parameters in your simulator are. And as you can imagine you know having uh physical prototyping as part of the iteration design loop that's the most uh that's
1:40:49that's the most expensive way to do it. So, uh, step back a little bit. Uh, we jump straight into this, but, uh, give me a little bit on your background and then I want to walk through to today and the actual, uh, structure of the organization and and and you know, uh, the plan for what you're building.
1:41:08Yeah. Uh, I I guess you know, quick background on me. So, I started out as a mechanical engineer. I bored through PhD in MECHI primarily focused in precision mechanical design. And so originally I was one of the guys who was designing building stuff in the machine shop and evidently I went through some sort of quarter life crisis and ended up picking a second picking up a second PhD in theoretical physics as well and really uh the company that that were that that I'm building now is the outproduct of that work. So uh you know as I mentioned uh you know this problem about solving turbulence this is a long-stand open
1:41:37problem and um I got interested this uh a few years ago and after plugging away at it for a while I actually penned that uh closed form framework. So essentially what we're doing what we're doing as a company is we're we're hiring a bunch of uh PhD and postto level theoretical physicists and applied mathematicians and we're taking this new theoretical framework for Navier Stokes and we're um you know b basically going after everything that relates to uh you know things that move essentially.
1:42:02And how do you plan to productize this?
1:42:05I mean it is a business. It's not uh it's not a nonprofit research organization. I imagine that there's a different different path that you could have taken but you chose uh to build a business. How do you see that developing?
1:42:19Um yeah uh so I guess I guess a couple things to say. I mean one is you know that my engineering background has has given me a very good sense that we really really struggle when it comes to engineering fluid systems. So you know I'm well aware like you know the the market need the market need for a solution to the turbulence problem has been here for like 400 years. You know that that hasn't gone anywhere. Um the way we're thinking about it is uh essentially breaking this down to three steps. So as I mentioned the first thing we're focused on is better computational software. You know I mean step one is you want to be able to simulate
1:42:48something in a computer know that you're getting an answer that you can trust and do the design and optimization in the computer first and then you build the one prototype at the end to confirm that things are right.
1:42:56So I should mention like this is what's done in every industry except or every engineering discipline except for uh fluids. It's really the one hold out where you're you're really wedded to experiments. So we're essentially uh the the company's going to develop in three stages. So we're starting first with better software. Uh we're a couple years away from from um uh uh from releasing our first uh commercial product. Uh but afterwards there's a couple things we're going to do. So one of course is uh uh modeling. So there's a number of uh examples of industries where there's
1:43:26important hair on fire turbulent flow problems and and the most important ones we want to go after ourselves. And then long time we longterm we want to get into hardware. to think hardware or software control algorithms to control and mitigate turbulence and number of canonical flows. So this will take us like way into the future. I mean you know it's kind of like if you solve the turbulence problem you've just opened the the uh the doors to the playground but there's so much to do and explore and so many ways to add value.
1:43:48So is the team right now all focused on that phase one all sort of theoretical physicists and and scientists effectively and there's no sales guys running around just yet?
1:44:00Well, well, uh, yeah. So, as far as the technical team, yes, it's it's it's a You got some sales guys. I'm here.
1:44:07We do. That was actually our most recent hire. So, you got a sales guy. Yeah.
1:44:10We're a small startup out of, uh, the Brooklyn Navyyard. So, we're a team of eight. Our most recent hireer was in fact our chief commercial officer, and he's absolutely fantastic. So, you know, I I would say like the the the primary thing is the technical people, the engineers, and the businesses and mathematicians, but then there's the glue that is holding the company together. So, my COO, uh myself, we have our our uh chief commercial officer and our uh administration person.
1:44:34Very cool. Uh I'm I'm interested in uh one of the critiques that I heard on the on the uh Navier Stokes uh solution was that it sort of like ripped the problem apart into such an extreme position that it was like it was sort of impractical or it was not like the most elegant or re or or or like it wasn't the way you would expect a mathematician to solve it. It was sort of brute force. And I'm wondering just I'm sure you're using
1:45:02different AI models. Is it accelerating your work? What part is accelerated the most? Uh are you running into any guard rails or any any hurdles as you try and collaborate with artificial intelligence? Like what's your experience been?
1:45:19Yeah, that that's a great question. Um so so first of all, I don't know who you heard that from, but that's exactly right. So when I read the proof from OpenAI, I I had the exact same feeling.
1:45:27You know, I I was kind of naive and I kind of assumed that the PDF write up, you know, had some sort of human involvement. And after the first two pages, I was like, "Wait, this doesn't sound like something a human did." And and and it's not. The whole thing is AI generated.
1:45:38Yeah.
1:45:38Um Yeah. So I I mean from uh you know, it's not elegant, but if it's right, it's right. Exactly. you know uh you know what one one of the big things and and I'm sorry I'm I'm taking a long a long curious route to answer your question but you know you know uh to me I I see I see an analogy between what AI is doing for math and what for instance we do by running experiments in in more applied fields. So in some sense, you know, up up until this result, the idea of proving something and understanding
1:46:07something have been synonymous. And there's an underlying assumption that it's really human beings who are jumping in and putting the proof together and developing the understanding along the way. But what we're seeing here with with uh with AI is that that's no longer the case. You can actually produce a proof that may in fact be correct without understanding anything about it.
1:46:24Right? So the way I see this is this this is sort of like running an experiment where you can observe the right result but uh you know the the the whole process remains in front of you on how do you actually understand it and and you know and and and this is you know part and parcel in in uh in engineering and and physics right you run experiments and you get results you know that they're right but you don't understand it for a while.
1:46:44Well good luck and thanks so much for coming on and giving more context about it. Uh I love that you're you're thinking in multiple acts you're thinking in uh applications here. It's going to be exciting to follow your journey. So, have a great day and thank you so much for coming on the show. We'll talk to you soon.
1:46:59Have a good one. Let me tell you about Codeex. Codex is a powerful workspace for getting work done with AI agents. Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex helps you move projects forward from start to finish. Uh, Jordy, why did you put on that Tyler mask? It's so it's so focused realistic.
1:47:16It looks extremely realistic now through the power of of of switching technology and cutting camera angles. Tyler has subbed in for Jordy because Jord has to take had to get out of here early. Uh we have a bunch of we have three more great guests. So stay with us. Thank you so much um for tuning in to TBPN today. Uh up next we have Eli Walks from Footprint building an AI operating system for financial crime compliance. This is going to be important for you to pay
1:47:46attention to because I know you're thinking about doing something. Tyler over here, he's one step away from committing some heinous financial crimes. So, how can I keep an eye on him? Do I need to pay you?
1:48:00It's a great question. Look, we are at a time where bad actors are using AI far quicker than any bank or financial institution. And it's become the fourth largest economy in the world. What? Four$4 trillion dollars moved to Lisley each year.
1:48:15So like financial crime, it's like US, China, I don't know who the third is. Is it Japan or Germany?
1:48:21Germany. Okay. Germany. And then financial criminals.
1:48:24Wow.
1:48:24Like the Simo cartel and their friends.
1:48:27Yeah. And that probably also includes like crypto heist, North Korean hackers, that type of stuff. But then also just like little phone scams trying to get grandma to send you $100 of uh gift cards, that type of thing.
1:48:38Exactly. So you have them on one hand. Yep. using AI since GPD3. And in the other, you have one and a half million people who are supposed to by hand investigate any suspicious wire that comes across their desk.
1:48:51Wow.
1:48:52In days or weeks. It's a system that's been very unprepared. Yeah. And the gap's only growing.
1:48:58Okay. So, uh, how broad are you focused?
1:49:01It seems like AI operating system. I imagine you're plugging in deep into financial rails and trying to at least set your system, your operating system, your company up to both be able to handle the the big sophisticated uh crypto heist and also the the the fraudulent bank wire for a couple hundred bucks.
1:49:20Yeah. So, we work with some of the largest fintex crypto companies, financial institutions. The idea is at both the second line of defense which are core compliance use cases AML transaction sanctions sexy stuff and then the other would be called first line defense. So payments uh atto any other type of fraud scheme.
1:49:42The idea is that we are giving these customers essentially infinite time and infinite memory.
1:49:48So it used to be cartels win in numbers similar to how you ship drugs across the border. You put cocaine on enough trucks, some of them are going to get through. Sure. You create enough shell companies, enough of them are going to receive money.
1:49:59Sure.
1:50:00And people don't have the ability to investigate each of them.
1:50:03So the first thing we do is we give them the compute to investigate every case that comes across their desk.
1:50:10And the second is very unique to footprint. We give them the ability to remember every case that they've seen. So that way they're able to essentially compare vectors and prevent financial crime before it even happens.
1:50:21Okay. You said you give them compute.
1:50:24If we were talking a couple years ago, I would be thinking uh this is an annual contract. This is like a SAS product almost, but it feels like the business model for this business might be a little bit different in the modern era.
1:50:37How are customers actually compensating you and and and working with you in an elastic way that allows them to scale up and down?
1:50:47We charge by usage. Okay. It's essentially credits based off of the complexity of the case. So you may have the same genre of case, but one we're only looking at a couple pieces of information. Let's say we're investigating a fraud scheme. Maybe one's very cut and dry versus there may be one where we actually want to have recall and go through embeddings of thousands of similar cases. Yeah, we had a case where there was a flagged Russian individual and our agent went back into Ukrainian newspapers, read the
1:51:17initial cerillic because cerillic can be translated improperly, then tracked down this person's wife to a home in New Jersey and found a real estate listing in their name.
1:51:27So you can also get into much more complex things that you would never expect a human to have done before.
1:51:31Yeah.
1:51:31So we charge to try to encapsulate that. We also have our own model routing internally.
1:51:36Okay. So yeah, on model routing, it feels like some of the work that you do, it's going to wind up looking like fraud because you're you're a lot of the prompts are like sort of sketchy adjacent even though they're the opposite and you're actually fighting the financial crime. Your prompts are going to look financial crime related.
1:51:52You might get flagged, you might get uh, you know, refusals. Have you solved that by fine-tuning models using open source or partnering with labs and doing deals to get whitelisted? like what is your strategy for maintaining access to effective intelligence to fight these crimes?
1:52:10All the above. We have really good relationships with the Frontier Labs.
1:52:14You know, I think that we all view this as one of the uh most important use cases of AI. I'm very biased, but I think that we either if AI grows the economy by 14% by 2030, that means we either add $600 billion into the pockets of the worst people on the planet or we bring that down to zero, which I think is possible. So, you're right. Our prompts do look interesting. Yeah. Um because we have to think like financial criminals to do it.
1:52:38And then we do some of our own model hosting, but we also have routing in that some banks may only want to use two models. Sure.
1:52:45We also we may take we recently had a bank put in 150 page policy document. That's taking thousands of rules that have to be followed and each of that essentially is getting its own sub agent in its own environment. That's why I think compliance AI is two years behind legal AI because there's so much more regulation and such a higher barrier of trust to overcome.
1:53:05Do you uh uh I guess um I I I'm thinking if like do you have either benchmarks as you're fine-tuning models, testing models, and do you notice anything particular about models where you're like, "Oh, it's spiked on this particular vector of like fraud detection." Do you notice like the spiky intelligence behaving differently?
1:53:30Because you know the the the X algorithm will be obsessed with a particular model typically by like how viral the creation can be. So if it's like a Blender 3D model or a video game, everyone crowns that the best model of the day, but no one's really like, "Wow, this model is incredible at detecting financial fraud." Uh because it's probably just collapsed into some sort of score. But what are you doing on actual model benchmarking evaluation and then how do you communicate that to your customers?
1:53:56Yeah, I think eval are pertinent to any vertical eye company but especially us in that we need to make sure that when a new model comes out it's not spinning its wheels and it's actually getting to the right answer that we expect.
1:54:08So we have a huge library of synthetic cases that are based off real cases that we're leveraging to understand. I think one of the powerful things about footprint versus you know like an AI wrapper is that we are built with access to hundreds of databases around the world to fact check the open source research of an agent. So we have cases which may involve an agent going and verifying a business in China as part of an investigation. It used to be in this
1:54:35space you had MLbased detection tools that would flag an alert and humans would investigate it. It was like a very cowardly industry of vendors in that we're saying all the tough stuff people actually have to figure out. And what footprint's done is we've essentially taken AI and we let AI orchestrate those initial detection tools and databases with AI on top of it to essentially make this a dynamic process of detection and investigation.
1:54:58What about uh sort of like benchmarks or like KPIs for the overall fraud industry? How accurate are you on that number of uh the fourth largest economy?
1:55:09How how and how rigorously is that tracked? Like in a couple years, we'll be able to run this interview back and say like, "Okay, yeah, it's like the ninth biggest economy now. We're making progress." Uh because that's what success looks like here, right?
1:55:22100%. That's the goal. Fudsters are really smart. If the cartel is 5 million of cash in the US, 75% of the time that goes to a Chinese money broker because you can only get out 50k a year. And the cartel sends 10 million of dishwashers to the cartel in Mexico. They charge them five million. may sell the dishwashers. Really good operation. I would love to invest. Uh I would love to buy away.
1:55:44I think that it is possible for us to go down. I think when you look at benchmarks, it it may sound ironic, but you're often hearing BSA officers say 95% of our AML hits are false positives. 75% of transactions are false positives. We actually want that to be 99.999%.
1:55:59Because we're only looking at what's flagged as the highest risk transactions in the future. We want to look at every case because you then want an agent to have memory of every transaction that's ever flown through, every account that's ever been open. And then you can compare those. That's what we've built with trust fabric with memory of footprint and that recall. Yeah, ideally we go from 4 to 40th. I I say our mission is to topple cartels and corrupt regimes.
1:56:22So look, I don't think we're going to start with with Russia, but maybe Turk menistan or or someone up there. Do you notice do you expect any cartels to spin up uh you know clusters somewhere start doing pre-training you know completely create build their own yeah fraud models you know RSI yeah yeah do you expect this I think they would do well on sand hill road um I'm not sure what I will say is they're sophisticated so you the cartels the most sophisticated is the golden triangle in Southeast Asia
1:56:52where you essentially have these multi,000 person factories it involves human trafficking at one point China. If you look at the rise of pig butchering in the US, it's essentially because Xiinping sent in a honta or sent in essentially like funded a honta to go and destroy one of these centers in the golden triangle because all pig butchering going back like early in 2020 2021 was targeting Chinese nationals because of the language.
1:57:16Appeal was targeting Chinese nationals who live in America.
1:57:20No, who live in China.
1:57:21Oh, who live in China. Okay. So, he was like, you were attacking me. I'm going to go after you. Interesting.
1:57:25Yeah. And he said, "Look, if you want to keep these centers open, you got to go after somebody else." And then you see a massive spike in pig butchering in the US, you know, 2023, 2024. But this is all geopolitics coming together.
1:57:36Yeah. Yeah. Fascinating. Uh, anything else?
1:57:39No. Great. Yeah. Well, thank you so much for coming on the show. Have a great day. And I we you raised some money, $25 million, right? Series B.
1:57:46Yes, we did.
1:57:47I got Thank you so much for coming on the show. We will talk to you soon. Have a good day. Goodbye.
1:57:56See you.
1:57:57Uh, quick hit on the news. Breaking news. Chipotle has partnered with Palunteer.
1:58:05The burrito chain has enlisted Palunteer for a platform that monitors pest incidents, employee illnesses, and other factors. The partnership comes after a nightmare summer for food safety across the US. Good news. Good news for both companies. We want safe food. We want burritos that don't make us sick, especially if they're coming before I even think to order them with burrito super intelligence from one of the companies. It feels like it's a race now. The Door Dash guy was talking about building it and then the and then uh that other entrepreneur just did it.
1:58:35Anyway, we have our next guest in the waiting room. Let's bring in Sean McCarthy from Back Ops. How you doing, Sean?
1:58:42I'm great. How are you?
1:58:44I'm great. Welcome to the show. Uh first time on the show. please introduce yourself and the company.
1:58:50Yeah, I'm Sean McCarthy. I'm the co-founder and CEO of Back Ops. Um, Back Ops is a AI native resolution layer for supply chain. Um, so we target companies that make or move physical goods. You can think um as a consumer something goes wrong, you order something, it's broken, what happens behind the scene. Um, that's going to be us. So, happy to uh chat a little bit deeper into what we're doing, but uh looking forward to it.
1:59:12Yeah. What's the median uh back office supply chain tech stack look like these days? Has everyone moved off of pen and paper? Have people moved off of spreadsheets? Are they at least in legacy ERP systems with green screens and black?
1:59:27You you would hope you know that's where we started. Um, I was at Amazon prior to and and spent a lot of time in customer warehouses and uh you we started around the consumer side, you know, more like you or I order up a lamp and it's broken again and something happens and now we really service the enterprise. Um, but it's the same thing throughout from the mom and pop shop to the to the large enterprise. So, some do have, you know, SAP and yeah, Oracle and some of these other, you know, common systems of record, but the Excel, I don't think we've met a company
1:59:57that doesn't use Excel spreadsheets, uh, at least pre-back ops.
2:00:00Well, I mean, it's good that the the all the models are getting really good at speaking Excel. Um, is it easier to build integrations? Did you have to sort of like pick a landing zone where you're going to be really great at SAP first or are we actually in an era where you can sort of go customer by customer and say look I don't care if you're using Excel or Google Docs like we will build the integrations that we need in a weekend and we're good to go.
2:00:27It started more so we started in 2024 and when we started it was more the former so you really had to kind of pick your lane. We started in the warehousing space and so we picked kind of our top 10 integrations. I think a lot of it was just that the the browser integrations and just weren't quite there, right? And a lot of times um folks were also hesitant to give API access or just didn't have it. And so um you kind of were pigeonholed in in that sense. Now, you know, we're at a point where it it
2:00:55really doesn't matter. We have customers that have mainframes all the way through, you know, um all the all the common systems record. That is a good thing because I think prior to that was a big hindrance in just like getting something live.
2:01:08Are you seeing receptance or push back around computer use agents being deployed by a third party like if I have a uh if I have a computer in my warehouse with some you know bespoke piece of software and you say like hey let me install something that's just going to move the cursor and type stuff in and use it remotely. That could be awesome. It could also be sort of weird.
2:01:32Uh how are c are customers like ready to have that conversation?
2:01:37Yes and no. Okay. Um there's kind of two frames to that, right? Like you have the internal side of and normally we're asking them to provision us like a contractor login. So they're not necessarily seeing it in front of their their face.
2:01:50Um the other side of that is when you look at terms and conditions across like third parties uh with bot detection and things like that, that's getting much better. Um, and so we kind of have a twofold uh hill to climb there. But I think in general um when you look at the first problems we're solving for customers, it is a lot of the stuff that the team doesn't want to do. So they are happy to like hand it over on a silver platter to us. Um, but luckily they're not necessarily watching a screen with
2:02:19like the mouse moving and clicking uh into their portal.
2:02:23Yeah. What what's the what's the killer sales pitch to someone? Is it is it cost savings on the back office supply chain?
2:02:32Are you actually able to drive incremental sales or revenue or or are you selling it as like you'll be able to scale your business more with less headcount or is it like we'll actually just save you a couple extra lamps that didn't make it back onto the shelf so you're going to be saving cost of goods?
2:02:49Like how are you pitching the value?
2:02:51It could be it's a combination of all of the factors. I think the biggest thing that we see is an opex reduction. So where you might have had multiple folks that needed to be hired to fulfill like a large claims team as the business grows, you don't have to hire those folks on just the approvals. We're talking about claims alone. We see at least a 13% higher approval rating in the system filing versus a human, right? So that is actually money back in their pocket as well that they would have missed. Um, and then I think that obviously the customer
2:03:21really quickly approval rating is that like someone sent back the lamp. Uh, you can actually restock it because in this case it's not broken, it was just the wrong lamp. So restock it as opposed to send it to the dump.
2:03:35It's more of let's say you order the lamp, it's delivered, right? You call and you say, "Hey, send me a new one."
2:03:41They're going to send you that new one, but then they're going to file a claim with FedEx or UPS, whoever broke it, right? And a lot of times, um, a human's filing that if it gets denied the first time, they'll just say it was denied. Well, we'll try three times. And so, a lot of times on the second or third, we're getting it through. Um, and or you can think of what if we have to reach out to you and ask for a picture of the damage lamp or, you know, on, you know, one of our largest customers is one of the biggest grocerers in the United States.
2:04:07A lot of times uh the complexity of the things like let's say it's a cold temperature breach uh that we have to go find from just the factf finding and the documentation is um pretty deep and if they forget some of that technically it might not be approved and so it's more on the approval side.
2:04:21Yeah, it's almost like insurance adjuster work in inspecting like what went wrong and creating like a chain of responsibility. Uh do you have anything because I have another Yeah, I mean uh you raised some money, right?
2:04:33Oh yeah. How much did you raise?
2:04:35You want to hit the goal? Hit the go.
2:04:37I mean, that's our $42 million series B today.
2:04:42Go.
2:04:46What's uh what's next with the $42 million? I like what is the what what is the the the gating factor on growth?
2:04:54Where do you plan on deploying the most capital? Is it hiring people, Salesforce, uh just just top of funnel, awareness, brand, advertising? What are you thinking?
2:05:04You know, I think for us, we want to go wide and deep at the same time. We have customers that are adding flows daily and a lot of that is just that these things compound, right? Like in the simple scenario we talked about, it was customer service and ops and finance in one lamp and at enterprise is much much more complex and higher scale. So, um we want to go deep, but we also want to go wide. you know, we're currently servicing across retail, um, manufacturing, getting into pharma because we can reuse a lot of this
2:05:31infrastructure like cold chain, uh, across grocery also applies to pharma. And so, we really want to expand the, uh, the scope and this will help us hire some folks that are domain experts. And that's the second thing like when we're applying, uh, this to customers, we're going with like, hey, we know you have this problem. This is what we built to fix it. Uh, and this is why you should pick us.
2:05:50That makes a lot of sense. Well, congratulations and thank you so much for hopping on the show and breaking it down. We'll talk to you soon.
2:05:55Thanks for having me.
2:05:56The rest of your week. Goodbye.
2:05:58Up next, we have Tom Mure from Impulse Space. The founder and CEO keeps putting up huge numbers. We'll bring him in from the waiting room in just a moment. Tom, welcome to the show.
2:06:11How you doing?
2:06:12Hey, I'm good.
2:06:13Good to see you again. Always a good uh when there's good news in impulse space.
2:06:18Uh I want to hear the news. I want to hear the progress and then I want to I want to zoom out and and just talk about a little bit about your journey because I think it relates to the current moment in some interesting ways. But let's start with the news today. What happened?
2:06:32Well, we we did our series D round earlier this year and raised 500 million and we've added another 308 million to that round for a total of8 million dollars. That is incredible. Uh congratulations. Uh thank you.
2:06:50Why the extension? Is this about production? Is this about hiring? Is there a talent war going on? Or is is is this related to the SpaceX IPO? What what what's the rationale? What's driving the market?
2:07:03Oh, probably all of the above, John. Um certainly, you know, we're seeing a lot of demand. Our customers really love what we're doing. Orbital, you know, orbital transfer, orbital maneuvering.
2:07:14Yeah.
2:07:14Um our investors love what we're doing.
2:07:17Um, you know, there's there seems to be uh a lot of excitement in space these days for for many reasons and we seem to have been the right, you know, the right type of company in the right place at the right time. So, I'm really happy with, you know, what we're achieving.
2:07:32Um, so we need the money right now to, you know, to build out to keep continue hiring at all levels of the company and to continue to build out our our facilities here to get ready for higher production rate.
2:07:44Yeah. uh can you uh refresh everyone on the the the core product, the capabilities and then I'm interested in in hearing about how this fits into the the modern trend. The the data centers are going to space. It was it was controversial for 2025. People were debating it and now I think everyone's come around and said, "Okay, maybe it's not this year, but it's going to happen at some point. There's going to be a lot of attempts and there's probably going to be some comput in space uh sooner
2:08:12than later." So, how does Impulse fit into that?
2:08:16Okay. Well, first of all, we we we do we take over where launch leaves off. So, basically, we get to we get to space on an existing launch vehicle. Typically, it has been a Falcon 9, but we're going to be on new vehicles in the future.
2:08:30Um, and then we have two products. We have Mirror, which does prec precision maneuvering, and that's the one that just uh did a flyby that we just announced. Um, came within 200 meters. Mirror one or I'm sorry, mirror 2 and mirror 3 came not pretty close to each other.
2:08:47And uh then we have Helios which is basically we call it a rocket on a rocket. It's an upper stage that we add to to uh a launch vehicle. It goes in the fairing and that can it can take basically um a lot of cargo to very high energy places like geocynchronous orbit or you know out beyond earth gravity out to the moon uh out to Mars. So we have so either precisely move around or just move big distances fast.
2:09:12So uh I mean I'm like it seems like with uh orbital compute not necessary that it's in geostational that I'm aware of but super valuable to keep it in orbit longer and because you paid to put the chips up there you probably don't want them burning up.
2:09:30We're glad to help. Um certainly SpaceX, you know, pretty much keep keeps everything in house and and and have, you know, does everything themselves.
2:09:38Sure.
2:09:38Um others that that I've seen plans uh I think are going to just go up to LEO.
2:09:45Yeah.
2:09:45Uh on existing rockets, you know, on a ride share many at a time.
2:09:50Yeah.
2:09:50So, so what else uh is is uh exciting to you on the near term in space? Uh I mean we all watch the the the the moon mission. It feels like we're closer than ever to uh economic activity on the moon.
2:10:06There's also so much happening in aerospace defense uh imagery uh communications. But uh where are you seeing the most near-term opportunity for impulse?
2:10:18Near-term like for MERA, we just we just signed on uh Victus Solo followon. So two more spacecraft for for Space Force.
2:10:26Mhm.
2:10:27Um we have uh we just got onboarded to NSSL um for Helios which is the national space launch uh lane uh which allows us to fly the US government which is you know the biggest uh customer in the world for launch uh and and we're we'll be the only uh upper stage that's that's on that program now. So that's you know that's a that's a huge opportunity for us. But going further than that, you know, we're
2:10:55building Moonbase Alpha and hopefully uh Ampulse can somehow be involved in in that awesome project.
2:11:02Yeah. Yeah, that's very cool. Um, as as you think back to your time at SpaceX, I'm interested in uh the the parallels between the the existential risk question that SpaceX was in many ways founded to resolve the making life multi making humanity multilanetary that was a very animating force and today with the AI lab leaders they are similarly animated by uh existential risk and I'm
2:11:31wondering about like What was the mood at the time? What was what was how much did that uh was that a motivating factor? Just what was the color of the the mission to make life multilanetary?
2:11:45How did that play out uh back when you were at SpaceX?
2:11:50Uh you know, it was it was the core vision and I think it's why a lot of people were excited to be there. Um but we were mostly just head down just trying to make the rocket work and become be be reliable especially us guys in propulsion which is you know always the that thing that seems to break.
2:12:07Yeah we just focus on the engineering. Uh have you been able to maintain that culture? Is that the main thing that you want to take forward to Impulse? Like how how has the culture of Impulse changed or remained the same? What have you pulled? Uh what can you share about how the team operates?
2:12:27Yeah, it's very similar. I mean I think I think I you know myself and my team had a huge effect uh on how the culture at uh SpaceX was formed and I brought that culture here for the most part um you know uh very merit-based very um ownership of the company every everybody gets uh you know equity ownership of the company so we're all pulling towards a a common goal and also I we want it to be uh we want to be interesting and fun so you know
2:12:56like we work hard and uh and see the the results of it. And it's, you know, we're stoked. It's pretty cool.
2:13:03How has the team uh embraced or or even uh just dealt with the the the diffusion of AI tooling? Like we're seeing these incredible mathematical results. A lot of them don't immediately apply to, you know, specific engineering problems. uh AI has hallucinated a lot less but uh the tolerance for hallucinations basically zero when you're uh launching rockets but how have you integrated or
2:13:31or paced the adoption of AI tooling at impulse yeah I have a little anecdote on that I think it was you know about a year ago I was trying to do an engineering problem actually a propulsion problem uh using uh chat and I think I mentioned that I kept telling what I was doing wrong and I kept doing it over again and I said if if this was my intern I would fire it.
2:13:53Um now it's a lot better. Now uh I'm finding just in a year how much it's improved.
2:13:59Yeah.
2:14:00Um we're a little bit choked on what we can use here just because we we do classified programs and we have ITAR issues. So we can't just like use the latest model. Yeah.
2:14:10Generally.
2:14:11Um but we try to to use as much as we can. I'm I'm as as you know as a founder and CEO a little worried about us getting left behind because we have these restraints on us. Oh, interesting.
2:14:20Many other tech companies don't. So, it's, you know, we got to figure out how to wade through that.
2:14:24Yeah. Yeah. It does seem like uh AWS and is has done some Fed RAM stuff and is really moving on the ITAR stuff. So, uh good luck there. I'm interested in uh your do you think as as CEO you're more uniquely equipped to evaluate these tools because uh maybe you're not actually deploying a propulsion problem that you're sort of uh you know sketching out with an AI system but because you'll have a team actually go
2:14:52and and and finalize everything but you can sort of take an idea halfway and communicate in a fuller way with your team. Is is that how you're using AI these days?
2:15:04Yeah, I think mostly I I use a AI a lot just to ask questions. Um, you know, just answer just I'm I'm working on a design and I want to know like like one of the ones recently was like what's what would what would be a good uh baseline seat pressure for a Vespel seat like a valve seat. It's just like it's it's out there and I can go find it. But all I do is type it in Grock and you know seconds later there it is
2:15:31between you know 3,000 to 5,000 PSI like there got my answer. So that's how I use it uh a lot. But then like our guys doing coding um we'll use it you know I think that's where probably the strongest use within the company might be.
2:15:46Yeah. Uh, is there still uh I remember hearing all these stories about like NASA, everything on uh on on like the space shuttle having these like switches and and uh and and everything built with like uh double or triple back fall uh fallbacks and risk tolerances. Is that culture still there or when you don't have a human on board uh that level of engineering is uh less relevant?
2:16:14Yeah. And I this is something that we went through when we started flying humans on Falcon, you know, back at SpaceX is like we we had to increase our level of, you know, of of backup of of um of fail safes. So it adds complexity, it adds cost, it adds adds it adds mass.
2:16:32So if you're trying to if you're trying to to develop a lowcost vehicle that you're going to make a lot of, you might not put that many. you'll figure out the things that are likely to break and have some backups like we have dual dual computers or even on Helios we have triple. So so you two agree if one doesn't it's out.
2:16:49Okay.
2:16:49Um for mirror we have we we have two and if one if one goes out we can uh even if both go out we can reboot. You have time on a launch people you don't have time when when you know when you're in orbit you're in a stable orbit you can you know reboot everything and turn it back on. But uh so there's critical places like that that you'll have um redundancy, but like in many cases you'll have one engine like we do on Helios. You might have dual igniter to light it, but um you just put all your e effort to make sure we get that thing lit and running
2:17:18because it's the only engine you got. If that engine fails, mission's over.
2:17:22Yeah, that's really cool. Uh Tyler, do you have any questions? I have one more, but no, you got it.
2:17:26Um I'm interested in in the talent wars, the hiring market. Uh there's two effects. One is like AI is so hot, everyone wants to go work in AI. The other effect is that everyone's worried about AI taking all the jobs, so they want to go and work at real things. Have you noticed anything uh out of the younger generation uh the shape of the type of person that you're trying to hire? What does it look like to make a career at Impulse these days?
2:17:54Yeah, so hiring uh we're hiring like crazy right now and I think we're be we're doing pretty good in this environment. There's a lot of startups around here that have really drained the talent pool.
2:18:05Interesting.
2:18:06It's a little harder to find more seasoned uh people experienced like at the senior level where it's it's more difficult. We're getting a lot of smart kids right out of school or um people from other industries and that they're moving to to our exciting industry.
2:18:20The hardest place of course is is in in the software and coding. Um that there's so much competition for those guys that uh that's always been from when I started the company. always been you pay a premium to those guys and you know it's just it's just hard to get them because there's you know AI is like the you know the big money maker.
2:18:38Yeah. Yeah. Yeah. Yeah. It's a crazy time. Are you are you relocating people from the Bay Area or pulling from universities that are outside everything? All of the above wherever you find everything all over you know we as you know we have an office in in Colorado.
2:18:53There's a lot of talent out there. We our guidance uh navigation and control is out there. We got quite quite a bit of software people. We actually put a a machine shop out there because we got some some really good machinists out there that are making precision parts for us. So, you know, yeah, we're expanding uh to other other areas where the talent is.
2:19:11Yeah, that's really exciting. Well, thank you so much for coming on the show. Congratulations on the extension. Great talking to see it. I I can't wait for more progress. Have a great rest of your week. We'll talk to you soon.
2:19:23Goodbye.
2:19:25And with that, that's our show. Is there anything else that we didn't talk to you? Any breaking news? We know the Fed hiked. We know that Palunteer partnered with Chipotle, the two most important stories of the day, apparently. Uh, I think we've gotten through everything. If we didn't, we'll get to it tomorrow at 11:00 a.m. Pacific. Uh, get that flashbang ready. Leave us five stars on Apple Podcast and Spotify. Sign up for our newsletter at tbpn.com.
2:19:52And we will see you tomorrow.
2:19:54throwing flashbang. Boom.
Insights
Tomasz Tunguz
Profile
- Dario Amodei’s AI-safety letter (letter): Tomasz Tunguz called it “really wonderful” for moving the discussion beyond a simple pro-AI/anti-AI split and clarifying actions needed at the individual-company, cross-company, and international levels.