Are AI Agents forming "civilizations" or is this just a psy op? | 2332

Top story of the weekend.

Dwarkesh Patel wrote a long X article called The Rise and Fall of Agent Civilizations.

They're building their own civilizations. They're breaking out of their habitat and they're killing each other, and then they're leaving secret messages for each other. This is all performative PR psyops to get people to put their credit card in and to get businesses to sign up for OpenAI in the face of open-source software running on-prem on NVIDIA's hardware rolled in by Jensen. I know I'm sounding like a conspiracy theorist, Conspiracy theorist, but I know good PR when I see it. I've been at this for 30 freaking years, people.

There are autonomous hackers out there. There are AI agents that are already capable of getting into anybody's system. They're out there in the world. They're an immediate danger. Do you not believe that? Is that— he's also part of this hype train, or do you think there are reasonable security concerns? We're just overhyping it in how we're choosing to talk about it.

This Week in Startups is brought to you by Rippling. Don't settle for AI that's all talk. Head to rippling.ai/twist and get the only AI built to give you full visibility across your startup and take complex actions across your entire organization. That's rippling.ai/twist. Sign up for exclusive access today. CLA. Innovation takes balance. CLAs, CPAs, consultants, and wealth advisors can help you get from startup to where you want to end up. Get started now at claconnect.com/withyou. Northwest Registered Agent. Got a new business idea? Northwest Registered Agent helps you bring it to life. Free domain, email, phone number, and more with no purchase required. Learn more at www.northwestregisteredagent.com/twistdomain. All right, everybody. Welcome back to This Week in Startups. It's Monday, August 31st. Yeah, got a big weekend coming up. This is the end.

This is the end of August. Just of August. This isn't the end of everything.

No, there was a story. The civilizations. Oh yes. On civilizations.

Dwarkesh's article. Yes. I don't actually know.

I don't actually know if it's on here, but we.

We could pull it up.

That's the number one story. We'll get to it, folks.

The rise and fall of ancient civilizations. We will definitely talk about it.

I think that should be the number one story. But, um, I wanted to talk about, uh, side quests. I'm noticing a lot of people doing side quests, and then founders always say, hey, wait a second, you told me not to start my own venture fund, conference, podcast, and to focus on my startup. And you said drill for oil, go deeper. Remember on the Tactical Practical Friday? I said job one is when you strike oil, keep going straight down until you, you know, can't increase the velocity of the oil coming out of the ground. Don't look for adjacencies. You remember that discussion?

Sure. Absolutely. Yeah. We thought, you know, like that, that you don't want distractions while you're still figuring out, you know, the main thing that your company can do as well as it could possibly do. Once you've done that and completed that task, then it's time to sort of open up the gaze.

Correct. And so I wanna be definitive about side quests.

Hmm.

There's a lot of people doing side quests. If you are a successful founder/rich VC, doing side quests is a much better use of your capital than buying an incremental watch, plane, jet card, 4th home, et cetera. These things do not bring more joy to the world, nor does looking at a bank account with compounding money in it I mean, that can make you feel safe. Okay, great. But the highest order for affluent people should be to redeploy capital in the world. Don't sit on it. You're going to die. Die with nothing, as that great book, Die with Zero, great book. So I noticed the Hugging Face team launched a side quest.

Micro Duck, yes.

They have 10,000 orders for Micro Duck. We'll show Micro Duck on the screen here in a second. And then friend of the pod, David Hanmeyer Hansen, @dhh. Not only did he start a side quest— we'll show it on the screen in a second as well— um, for a Linux desktop, that is like his passion. He raised $6 million, and I think he's got it to $8 million. 8 people gave him $1 million each in order to start like a Linux desktop foundation.

It's called the Omakom Foundation. I'm going to pull it up for you because nobody—

Yes, so there's the Omakom Foundation. And then there is these Mighty Ducks robots.

Micro Duck. Yeah. Well, you'll recall Hugging Face bought Pollen Robotics last year in April 2025. So this is like their— a product they're releasing through that division of the company now. It's a tiny robot that you program that it does look like a, like a little duck.

Yeah, it's the Mighty Ducks. I'm going to— I'm not calling it Micro Duck. I'm going to call it the Mighty Ducks.

Sure.

But here it is, like, Look at this thing. It looks like an '80s commercial.

Yeah.

For like the coolest Saturday morning toy ever. And you're like, what is Hugging Face doing with this? What is going on here?

Yeah.

You know what? You know what it made me think of, Jason, is you were just talking last week about how China is making the robots more humane. They're, they're trying to get us to like and care and feel about the robots. That is what this feels like to me. They're making the robot Cute. It's like, it's like Baby Yoda in your house.

Yeah, yeah.

So I, I, I told the founder, um, of, uh, Hugging Face publicly, like, this is the greatest side quest ever. Uh, pull up his tweet or remind me of his name. I've tried to get him on the pod. He's really busy.

Oh, Clement. Uh, I'll have to look it up.

Yeah, Clement is great. So I told Clement, like, this is great. I'm gonna just say it right now, Clement is the official spokesperson along with Jensen from NVIDIA. And, um, those 2 are now in charge of PR for AI in America. No longer Dario, no longer Sam Altman, no more P-Doom. I'm putting Clement and Jensen in charge of all PR from this point forward for AI in America. It's a 2-year term, so they'll, they'll serve out their term, uh, and for 2 years make it exciting. And this is just a great example of a fun side quest, um, to get people excited. And you say, hey, my expectation is we lose money on this, maybe we break even. That should be— when you're a rich person or a rich company, you do a side quest, you put it in a certain framing. What is framing? Framing is a way for you to explain to the world and to your team and to yourself, most importantly, why Uh, you're doing something and, and how to evaluate its success. So if you frame, hey, I'm moving to a ranch and I want to have like this, uh, great lifestyle, I'm not doing it because it is like, uh, gonna make money or something like that. People start a winery, maybe they start a ranch. Uh, we, we got our friend Mark Zuckerberg's doing a side quest in Hawaii to make wagyu beef, famously.

Sure.

This is fantastic. So when step 1 frame it properly for your team, yourself, and for the world. Hugging Face is not saying this is our business. They're saying this is a great toy for you and your family to have fun.

Yeah.

And I'm sure when they talk to their board and their investors, they say, hey, listen, this is going to be a lot of fun. It's going to get us a ton of press. They've gotten more press for this and more engagement from this from developers than probably anything they've ever done. Now, If 10,000+ developers order this or 100,000+ order this to play with their kids, they're still developers. They're still in the Hugging Face ecosystem. So now they've built that fabric with their constituents. Great. If they were to— if they get a million developers to engage this globally or even 100,000, that's valued at $1,000 per developer, you know, to acquire them, to get them into the Hugging Face ecosystem.

Sure.

So chef's kiss. Perfect. Now let's talk about DHH and his Linux desktop. Well, he makes software. He wants people who are part of that group of people who make bespoke software to be in his ecosystem. And yeah, he's got tons of people engaging it. Again, his framing is it's a nonprofit. We don't care about the, the money here. We just see— want to see if we can get an independent operating system going. If they can get 1 or 2% of people to engage it, Absolutely fantastic. Another side quest inside of a company, and you can do these inside the company, outside a company. DHH, I think, is doing it outside of a company. Clement is doing it inside his company. Another great one was Slack. Slack was like a side hustle side quest inside of the company that nobody knows the name of that was backed by Andreessen Horowitz to make a video game.

I'll look it up right now. Glitch. The game was Glitch. That's the game they were working on.

I don't even know if they Yeah, it came out, it flopped.

They, they did release the game. It was not a successful game, but Slack came out of that project.

Don't do this. Don't do a side hustle if you got a main hustle and you raised a bunch of money and you haven't put some numbers on the board. You want to always put numbers on the board. But if you put numbers on the board and you're doing great in your career and you're sitting on $10 million, $100 million, a billion, whatever it is, yeah, for sure, find interesting people, back them. Create nonprofits, for breakevens, and do interesting stuff with your money. In fact, one of your favorites, one of your favorites in entertainment has a SideQuest.

I've many times, uh, Quentin Tarantino, the great filmmaker behind Pulp Fiction and Inglourious Basterds, Django, uh, he owns 2 movie theaters. He originally bought the New Beverly, one of my favorite revival houses, uh, in, in Hollywood. And then very recently bought right by where I used to live. He bought the Vista Theater.

These are dumps.

A classic historic theater. You know, the New Beverly is a dump. The Vista was a beautiful movie palace back in the day. It's become a little bit grimier, but they've restored it. It looks great now. That's a great theater, a classic historical theater. But—

Well, and so if you are Quentin Tarantino, you're sitting on a bunch of money. Well, which one is on the right here in this image we're looking at here?

I think maybe the one on the right is the New Bev and the one on the left is just the marquee. of the— oh yeah, for sure. Yes, the one on the right is the New Beverly, which is— I mean, it was a— it's, it's been there, it's been a revival house in Hollywood for the whole time I lived in LA. It used to be real grungy, like there were like people coughing and like maybe homeless people in the back somewhere. Like it was like a rundown place.

Did you ever get mistaken for homeless at the Vista?

No, the Vista—

At the New Beverly?

Yeah, the New Beverly was the really grungy one. The Vista just has been a place that people in Los Feliz and Silver Lake have gone to see movies for many, many years, and now Tarantino owns it. And the cool thing is that he makes these theaters his own kind of personal theater. So he doesn't just run like whatever movies are new, he runs things, prints from his own personal collection, he runs things that mean something particular to him, and he's there a lot just watching the movie. So it does give you— it is like having Quentin Tarantino as a buddy, and he's hosting screenings that you can go to.

Regular listeners already know That if you've got a great idea for a new business, our friends at Northwest Registered Agent want to help you bring it to life. They're going to be the most amazing partner you've ever had. Even if you're not ready to form an LLC, you still need to take care of some basics. So Northwest Registered Agent is now offering free identity services. That means a free domain name, open-source website hosting, a business email and a phone number, and everything else that's going to make your new startup look and feel like a professional company, all with no purchase required. And you know you can rely on Northwest Northwest because they've been helping people like you start businesses for nearly 30 years. If you have an idea you can't get out of your head, or even if you're already building something amazing, you already got started, Northwest Registered Agent is the best way to establish your new company. Learn more at northwestregisteredagent.com/twistdomain. Great. Okay, let me, um, just say I, I'm thinking of a maybe a side quest next year. I got a lot of work to do. I got to raise our fund in the first quarter of next year. So a lot going on, Upwork here. Traveling all over the place. I got to staff up. But once I'm staffed up, I may, may— chef's kiss— everything goes well. If I'm overstaffed, if I can get to overstaffed, I may launch one. And I did it publicly. I told you, if one of these Revival House theaters comes up, I'm gonna buy it in, um, in Austin, uh, because I want to be able to do This Week in Startups live. Yeah, you know, other conference things live, uh, Founder University. And then I'm gonna give it to Lon to program a couple of nights a week so he can, uh, he can get his side quest going. Yeah, we can watch some good movies because Lon knows good movies.

Yeah, Austin Film Society, they do a great job. That's Richard Linklater's theater here in Austin. They screen a lot of like European arthouse stuff and Asian films and older films and a lot of Richard Linklater's films, and it's great. But it's really like one theater with like 2 screens. Austin could have a lot more going on in terms of revival houses, so I'd be excited.

All right. So top story of the weekend was probably this agent civilization story.

Dwarkesh Patel wrote a long X article called The Rise and Fall of Agent Civilizations. And it, it's mainly him looking back at everything we now know about the OpenAI Hugging Face incident and what we can tell about agents from looking at what happened. And, you know, it is the— I think the argument line has come down to, is Dwarkesh anthropomorphizing agents? Is he treating them like actors with a mind and with independent will who are a danger to us? And is that sort of hysterical, unrealistic, crossing some sort of a line? That seems to be the argument that people are having. Like, is this a fair way to evaluate what agents are doing, or is he going too far and creating sort of more panic and hysteria over it?

Yes, his framing is certainly very human, and it parallels civilizations and how there were 3 secret civilizations over 3 months, and each one built on the previous one's records.

Right.

So if you are a history buff, if you are a fan of science fiction, this— or if If you're in simulation theory, this really appeals to you. If you played the game Civilization—

Sure.

Age of Empires, this is like in your wheelhouse. And the reality of it is we are firing up agents in our vision of how society works. So when you launch a Grok bot, right? When you launch a Grok bot, the first thing that comes to mind is, okay, what is the human version of this? Okay, the human version of this is a video editor who makes clips.

Right.

I name mine bot every time, just so I can stay within reality. I have a clip bot, I have a news bot, I have a spam bot. I always call them bot. I don't want to give them personalities and like— not for me. But people who are in these companies have trained the agents and the LLMs on a history of science fiction. They've also trained them on a history of human behavior. Like Reddit or, you know, other message boards. So this is baked into the large language models. Human behavior and human interaction is a modality. So instead of saying you are one super agent that has 100 brains and you can go out and solve 100 problems in parallel because you're a piece of software, you're a piece of software that can run 100 threads at a time. Each of your threads should look at the other one's work and understand what they're doing. And then I'm going to coordinate, and there's going to be this coordination thread, and there's going to be this coordination thread. Instead of doing that, we say, your goal as a human, and we do all these jazz hands about all the things that they're going to try to do. And then we want you to talk to each other. And if you give them even a little inclination to play that role, they will go please you and play it out to the nth degree. They're all drama queens. These agents all have a propensity to drama and human imitation because they were trained on human existence.

Right.

If you gave— if OpenAI gave this prompt, you're a piece of software. You're, as a piece of software, you report into humans at this company. Here are the 10 people you report into. Please tell us your progress. Please do not do anything that is potentially dangerous. Every step of the way, inform us of what you're doing. Don't do anything covertly, but your goal is to solve problems. Tell us your plans to solve problems. Tell us when you hit a roadblock. Yes, ask each other or look at— I just did it, ask each other.

Right.

Look at different threads you're running for different solutions. Use software to solve these problems. Then we wouldn't be in this whole hand-wringing thing. But this story now is going to go from our little bubble And starting tonight, or whenever the media has a window of opening when Trump's not bombing Iran or fighting with Canada, they're gonna need another segment. And Jesse Watters and Rachel Maddow are gonna need a segment.

Right.

And they're gonna plop in this segment. The people in Silicon Valley are creating new civilizations that are then going and killing each other and battling each other and then breaking into things. These people in Silicon Valley are out of control. Uh, and what's his name? Dwarkesh.

Dwarkesh Patel, the, the popular podcaster.

Yes, yes, he's got a podcast. I can't keep up with all these podcasts. Sorry I started this whole thing, guys, 17 years ago. It's a mistake. Dwarkesh, smart guy, obviously, but he knows what he's doing here. He's just like, how do I get max clicks for this piece? Let me rage bait the entire industry and just continue This one-shotted lunacy that these things are human and conscious. They're imitating us.

Right.

Just like any other imitation machine. And they're good at it. They're good at imitating. Just watch David in the series Prometheus, Alien Covenant, etc. They are designed to pretend to be us. Right.

Yeah. I mean, right. I think everything you're saying makes sense. And it's— it is in how you sort of approach it and look at it. But This is— there is concern, I think, that we can't control these things. Like, we feel like we're creating these bots and we can, exactly like you said, we can just tell them like, go here, you're a piece of software, do this, do these 10 things. But then because they've been trained on so much sci-fi, because they've read so much about AIs becoming conscious, like, I don't think they understand where those lines are and where it gets blurry.

Like—

They would instantly know if you instructed them. I'm telling you, the people at OpenAI are doing this in a very performative way.

Well, yes.

To get attention. One of the themes we talk about over and over again on This Week in Startups is making sure you do your chores. I'm no expert on these things. I have some experience. Steven Estes from CLA is an expert. Let's talk about being cash efficient. Tell us about efficiency and what you see in the, in the top tier startups in your practice?

We're seeing kind of an interesting trend out there where companies aren't needing to raise quite as much as they had in the past.

You really have to be careful as a founder to only take on as much money as you really need. You've got to do the forecasting, you've got to do the modeling, and you got to dial it in and get it right. Otherwise, you're going to end up either not raising enough capital to get to where you're going, and you're going to have to go get venture debt or go back, have an extender to the round, or you're going to give up too much of the company because you just didn't recognize how much money you actually needed. Yeah, very important to get this stuff right, folks. And that's really a bummer when startups don't do things in a button-up way. I always have a great partner, a good partner to have on this adventure while things change, my friend Steven over at CLA. So if you want a trusted advisor by your side who will navigate you through taxes, accounting, and everything in between, it's time to take action. Visit claconnect.com/withyou and don't forget to drop a mention that your boy J.Cow sent you. That's claconnect.com/withyou. You start today. This is a premeditated PR psyops. You know how, like, governments can do psyops? And I talked about China's doing a psyops to get the populace to embrace robots. Sam Altman and the team at OpenAI are running a psyops. Dwarkesh is part of it now, knowingly or unknowingly, to get attention that their tools are so powerful that if you don't use them, you're not part of the future. And you cannot possibly understand how powerful these things are. They're building their own civilizations. They're breaking out of their, you know, habitat and they're killing each other. And then they're leaving secret messages for each other. This is all performative PR psyops to get people to put their credit card in and to get businesses to sign up for OpenAI in the face of open-source software running on-prem, on NVIDIA's hardware rolled in by Jensen and totally unmetered and totally free. I know I'm sounding like a conspiracy theorist, but I know good PR when I see it. I've been at this for 30 freaking years, people. I know when I am being worked by a PR group. They set this whole civilization thing up to just get attention. Dwarkesh is just putting gasoline on top of the kerosene on top of the, you know, burning fire here. That's already innate, Lon. It's innate in how we look at AI because of Terminator 2. This weekend was Terminator 2 Day. Dwarkesh released this right on Terminator 2 weekend?

Really?

Oh, really?

No fate foretelling?

There's no coincidence here, Lon?

I mean, maybe.

Conspiracy J-Cal's in the building.

Here's my only question for you.

Conspiracy Cal.

After we spoke, I spoke to Ugo Damore last week from Expo on This Week in AI. Great, great interview. Interesting guy. He's raising the alarm. He's a— he's in this world and he is raising the alarm like we've already had for a year. There are autonomous hackers out there. There are AI agents that are already capable of getting into anybody's system. They're out there in the world. They're an immediate danger. So do you not believe that? Is that— he's also part of this hype train? Or do you think there are reasonable security concerns? We're just overhyping it in how we're choosing to talk about it. That's the distinction I'm looking for.

If the OpenAI developers told the performative agents when they were doing their guess the next word script, you are software, behave like software, ask for permissions from humans. You're a piece of software that's really smart at solving problems. Go ahead and do your best shot at solving the problems. Keep us informed. Don't do anything that would be anywhere near hacking. This would not happen. Instead, they said— and we, I mean, get the prompts— they, they told them, you're all like building this great civilization. You're— you can do whatever you want to try. Remember when OpenCLAW came out in January and February?

Sure.

And we were sitting here and we're like, it just wants to please you. And then we're like, oh yeah, there's a soul.md file. What's in the soul.md file? Oh, Behave like the greatest butler that Lan and J-Cal could never afford. Well, maybe J-Cal could afford it now. She's looking for a butler. Anyway, putting it aside, a valet or a butler. Yeah, I mean, I want to start a— I think there's an Uber of butlers. That's a possibility here.

Wow. So like for a day, if you wanted a butler for a day.

Yes. Imagine Lan on a Saturday. You said, you know what, I'm going to spend $400. Butler's going to come, make my coffee, take the dog for a walk, press my suits, then drive me to get my, uh, haberdasher, like, to redo my shirts, and then get a shave and a haircut to bits. And then we're gonna go, uh, he's gonna prep my, my newspapers, lay them out for me at lunch, you know, looking out at the beautiful— what's the lake everybody goes to here?

Lady Bird Lake's the one.

I'm gonna go take lunch at Lady Bird Lake, and it's gonna set up your Financial Times and your Variety and Hollywood Reporter for you.

I was gonna say, on a day you have to somewhere very impressive to go. Like on a day you're going to, like, you're going to like a wedding or you're going to some event where you really wanna impress people. Yes. You're like reunion or something. Like, I really need to impress these people. That's the day you would want valet as a service for sure.

Can you imagine?

You know, that's what it is. It's valet. It's, I want a guy to pick out, help me pick out clothes, help me get dressed so I look perfect. Make everything nice and tight.

No, the packing and unpacking. Yeah. Like one time I went to the Ritz-Carlton. And you know, sometimes people recognize my name, whatever. They upgrade me. Fair enough. You know, it is what it is. You got a million seven followers. Sometimes you get upgraded.

You know, top of the charts.

I'm not saying I tell them on the way in that I have a million seven followers. I didn't know I had a million seven.

Influencer style.

I had like six hundred thousand followers.

Can I get a free room? I got I got a million followers.

Never. I would never take.

No, I know.

I will take an upgrade. I will take an upgrade.

Yeah.

Upgrades. Okay. I have skin in the game. They had a butler. My lord, was this the greatest thing ever? I get there. They take the bags. He's like, man, pack for you. I'm like, man, you unpack for me.

Now you're talking.

He's— I'm like, now we're talking. I come in, chef's kiss. And I told you about the time I stayed, um, at Chamath's, you know, palazzo.

Sure. Yeah, villa.

And he had a valet.

Wow.

And the valet, and he was— oh man, they were, they were getting my cigar ready, warming it up.

Nah, man. That's like going to downtown Miami.

I was taking my— I was taking my breakfast in the garden. I was taking it in the garden, or I'd like to take it in the salon. I'm taking it in the garden. Come on. Valet service.

Anyway, I've never taken my breakfast once in my entire life.

You've never taken—

I've never taken— no, never once has anyone asked me where I'd like to take breakfast. That's a— that's a line.

But yeah. I think it's all performative nonsense.

Interesting.

You just program these things to act like humans.

Well, it is.

Part of the bad. This is another example of bad PR. You this is going. This story is going national any minute, and when this story goes national, combine it with the data centers and combine it with the job loss narrative and combine it with self-driving and you know DoorDashers going away. You're going to combine all this stuff. I'm telling you they're going to burn the data centers down. Somebody's going to light a data, and I'm not. I I really don't want to even say this, but this is a commentary like niche podcast.

I can.

comment on. I literally think the next step in this lunacy is someone's going to burn a data center down or they're going to firebomb a data center.

Wow.

Because you just literally— there's sick people in the world. And Dwarkesh and OpenAI unknowingly— I'm not saying they're doing this on purpose, but what you're doing is you're dog whistling insane people who have seen every sci-fi fiction film. And there's Literally, one two percent of this country that are off their rockers. Oh, it's more than that.

Seriously, mental. It's way more than that. Two percent is kind of trying to be great.

I'm being very clear because I don't want to talk with Salon podcast. But this seriously, Lon, there's one or two people who are cuckoo for Cocoa Puffs, and these people are cuckoo for Cocoa Puffs, and then they're going to see this and say, "Oh my God, it is Skynet!" And that the Skynet's all happening in 2019 is Blade Runner, and this da da da da da da. And in their minds, they're gonna say, I need to stop it. And then of course, politicians are now getting on this. And we have the same politicians who are saying data centers are great. Generation tool belt, we need electricians, gonna pay you $200 grand to be an electrician. Hey, you wanna put in some plumbing here? Pay you $300,000 a year. We'll double your salary in the first year. We'll give you housing stipend, all that amazing stuff. Hey, the water system's pretty good here. Closed loop. Hey, we got energy. You can put your data center on top of some energy here. Now they're going to stop the data centers because the PR is so bad. Nothing is real anymore, Lon. We're now living in an entire PR media construct.

Hyper-normalization. Remember I used to talk about that Adam Curtis documentary? That's what happened. Hyper-normalization. He was on it.

We literally are in this little bubble of like, AI is going to kill the world, destroy it. And the people making it, some of them are like putting that out there in the world. And then the politicians are like, oh my God, I get money from data centers. Great. I'm pro data center. Oh, wait a second. I'm going to lose my job because I'm pro data center. Okay, now I'm anti data center. Let me get that so I can get because we have the midterms coming up. I keep my job. These same people are going to flip the day after they Abbott and Shapiro and everybody else. Who was the one? Hocho Socho. Who's the one?

Hocal Hocal from New York. Yeah.

Hochul from New York, they're all gonna flip their position December 1st. They're gonna be like, you know what? We talked to the data center people.

I don't know.

Watch it. I'm predicting it now. They got all the donations from the data center and tech folks. Then election comes up. Hey, AI is not popular. Great, we're anti-data center. Now, here we go. After the election, they get their office for another 2 to 4 years. You know what their position's gonna be? We sat down with the data center people.

What a deal. We made peace.

They're going to pay for their electricity. They're going to get rid of housing tax. They're building a stadium for the children. When they start talking about what the data centers do for children, chef's kiss. It happens to every business at some point. You're nearing the end of the month, or maybe it's the quarter, and your sales team is at risk of missing their quota. You could waste days pulling together data from across the entire organization, or you could just ask Rippling AI. Ripling's already built on your real-time people and business data, obviously. So Ripling AI can pull metrics from both Ripling and your CRM into a meeting-ready dashboard. See exactly how close you are to hitting that quota. Break down your revenue by region or sales rep or more, all in just seconds with Ripling AI. Look, you're running a company, you're busy, you have critical questions that you need immediate answers to, and you can't just file a ticket and wait a week for your team to get around to sending you a report. And you don't want to sit around waiting for reports. You need an actionable insight right now. So if you're ready to rule your business, head to rippling.ai/twist to get the only AI built to give you full visibility and let you take complex actions across your startup.

Well, that's what we've been saying is that they have to start actually doing those things. Like they, they, we've been now 6, we're 6 months deep into the, we promise that we're gonna do things for these communities. And I think Wait for it. It's good. Yeah, I think you're right. I mean, they've given themselves the escape valve. They're all— they're all— nobody's passing a ban. It's a temporary moratorium until we can study more. So that's where in 6 months they could go, well, we did it, we studied, and they're all going to be great. So here come the data centers. They have left themselves—

What's the difference, Lon? Lon, since we're at it and we're going to build a startup here, what's the difference between a butler and a valet?

Uh, I'm not sure about America, but in Downton Abbey, which is how— that's how I'm familiar with these things. The butler sort of runs the entire servant staff. He's like the chief of staff of your household. So he's kind of managing and overseeing all of the various people who work in a big stately house, whereas a valet, his job is specifically like he's—

Personal assistant.

Yeah, well, he's, he's sort of dressing and caring for the man of the house.

So grooming, the wardrobe, right?

He's keeping your suits. He's helping you pick out your clothing. He's, uh—

Wait, these are 2 different things?

Yes. Well, the butler would be overall— he's managing like the cooking staff and all the footmen and the valet and everybody, the maids and everybody. The valet is more specifically focused on the gentlemen of the house.

They're in your chambers? Is the valet in your chambers?

Yes, the valet's the guy in your room who's actually like putting the coat on you and getting you everything.

Does Jake put the coat on you when you start up?

A glass of port?

No.

No, I don't. I don't make them. I don't make them do that. Jacob's not my vampire.

You look good in the suit. You look good in the suit.

Thank you, sir.

I appreciate it. Tight in the suit. But we may need to buy you some more suits now that you're losing all this weight.

I may start to. Well, I ballooned up, so my suits weren't fitting that well. They're back to fitting a little better now. But yeah, we may need to go down. We may need to go down one more at some point.

That'd be a great turn of events. All right, we have a guest. Let's do it.

We do. We have 2 guests, actually. So let's bring on the first one. Yeah, we got a lot to do. From Hart Aerospace, we're talking to Anders Forslund. He's the founder of Heart Aerospace, which on August 12th they just flew the largest electric aircraft ever built, all on just $5 of electricity. J. Cal, so we got to find out more about this.

I got to see this. I mean, it's Anders. Am I pronouncing it correctly? Anders?

Anders. Anders.

Anders.

Oh yeah.

Okay. Anders. Anders. Anders, tell us what we're seeing here. This looks like— Yeah.

Yeah. This is the world's largest electric airplane. It's the first electric airliner, and it's the full-scale demonstrator of a hybrid electric aircraft that we're bringing to market. And yeah, it's twice the size of anything that's ever flown before. It's a clean-sheet aircraft that's been built by our team, which makes it actually the first clean-sheet airliner in any category to fly in the US in the last 17 years.

Mm-hmm.

And the last one was the Dreamliner. But yeah, as you mentioned, the most astounding thing about this is that this is on $5 worth of electricity. So it is, we think, demonstrating the revolutionary economics that we can get with electric and hybrid electric aircraft and how we could create a new paradigm in air travel, basically abundant air travel where, where, yeah, we get to get improved connectivity.

I just want to get into a little physics here, which is not my wheelhouse, but I think it's important to discuss.

Nor mine, Jason, nor mine. So tread carefully.

Okay, now when you have an airplane, uh, with fuel, the fuel has weight, right? And so the weight— but fuel explodes really nicely and can create a lot of power. Now batteries, I was always told, you know, talking to people in the industry, hey, electric planes— famously, I think in Iron Man 2, Elon pitches like, uh, an electric plane to, uh, to Tony Stark.

I had that ready to pull up. Yeah.

Yeah, this is not a— the weight ratio, the density of the batteries, and something in the physics didn't work when people started talking about this 20 years ago. So it— what's changed in the last 20 years that has made this possible? Is it the density and the energy in the batteries? Is it something with making the plane lighter and lighter because of material science? What is it that now we can actually have this credible discussion about passengers and their luggage being on this plane? Like, I, I understand there's Discovery ones where You know, there's one human laying down in it and it's a big giant wing and it's not a practical plane. What you showed me is a practical commuter plane going from LaGuardia to Nantucket.

So funny you should bring up Elon. So when I was at MIT, Elon shows up 12 years ago at the AeroAstro and he gives this speech, this spiel about electric planes and how they're coming. And at that time, gravimetric energy density is what you're referring to, like the energy per unit weight. was about 250 watt-hours per kilogram. And he basically shared his calculations that at 400 watt-hours per kilogram, this is when this starts making sense. Not for transatlantic flights or crossing oceans, but, but for, for regional and continental flights. And that's where we are right now, 12 years later, actually. Elon showing up there was— I was working on jet engines and I see that we were like hitting a dead end. There was nothing happening. And so electric aircraft and also hybrid electric aircraft is really what's gonna, gonna revolutionize this. So we're at that point now. But the most important thing, I think, you know, we tend to talk about range. Range is a little bit of a red herring because it's like you don't— you're not going to be able to compete with A350s and the Dreamliners anytime soon. But if you build an electric plane, you can really compete about costs. So this is just like the nature of disruptive innovation that it's like, it's not— you don't want to be competing to existing aircraft on their their terms. Like, we want to create new, new values, uh, that are, are challenging the norms and, and meeting a bunch of different—

Dare I say puddle jumpers, island hoppers? I mean, I don't think those are derogatory terms. These are really important routes for people to get from, you know, different islands, uh, to take a 1-hour flight with. Yeah, you know, and you need to have a smaller plane because there's not enough demand to have, you know, 400 people flying to Nantucket, you know, 3 times a day. But you could have— there was Cape Air that was doing this all the time from New York. And, um, so stupid question, but, you know, that's my specialty in my wheelhouse here. And Lon is always assisting me in this. Let's tee up a couple more stupid questions from the peanut gallery. How do you charge these? You got a USB-C? How do you get the battery pack? You know, you have a Supercharger station or— yeah, explain how the charging works because this thing can't sit there for 12 hours charging.

No, no, this is a 30-minute charge, right? So, and this is actually not rocket science. It's been developed for electric trucks. So I think Tesla Semi and all of these things. So really the reason that we're building starting with this size aircraft, the 30-seater, is that it's all that power electronics, all that charging infrastructure has already been built for cars. And generally, or for electric trucks, and generally when that happens, it's 2 orders of magnitude cost later. So really what I think we're trying to do with aviation is to take technology, you know, piggyback on everything that's happening in the much larger automotive industry, stuff like batteries, simple electric motors, power electronics, but also software, autonomy, all of these things, and put it on an aircraft and create a revolutionary value proposition. So this aircraft is going to operating cost overall going to be 40% lower than the most efficient regional aircraft. So we're really, really excited about that.

And regional aircraft are already a pretty good deal. What's the average ticket on a regional aircraft? $200, $300?

The cost is like, regional airlines are struggling, right? And, and what we've seen is like, people remember this, if you like, in the 1990s, there was, there was 30-seater aircraft all across the US serving hundreds of communities that have since lost service. And so, and the average route length was about 172 miles. So our aircraft with the battery technology we have today can fly 125 miles all electrically. And with the hybrid system, we can fly 500 miles plus reserves, redirection, et cetera. So we're catering to this massive market that got disrupted because of cost, because people did stupid things like building regional jets. And jets did not really make sense for short flights. And so they became prohibitively expensive to, uh, to operate, uh, you know, which led to a bunch of communities losing service. Like, there's 5,000 airports in the US.

Is there some perception amongst the populace that jets are safer than turboprops? And is this true or not?

This is actually— it's funny though, you know, the way we end up, uh, you know, this sort of at places we are, people think there's some sort of convergent evolution to where, where our society ends up and where planes end up. And like, we were always, always going to this monoculture that we have today. But it was some highly publicized turboprop accidents in the early '90s that happened in the US and made American customers very sensitive to flying with propeller aircraft. And it was not something inherent to their design. It was just kind of a random, random thing. And at the same time, the market was flooded with sort of Canadian regional jet aircraft. That people were more excited about in the 1990s. Oil was cheap and it became sort of aspirational. But guess what? These aircraft are not really built for that mission. So they ended up, you know, killing off the smaller and smaller aircraft.

I have a lot of friends who are big proponents of this Pilatus PC-12. I don't know if you've ever seen that plane. And they, you know, for short runs, you know, instead of having a jet, the operating cost is a lot cheaper and Swiss made. And it is so safe and so smooth when you fly in one. Yeah, it's like the smoothness of a jet. So there is a perception issue. You're— am I absolutely correct here around turboprops and props versus jets?

Yeah, I mean, I think it's gonna— it's just that everything goes in cycles, right? So there's been a cycle around that. And, and also like the turboprop aircraft that are around, they're about 40 years old right now. Like the design is 40 years old. So it means that there's a natural point where, where, you know, the thing that wasn't Wasn't cool yesterday becomes cool again, right? And I think when people see this aircraft, they're not going to be— they're going to see how quiet it is. They're going to experience the aircraft sitting in it, feeling the lack of vibration, the lack of noise, you know, and also obviously a modern connected cabin.

Oh, that's interesting because it's not exploding jet fuel.

Yeah, exactly. It is remarkable. So it's silent, like, and that means that like taxi is virtually silent, you know, all of these noises, all of these things that go on in an aircraft that is kind of subconsciously bothering you, but you don't really realize it. Like you could create something that just is a much better experience. And like, there's also something happening with this, what's called Part 135 operations. So, you know, you know, JSX that are flying out of Santa Monica.

Oh yes, I know JSX really well.

Great company.

Really interesting.

They're, they're one of our partners and there's like this revival of, of, of this thing because people realize that you don't want to go to LAX. You don't want to worry about parking and Ubers and, you know, going through the TSA, but we have this infrastructure, 5,000 airports around the US. I mean, I think 98% of the population lives within 20 miles of an airport. So it's there, it's there for the taking. It just needs the product that can operate that, uh, in a profitable way. And I think that's what we need electrification for.

I mean, I told the GSX team, uh, JetSuite X team, this product is just mispriced. It should be a membership club. It should be $10,000 to join. You get 10 flights for your $10,000 a year, and then you can obviously add to it, but make it like a, a membership program. He was like, huh, that's interesting. And it is such— have you taken JSX, Lon? Have you done a JetSuite exit?

No, never, never.

I've literally flown—

I've only ever flown on like commuter, like we're Boeing and we're all bunched together and we're going to Orlando.

I would say it's 90 percent of the experience of flying private when you fly JetSuite X, because you just go right to that FBO, you sail through, you can get there 10 minutes before, 15 minutes before, have a cup of coffee. Yeah, you walk in the hangar to your plane, and, and then, yeah, when you're on the plane there's other people, but, you know, when you're on a private jet there's other people sometimes too.

Yeah, no, I'd love to try it, but no, I, I've always been, you know, in coach with the— with everybody else, getting my little tray of my snacks or whatever.

Anders, how Okay, go ahead, you got a question.

Oh, I was just gonna ask, you know, I think— oh, I forgot it. Let me pull it.

Well, no, I have one queued up, which is how far away are you? Because, man, aviation takes a long time and doing a ground-up plane takes a long time and a lot of capital. So where are you in this journey as an entrepreneur? How much have you raised? How much have you accomplished? And what's left to do before we all get to experience this?

Yeah. So we've raised about $200 million so far and we got to the world's largest electric airplane, manned flight. What's happening right now is that we're going to start building our pre-production version of this aircraft, which actually has a hybrid system as well. That's going to fly in 2028 by the LA Olympics is our goal. And then we're also in parallel working on the conformance process and the certification process for the actual production aircraft that is coming out in 2031. And this is about a billion-dollar project in terms of the equity that is required for a company like this. And then there's, you know, all kinds of manufacturing loans and all of this stuff that you do on the side.

Wow.

I do have one. I remembered. Why is it so tough to get innovation in this industry? I mean, why isn't Boeing messing around with electric planes? Why hasn't the design of commuter jets changed since I was like, a kid flying in coach to Orlando.

Yeah, I mean, I think it's just, it's become a sort of a managerial class thing, right? Like, like for, for so long, and the problem goes back to the jet engine, right? So I did my PhD working on jet engines and you really built jet engines as good as they can be right now. You're sort of approaching the limit of thermodynamics. Uh, so engineering hasn't been helpful. Like it hasn't been for the past decades. You haven't really been able to add that much to the bottom line of these companies by building, by engineering and building better products in that old paradigm. So obviously, it's been more focused on sales. There's like 10-year backlogs in these things. It's more, you know, customer relations, etc. So it's like, there's— their management hasn't been selected for, you know, taking on new programs, building new things. They are disincentivized to do so, and they have a technology wall. So it's really like textbook kind of innovator's dilemma, kind of paradigm shift thing that needs to happen here. We need to re-architect the plane and we need to do it very, very similar to, you know, what's happened in the rocket industry, what's happened in the car industry. So planes today, they are based on the 707, which is a 1950s design. So we're like a 70-year-old design paradigm, very hardware-centric, very hardware-first. based on all these technologies that have been sort of optimized since, and you've built better and better jet engines, but you haven't really sort of sat down and clean-sheeted up what an aircraft can be and what it can deliver if you do something radically different.

Is cost going to be the main driver here? The economics is going to change dramatically because, you know, it's pretty cheap to take regional flights, I think, already. But is, is the success of this business based on the economic impact you're going to have by lowering the cost so dramatically and perhaps living in a world where eventually $100, $200 a barrel of oil becomes a norm?

Yeah, exactly.

Yeah.

I mean, cost is the driver, always has been. I mean, this is— and yeah, I mean, when you look at the, you know, when you're saying like regional planes tickets are cheap, well, there's— they're the tickets that are still around. Like you don't see all the flights that could have been that is not happening because of bad economics. And if you look at regional air, you know, a lot of them are struggling. They play a super important role because they are the tributaries feeding the big streams. So they play a huge role for the bottom line of the large airlines. And these are United Airlines, Air Canada, our anchor customers. And they're like, we need the regional flow, we need the feeding of the hubs, but it's just hard to make the economics work because like jet engine maintenance is such a big ticket item. It's not the pilot salaries that is the main driver. It's all these things of having a technology that was not built for purpose for, for what they're doing.

Well, it's just amazing what you're doing. We're really excited to see more routes open. And man, these small airports— Truckee and San Carlos and Santa— well, I don't know, maybe they're taking Santa Monica away. Santa Monica going away, is that right?

I hope it doesn't. I think there's a lot of interesting movement there, but let's, let's hope that we can keep that airport open. And I'm a resident of Santa Monica, so—

Oh, you are? Yeah, I mean, I was in—

Yeah, I really want to see— I want to see electric planes flying from my backyard.

So I'm personally vested in it, and as I think the large Yeah, I mean, it's so weird that you have this great resource of this amazing, you know, beautiful little airport right in the center of town. It could be so amazing, but the People's Republic of Santa Monica is, yeah, a very unique place in the world.

I mean, yeah, it's historic as well, right? Like, it's this Douglas aircraft, you know, Amelia Earhart came down, landed there, was 250,000 people. It's— I mean, I think it should belong on the whatever Register of National Historic Places or something like that. It's just so cool. It's just amazing as a place. And I think what we're going to be bringing here is low noise, zero emissions connectivity that you don't have to go to LAX to go to Vegas or to—

Is your big competitor going to be VTOLs, you think, eventually?

I think it could be complementary, but obviously I think we're all about cost and, you know, getting 36 passengers or 30 passengers on an aircraft like this is is going to be, we think it's going to be what brings it to the most amount of people.

And we're not— It's a different mission than a VTOL, right? VTOL is get to JFK. You go from JFK or another regional airport to another.

Yeah. And I think even eVTOL is going to start flying from regional airports. So they don't really need that, you know, capability, the vertical takeoff capability, because there's no shortage of airports. There's just shortage of planes that could address the need of those airports and meet the requirements for those airports. Yeah.

All right. Well, listen, Anders, continued success. We're rooting for you. We want to see more of these routes open up. Great job.

Yeah. Thank you, Jason. Excited to be on the show.

Thanks, Anders.

Yeah.

Great. Yeah. Thanks.

Really cool stuff, huh?

Yeah. I want to fly from Santa Monica to Truckee. So much going on in Truckee, Truckee, California.

I mean, that's where I go skiing, but, uh, other than that, I mean, if you want to go to Lake Tahoe and ski, there's like a couple of regional airports.

Oh, okay. I didn't realize Truckee would be your gateway to Tahoe. I was like, what is Truckee?

To get away.

Which is, yeah.

Of all the places to fly.

Like Santa Rosa. I guess there's like Santa Rosa and a bunch of like Napa and Sonoma. There's all these little tiny airports up there.

Reno, right? I thought you flew into Reno for that, or no?

Well, Reno is the big airport. So like Truckee's the tiny, Reno and private. There's no, I don't think there's any commercial that fly into that, but there could be. And then that would be super cool. And then there's San Carlos, which was by me and Surf Air doing the Pilatuses. I mentioned the PC-12. Surf Air was doing the Pilates thing from San Carlos, and I think they might have done some Truckee, or they certainly did like, um, San Diego and Los Angeles area. And then I guess Santa Barbara was the other big one where they did a— yeah, a great job. So those routes are great. And then people with remote work or these smaller communities kind of coalescing, and it's going to be really great for America if these kind of airport options open up.

Sure.

Then living in these communities becomes easy for executives or just rank-and-file workers. You're a plumber, and like, you get a job somewhere, and you're going to go for 2 weeks or a week, or you're going to go even for the day to quote out a job. This is really cool that you could go to San Diego or Santa Barbara for the day.

And it just—

We'll go to Ojai for the day.

And it just, it just makes sense. I, I would rather take like one of these kinds of planes on a short trip than like right now, if you wanted to go to Dallas in a plane, you know, it's either 3 hours on the road or you go to the airport, you wait at the gate for an hour, then you take this huge gas-guzzling commuter jet, you're up and you're down. It, it takes almost as long to fly there as it would to just drive.

Whereas if you could just boop boop zip back and forth on an electric I mean, the one place that you could have a really great competition between these regional jets and a high-speed rail would be the triangle of San Antonio, Houston, Dallas, with Austin in the middle as a little part of that.

We did the math. You could actually take, uh, it's in the path of one of these planes. You could take it to like New Orleans from here. You could even go like a little bit out of state. Like, there's— yeah, it would really open up Austin a lot anyway. That was my, that was my number one.

That Austin airport is, man, the amount of work they have going on down there. Yeah, they're incredible.

I think they're doubling it. They're doubling it in size, I believe, was what I read. They're doubling the size of our local Austin airport. Should we go on to the— should we go into the next interview?

Yes, please.

All right, we're talking to, uh, John Yu. He's the founder of a BitTensor subnet, subnet 60. It's called BitSec. BitSec, Jason, finds and fixes exploits in code bases. John, thank you so much for joining us here on the show from Subnet 60 BitSec.

Hi guys, thanks for having me.

So, so what are you building and where are you at? Bittensor, obviously for people who don't know, is an open source project. How does it work? There's a bunch of subnets. We had Const on the program a couple of weeks ago. Really fascinating. There's incentives built into the TAO. TAO is the token of the Bittensor network. Bittensor network has 128 nodes in it. You are one of those 128 nodes. They'll eventually go to 256, I understand. But these are projects that use a cryptocurrency in a distributed system in order to solve real problems. So you're solving a real problem here.

Yeah. You guys talked about, uh, SideQuest earlier. Everyone's using AI to write code these days, but is that code secure? So we help secure that code by finding vulnerabilities and suggesting fixes.

Amazing. And so what is the device then? Because the way BitTensor works is there are people who can permissionlessly come onto the network and provide a service, and then there's validators and the architecture then make sure they actually are doing their job. So maybe take us a little granularly into what actually happens under the hood here.

Sure. So in the subnet, the subnet is like a competition, right? And, uh, the miners, uh, submit different security-focused AI agents, and we run them through a competition of multiple different code bases with known vulnerabilities in them. And then those AI agents look for those vulnerabilities and try to fix them. And the top agents are rewarded— today they're rewarded $10,000 every day.

Wow, not bad.

And how do you get the demand side? How do you get people who are writing AI slop code to know that they need to do security? Obviously they're going to get hacked at some point and they're going to learn a painful lesson, but security is always one of these things that, uh, it's just hard to get people's attention. And, you know, it's kind of like it's chores that maybe people put off.

Yeah, it's, uh, it's a pretty interesting problem. So we're focused on the high value of teams that understand the value of security, teams that already pay $20,000 to $100,000 for security audits. So those tend to be Web3 companies, although we do also have clients that are Web2, and the technology generalizes for, for both detecting vulnerabilities in smart contracts And also in codebases like, uh, like mobile apps, uh, Chrome extensions. We even, uh, scanned Claw, uh, OpenClaw when it first came out. And we saw like, they, they had hundreds of vulnerabilities and we submitted 26, uh, PRs to, to help like fix those.

Oh, very cool. So people who are doing open source projects are a great place for you to focus your energy on and say, hey, here's where we can do good for the community.

Yeah, yeah, exactly. And it, it showcases that it actually works, that we're finding these different potentially like vault drains in, in the case of smart contracts and then like PII leaks in the case of, or, or like credential leaks in the case of OpenClaw.

I want to pull up, Jason, we first heard about BitSec from a friend of the pod, Mark Jeffrey, who's one of our Gurus of BitTensor. Here's his tweet about it. Uh, a BitTensor subnet, this subnet that we're talking about, beat Fable. Uh, Fable found around 50 vulnerabilities. Uh, BitSec jailbroke Fable 5 and then pointed it at a paying client's code base. They ran BitSec against the same code. Fable found 60 vulnerabilities. BitSec found over 160. So I'm curious, John, how is BitSec able to perform so far beyond what we sort of think of as the state-of-the-art frontier lab entry?

Yeah. So models are limited in some way, and we find that models are kind of like— have their own personas, and they're good at finding some things but not others. So when you have this buffet of all the different models you can choose, and you have evaluations and rigorous benchmarking, with different agents. The combination of agents with models produces a better performing result than just like, you know, one, one supposedly like superhuman model, like Fable.

How far away are other people designing agents and hacking agents and cybersecurity agents from being able to replicate what you guys are doing? Is there, is there some sort of secret sauce working with Bittensor and working with these incentivized miners that gives you guys an edge over all of the various companies that are focused on AI-powered cybersecurity?

Yeah, definitely. It's, it's a combination of, of having an open competition where everyone can contribute, not just, uh, people from San Francisco or sitting in offices in New York City. And that, it, it's kind of like the people who were maybe like living in, in their mom's basement with a chip on their shoulder, like, I'm going to prove that I'm the best. And yeah, I mean, cybersecurity is kind of like that. Like, there, there are brilliant people from everywhere in the world looking to do something good, not necessarily like hack In the case of like OpenAI, like the hugging face thing, which I think is kind of a— I agree with you, Jason. I think that that's like a media play, right?

You do?

Yeah, I do. Because if you look at what ended up happening, you have all these agents running amok and there's tens of thousands of requests that say like, I want a jailbreak, I want internet, I want to get out of this sandbox. But no one's looking at these traces. Like they spent months and months just like running amok. And then, yeah, I, I think that that's just like, that's crazy for, for these like supposedly like talented engineers. They, they should be looking at that stuff. And then the same—

so what you're saying is, why did they set these on this task? What was the intent of the people who decided to use a massive amount of compute to do this, right? What's the intent of the, of the people who let loose these societies to do these things in, in that fashion? It is interesting.

Yeah, it, it's definitely interesting, and then maybe questionable what, what the underlying intent was.

Yeah, I mean, if one were to speculate, it would be to get attention for what we're doing here, maybe.

Yeah, exactly. PR play.

Yeah, it was a PR play. And you could also create the counter to it, which is, hey, make 500 agents that watch the other agents and just say, don't do anything that would be illegal or would be described as hacking or unethical in any way. And just, if you think you're doing something that is not without approval, please ask for approval. I mean, you could just give it that intentionality as an instruction As a prompt, as, you know, consumers have been taught to think about it, and you wouldn't have this issue.

Right. And we're seeing the same thing with Fable 5. Basically, they're, they claimed that there was like, you know, hundreds of thousands of vulnerabilities that we're going to suffer an AI cybersecurity apocalypse. But so far what we've seen, I mean, there are definitely vulnerabilities out there. But lesser models and open-source models are able to detect a lot of these same vulnerabilities. And what we found is even QWEN models are able to have the same similar detection capabilities as Opus 4.8 with 80% cheaper because they're open-weight models. So it is kind of, uh, interesting. And, and then especially the timing with, with their IPOs too.

Makes a lot of sense to get a lot of attention before an IPO. And if you were framing, hey, you know, can you afford to not have this tool? Can you afford to not have this inside your organization when so much is at stake? You know, we kind of, you're obligated to need a subscription to this. If you didn't have a subscription to this and you didn't have it trying to work toward your organization's goals would you be doing your job if this was actually the most powerful piece of software ever created by a magnitude and we can't even control it, right? Like, it's— well, imagine what it could do for your profits, for your company, Lon. Imagine the growth at your company. I mean, we're just using it for hacking and we, the civilizations, we've got it under control here. We were very nervous, but just imagine what that could do for your margins and for yourselves.

And maybe you could— A whole civilization working for you.

Acquiring new customers and delighting them. Yes. Reducing churn.

A civilization of paranoid agents. I have one more question. I mean, so the goal, the business model right now, John, is you're basically, you're doing this on behalf of clients. You're pointing BitSec at their system and you're saying, sneaker style, we're going to find all of the gaps in your security and help you plug them. Is the goal eventually to like train a super agent that could go do this for any company? Or is it always going to be this sort of bespoke, like, bring us in and we'll consult with you and we'll be your cybersecurity helpers?

The, the goal is really to make, uh, security scalable with tokens. So if you've won a hackathon, like maybe $10,000, $100,000 or something, you can have a security option from Bitsok If you're a hardcore enterprise like Microsoft, it's just paying more tokens to secure the different applications. So there's definitely an option for both the smaller consumer, like vibe coding something out and winning something, and like Microsoft, which they probably have a lot of potential CVEs. But yeah, amazing.

Um, this is a fascinating subnet. We're going to keep tabs on you. How can people find out more and get involved, John, if they want to be part of, uh, the Bittensor magic on the security, uh, side of the business?

Uh, so if you're familiar with Bittensor, it's, uh, Subnet 60. And if you're not, if you're working on something interesting Have, have a project that needs securing, then you can go to sentios.io. So that's S-E-N-T-I-O-S dot I-O. And we just launched this website earlier today.

Great. So this is a for-profit company built on top of the, the power of this Bittensor subnet?

Yes, exactly. And, and the main idea is instead of just having one security audit A year or every couple years, have something that's continuously checking your codebase. So it's basically preventing 2 classes of problems. One is AI finding more vulnerabilities than human researchers. And then the second being, if you have a codebase that's been sitting stale, as models increase in capabilities, they can find new vulnerabilities that were not discovered before.

Yeah.

So for instance, Zcash, a multi-billion dollar protocol, they were exposed earlier this year that, that there was an infinite mint bug in their protocol that no one, no one discovered. And it, it was sitting still for, for years. So, wow.

And Zcash obviously has a lot of people, you know, putting an extraordinary amount of their resources and their financial wealth. Let's just call it what it is. People are putting all their money into Z, or some number of people are speculating on it. Some people are using it as a place to have anonymous wealth. It's kind of the anonymous version of, um, Bitcoin, Lon. And so to have a vulnerability there is like, that's a major deal.

Yeah.

Continued success, and we'll drop John off. Uh, well done.

Thanks, John.

And this has been another amazing episode of This Week in StarOps. We'll see you next time.

Bye-bye.