How Ethereum May Have One-Upped Bitcoin in One Big Way - Uneasy Money

agents can't go and make a Wells Fargo account or a Morgan Stanley account, right? Be on these payment rails. And in reality, it should out that. I mean, that to me has to be where I think the next bull run is going to come.

I think we just need to convince the claw bots that Pangu is the social medium of exchange.

Once your prefrontal cortex kicks in, you're like, "Wait, hold up. Maybe I shouldn't be advertising my entire criminal endeavors linked to my handle that I'm very proud of and posting it on the internet for everyone to see. Maybe I shouldn't do that."

Hey everyone, I'm Kane Warrick and welcome to Uneasy Money because what happens on chain never stays on chain. I'm here with my co-host Taylor Monahan, security at MetaMask and Lucanet, CEO of Pudgy Penguins.

One quick thing before we start. Nothing you hear on easy money is financial advice. We're just three builders talking about what's happening on chain and we want you to always do your own research before aping in. You can find all our disclosures at onchaincrypto.com/oneasymoney.

And before we begin, here's a word from the sponsors that make the show possible.

The energy network is an intelligent decentralized grid that coordinates smart devices to balance supply and demand. Energy Dollar is the native token of the network from one of Europe's fastest growing energy startups. Followenergy on X to find out more.

Multi-chain Advisors is an emerging technology growth firm that has helped create 50 plus billion in enterprise value for 80 plus clients over the past 4 years. They're the partner to help navigate markets. Build real traction today at multi-chainadv.com.

All right, first things first, we're going to kick off with a story that is not really about crypto, although I guess the intersection of crypto and AI is becoming more and more interesting.

So maltbot or clawbot as it is called, which is its own interesting thing, right? The claw people were like, you can't call it clawbot guys, you got to do something else and they had to change the name. So Agentic AI meets memecoins.

So what is Claudebot? Claudebot is an open-source agent that lives inside of your computer, and there's actually a security component to this as well, which we can talk about.

But you basically install this AI, runs on Claude, the LLM, and it can do kind of anything that your computer can do as an agent. You can talk to it through various channels like Telegram.

I think as someone who's been vibe coding pretty hardcore for a year and watching how much smarter these models have gotten, I'm not surprised that this went so viral because I think the gap between people's expectations and reality has kind of collapsed a little bit.

Like you go through these waves, right? There's the hype wave where AGI is coming and then you hit a bunch of issues and people go, "Uh, no, they're actually really dumb," and then everyone kind of fades it for a little while and then they get much smarter and then everyone's like, "Holy [ __ ] AGI is coming. It's like inside my computer right now." So that's what this the last week kind of felt like, I guess.

Yes. Everything's moving so quickly. I tweeted this this morning. It feels like pre-DeFi summer Ethereum.

Every day there was some crazy thing and it just felt like if you were sleeping for like 20 seconds, you were missing out on so much alpha and it just feels like that right now. The market is moving so quickly and there is something about this recursive self-improvement loop that's going on where you make coding easier and people start building more tools and then those tools make it easier to make more tools and you just get this crazy loop, right?

So this guy wrote a post two days ago and was like, "Hey guys, maybe don't expose all of your reports on the internet." It's even worse, right? If you read through it, it's quite an interesting read. It's on X. We can link it later.

But he's like it's not just that you're exposing ports. Like that would be a dumb thing to do, but there's a guy sitting behind those ports who's like I will do anything you tell me.

And the guy who will do anything, you've granted explicit permission to do anything on all of your most valuable accounts. It's crazy.

So I think there's this very weird thing. It made me realize once again that people even the people who shill security don't necessarily have a deep understanding of the foundations of security because they're like I bought a new Mac Mini so it's isolated and I'm like bro the number one reason that you don't want them on your computer is because your computer like your daily driver device has session cookies, passwords, authentication, and that all like all your most valuable things live in unlocks everything.

Those are keys. Those keys unlock all your [ __ ] right? And so that's why we recommend like don't having separate devices for things is because your daily driver device that you use all day every day has everything logged in and if they get on that device, then they can go and get anything from any of your accounts.

Now, just because you have a separate device that has all of those same session cookies and all full permission to read and write to any of your accounts, to log in, to grab the data, to make changes, that does not protect you. That's actually the core.

It's not only does it not protect you, it's actually worse, because it is the illusion. You wouldn't grant those permissions on your main machine because you're like, "Ah, like this has all my stuff on it, right?" And then you walk over next door and grant the exact same permissions on a thing that has a pipe directly to your stuff.

Yeah, and these days, people don't realize these days everything is syncing all the time. So the login and the authentication is really the most valuable thing.

Yes, you might have some old files on your computer, but increasingly the reality is all cloud. It's all in the cloud. Anything that you're keychain that has all your stuff, right?

It kind of reminds me of the hardware wallet people that are like, "No, no, bro. It's fine. I've got a hardware wallet. You don't understand." And then you start talking to them and it's like you're just signing bytes. Anything that pops up on that thing, you just press all the key. You just press it as quickly as possible. You don't look at anything like you've solved one attack vector of key exfiltration which is not the main problem and exposed yourself to this illusion of security where you're like, "No, no, no. I'm fine. I'll just I'm going to suck."

So, no, that was my. The Ledger CTO Charles likes to go on Twitter and be like, "Ledger would fix this." And I was like, "It didn't fix it. It didn't fix it, guys."

So I think there's so many things to this story. One is, yeah, bot C L A W D bot this play on Claude. Anthropic did not appreciate that and presumably he sent them a nice cease and dist. I'm guessing and the guy was like don't worry about it we're going to call it something weird that's not related to Claude. So I think that one was quite funny.

But then maybe this was probably the most eye opening thing to me and I kind of knew this but it had been a while since someone had pointed out to me and it seems like this technology has gotten better. There are these port scanner people who scan ports on the internet like just constantly, right?

It's almost like an indexer for crypto, like indexing every port that's on the internet and the response and you can just go and search and be like, "Hey, like I'm looking for this keyword in this port scanner and it gives you all of the things that are responding." I was kind of blown away by that one. That was pretty crazy.

It's fascinating because there's an incentive, right? Because there's a huge market if you can get on basically any server, any computer. there's so much you can do with it these days and thus the tools evolve rapidly in response to that.

There's also a large number of security researchers and stuff that do this for legitimate reasons but it kind of goes back and forth like they build like really robust tools and then that threat actors are like oh that's nice. I'll take that.

Luca, what's your take been on the claw bot thing? Have you got one running?

No, I'm out of that sphere a little bit. So much so that I got a guy on our team on the abstract team named Jared Watts. I highly recommend anyone listening. I'm gonna I'm learning. I'm starting. I'm not gonna. It's very obvious to me that if you're not vibe coding and if you're not just completely utilizing the Claude interface that you're going to be left behind in a big way.

Obviously I became a kind of a big pro at least on the LLM side, but that's not rocket science. But honest, I've really not taken the vibe coding things very seriously. And so Claude B outside of buying the I bought the to the the Fugazi Claude token. As for my when I see something trending, my mind instead of just what? Yeah. What's the token?

Well, it's funny. I said this last week, right? But like I've been I must have spent at this point I would say across like cursor codeex cla code or whatever 100k this year in tokens like easily 100k. And I'm like there is no direct return on investment for me. Like what am I doing with my life? Right? Like it's all like intangible. I'm like I could have bought so many to I'm buying the wrong kind of token here. Like this is these tokens. I'm gonna buy more pudgy cards, dude. What are you doing?

Are you guys using it at Infinex? Like, are you guys like I talked to Saigar the other day and he's like, "Dude, you know, things that used to take us two weeks to ship, we're shipping in two days. It's great."

So I think my hot take on this is like everything's coordination problems. The lens that the thing that got me so excited about Ethereum back in 2016 is all of a sudden there was this giant coordination problem when the Dow hack happened and immediately my coordination Spidey senses went off and I was like people are coordinating here around this thing and it's like a big deal and I that got me to go down the rabbit hole.

This feels very similar. So you have a team of people five people, 10 people, 50 people, 500 people, 50,000 people, 500,000 people. And you got to coordinate them, right? As that scales, it gets harder. The bigger the org, the harder it is to coordinate people.

Coordinating people has been the primary thing that we've needed to do, even down to the level of an individual person. The reason why Clawbot is so cool is because it helps you self-coordinate. Why people are so excited about it because it's like I've got all these things that I need to do to coordinate my own life and this thing goes out and solves those problems for me and deals with my email and deals with like so that coordination problem even down to like a single person is really challenging, right?

But then you get a team of 30 engineers and you try and get them to do something and god help you like figuring that out, right? It is really really hard. There's all these social elements whatever.

What's interesting about the last honestly like six weeks I would say right and for me I think the big switch that flicked over is claude code went from you used to have to type ultraink to get it to go into like deep thinking mode even when you had it on max mode right and they switched this about 3 weeks ago where it's on by default there's no more ultra think and that to me is where it went really parabolic and all of a sudden it was like oh wow these things are that much smarter when you make them smart all the time.

So the big question for me is how do you coordinate agents? That's actually now the question because a single agent can do something sequentially but context is really hard. They get context poisoning really quickly.

So Ralph, which was the thing that everyone was excited about two weeks ago, was about this coordination problem, right? Like instead of having one agent that tries to grind through a 200page plan, you have individual agents that spin up that are ephemeral that do a loop of work and then they hand over to the next guy and they just loop through it and loop through it and like that was like a very coarse grain solution to the coordination problem.

I think that there are many many other solutions and even Kimmy, the new, I guess it's like it looks to be people like I'm not 100% certain about this, but it looks to be like someone has jailbroken Claude and gotten the weights and has published them as an open source thing.

So Kimmy is this open source model that is almost as good as Claude. But one of the things that Kimmy has that Claude doesn't have, they've done this training mode where you have agent swarms and the agents like swarm over a task and find ways. The I think the big unlock over the next like 3 to six months is going to be this agent coordination problem. Whoever is at the forefront of that is going to be able to just blow through some [ __ ]

So Kane, you seem to be on the ground level here and I again I remember meeting you in New York, Utah vibe coding a Pudgy game. Let me get like a really highlevel macro of where you think this all goes. Obviously Daario just published that article kind of the flip side of where you think AI give me the like where are we three years from now. Where's the disruption going to lead? Give me the highest level out of the weeds perspective based on somebody who's obviously been using this stuff and I have a lot of respect to your macro takes. What do you think is going on here?

So I think I should probably caveat with like my macro takes have a history of being incredibly wrong. So you know I'll just throw that out there, right? Like ETH is not at 25K as I bet KL Samani 18 months ago that it would be. So you know take anything that I say from a macro perspective with a grain of salt.

But my hot take on where we're going is that we have very very smart individual people, these agents, they're very very smart individually for legitimately 20 minutes and then they get exponentially dumber every minute that you're interacting with them.

This is just it's the weirdest thing. and there is no real mental model for this. It's not how we think about humans, right? Like you get a human and you start teaching them stuff and they get better and they stay good and they have this upward trajectory for like 50 years before they start to like taper off, right?

Early on there's like a you get like a three-year-old, right? And you teach them some [ __ ] and then like at four they're like 20 times smarter than at three. And so there's this exponential thing. It's weird to interact with an entity that spawns super smart like autistic level like genius knows so many things and then you start talking to it and like 20 minutes later it's [ __ ] Like that is a weird thing to manage and think through, right?

That's why I say that you know the challenge is going to be in coordinating these things. So once you figure this out, and Kimmy is the first time that I've seen a really focused effort to coordinate swarms of agents. There's been people that have been trying to do like sequential stuff, but like coordinating a swarm of agents to go and tackle a task.

The interesting thing about this is that right now there's a fairly high cost even in tokens to getting a single agent to do something. So getting a single agent to do something and then getting other agents to check it is the hard problem here, right?

So all of the time that I spent over the last nine months learning to vibe code and trying to figure out where the edges are, the edges are all me. Every single problem that exists in this space is every time I have to do something. So the key thing is trying to figure out how do you remove the human from the loop and throw agents at the problem and have them kind of self-correct.

That's the hard thing once you have this recursive self-improvement and that's what you know Dario's thing right he's like we are using these models to build smarter models right once there are no humans in that loop that's where you get to like a fast takeoff scenario right where like on Tuesday the models get together and come up with like a new theory for how maybe they can make a smarter model and then on Wednesday it's 10 times smarter and then that model's like I've got some ideas and genuin ly like we are in this like early part of this curve where shit's going to get crazy like the like pivot to being like a personal trainer or something at this point.

If somebody ETSD from the early 2000s is like this is just going to be this similar you know just as just as similar to the tech bubble what would be your contrarian take to that statement was so I'm old enough to have been through a couple of AI winters right Like AI's had winters just like crypto, right? Where and this is why I say like there's a hype cycle and then it dies off and then a hype cycle and dies off.

What's interesting about the last 3 years of AI hype cycles is it's been very compressed, right? And you know Dario even said this in that recent post where he's like people get the hype cycle going like this. Meanwhile, if you actually plot the skill level of the thing, it's a linear thing that just keeps going up, right? It's just up and to the right.

My hot take is that if you look back at some of the things that people have predicted will not happen, right? So chess, you know, humans are too strategic for a computer to ever be better than us at chess. Like how could a computer ever beat a grandmaster? Well, that fell over in 1997, right? And that was not even a smart model. that was just a brute forcing algorithm basically, right?

Then we had like Stockfish and you know all of these models that got better and better and better to the point now where like at a game that we've been playing for thousands of years, the best entity that can play it is a computer. And not just a computer, but like a model, right? A model that has the ability to play this thing.

Then people are like, "Okay, sure, sure, sure. Chess, but like what about Go?" Like Go's way too hard. there's like too many positions and atoms in the universe and all this nonsense, right? And then Alph Go crushed the best Go player to the point where like it's laughable, right?

Then you go, okay, well, what about these openw world games? Then it's like, okay, well, now we're winning Starcraft, right? And then it's like, all right, what about protein folding? Yeah, okay, it can do that, right?

There is no thing that anyone has predicted that and it would be wild if you think about it to it would be like saying that yeah sure there are aliens on another planet that are smarter than us but they're not going to be better at us than like writing the book. And it's like what are you talking about? Like of course they're they're better at everything.

It's like okay okay but like not better at like Atari games. And it's like what? No. like they're what? Anything that you are good at, this thing is smarter than you. It will be better than you at. And that's the world that we're in. We have aliens that are inside of our planet right now that are smarter than us in very narrow domains and those domains are rapidly expanding.

Yeah. And we are going to get to a point where they're just smarter than us in every domain. And then what the do we do?

Well, yeah. And like you were talking about like the coordination thing on once you for lack of a better word stop making them so like rapidly like ADHD devolving and like can like refocus them and like re recoordinate them. Humans are not great at coordinating this. Like the that's the hardest thing about scale organizations is like like we're really bad at it. It requires like this very horrible modes of communication and like checks and balances. You put in processes. You create committees. Like it's it's the worst thing in the world.

Agreed. So yeah, that's I think like the major unlock for me is like the coordination game which is obviously like this is what blockchain also purports to like help solve, right? Is like remove the coordination.

I think that we vastly underestimate what that results in. I think we haven't I think we understand to a certain extent just how inefficient the coordination is in human land, but I think that there's going to be a tipping point where yeah, so everything just gets completely unlocked once you once you kind of quote unquote solve coordination, whatever that means.

Wild things can happen. And you know again the the thing that we have now is this recursive self-improvement loop that is very human driven right in the last two months the ability to build tools for AI for AI to get better has collapsed in terms of cost and time by 90%.

Any kid who has access to cla code can cook up some scheme to try and figure out how to make these agents coordinate themselves better. So you have tens of thousands of people who could not have previously done this who are now as good as the best engineers who were running these labs 3 years ago building software and tools for these things and there's thousands of people out there.

It's wild. It's really wild and it's going to change everything and that's not even assuming that we get a better soda model in the next six months which we will of course like Claude 5 or Codeex 6 or whatever.

This is the most interesting thing to me. I was talking to a friend of mine about this last night and I was like it feels unlike crypto in the early crypto period where it felt like there were still moes. Moes have kind of disappeared. Your moat is you're a smart person building things and you got a lead on the other smart people who aren't building the right thing and you can build a moat. Anything that you can build can be replicated by the model in like 20 minutes.

Like there are no moes and that I think is going to be the most chaotic thing is realizing that like moes have disappeared at least from like an engineering perspective. Anything that you can conceive of will be buildable in minutes. So yeah, it's going to be wild. Good times.

Distribution it is regardless of this like you know what you know what you can't replicate? You can't replicate pudgy penguins. That's what you can't replicate. You know what else you can't replicate? Apparently not getting scammed because there's a whole other side story to this. We're going to be talking about scams a lot.

Okay, so yes, AI is magical, etc., etc. Back to cryptolandish in true crypto fashion. This all the clawbot is exploding. He had to rename. Anthropic is mad that they he's whatever showing them up. So, he renames everything. Of course, someone grabbed all the old handles, right? Because why wouldn't you? Yeah. There's this thing that's like the future and everyone loves whatever.

Of course the person who did that is a crypto person because why would they not be? And of course the thing that they decided to do which actually I'm not opposed to this as given the available options right you can go fish people you can go try to hack people in terms of what malicious things you can do this is the least malicious but I would prefer people used it for like good faith things like truly good faith things but we'll take it. They launched meme coins or promoted memecoins including the claude memecoin.

There's multiple levels of irony here. One is that anthropic was like oh god I don't want our brand to be like tainted by this because it was like hold my beer. And this only happened because they forced the bro to rename, right? Like if he had it renamed, then you wouldn't have a $16 million market cap memecoin that went like of course.

The other side of the story obviously is well I mean it's just like the most crypto thing ever of course, but like these Claude guys like they are not not anthropic but like the Cladbot builder and like the AI guys, they're smart guys. They're not crypto guys. He's sitting there going, "Bro, I just wanted to build something cool. What the hell did you do?"

The interesting thing about memecoins, I think, right? And like this this whole vector is, when I first got into crypto, I was like, I've been in adversarial environments. I know I know what that's like. And like 20 minutes in, I was like, this is insane. Like there are people trying to kill you just everywhere all the time, trying to steal your money, trying to destroy everything that you've ever done, and they're like one second away from pulling it off.

Then you live inside of that environment for like eight years and you be and you forget that people live in a different way. This guy's like, "I'm just going to change Twitter handles. What could possibly go wrong?" What could go wrong?

It's the most jading part about being here, frankly. It also like I don't know. Again, I'm It could be way worse. I'm glad that they didn't fish a whole bunch of people and destroy people to a certain extent. I don't enjoy the meme combing culture or this like I just I really don't like it. I don't think it's good. I don't think it's valuable. However, at least it's sort of like limited to the sphere of people who like certain not into. You're playing the game.

Agreed. I think like the most lasting sort of impact is like it is such a major turnoff like from the outside or even the people that are like one foot in watching and observing who really don't understand the game and stuff to them especially when they see the charts they're just like okay so wait you usurped this bro's brand after he had been like you know screwed by the big bad anthropic corporation you usurped his brand and then rugged a memecoin on.

Also the problem is for out for for people outside, right? Crypto is undifferiated. Like the the the guy who you know like sniped that handle, launched the memecoin, whatever. Like they're the same guy as like Luca and and Taylor and I. Like you're all crypto people doing bad things, right? It's becomes this undifferiated mass of bad stuff that people don't like.

So I mean look at the end of the day we just need to convince the claw bots that we're not bad guys and then it'll be fine because they're in charge now. So yeah exactly. I had one I had one brief let's try to keep it brief. Okay. We talked about the macro AI we talked about right all this different stuff. Where is crypto gonna fit into this though? Like we are going to like that's the next that's the unlock that I feel like we're kind of dropping the ball on at this point.

We're using it to build tools. We're using it to like coordinate people's management of their Telegram conversations that they can't keep up with, etc. But like what the hell are we doing with wallets, with crypto, with onchain trading like our bots, right? I feel like we're there's a there's there's a big open table someone should probably sit down at, right? Is anyone doing this?

So, my hot take I was talking to probably the person that I think has like the best long-term kind of view on crypto of anyone I know. And he was like AI is just so much more fun than crypto right now. That was like his it's like the innovation. It's like like you said like Defi summer like it's just it's this constant like it's fun there's not a huge amount of risk yet there's a huge amount of opportunity like you just have to show up this output you need to show up and learn things and it's moving so quickly and you're in that like early part of the curve that like you know a like young enterprising autist can really like speedrun a lot of stuff right and I think it feels exactly like early crypto.

To me, it feels like the 2016 2017 era of like Ethereum turns up, you can do stuff now rather than just forking Bitcoin, right? You can actually build software on this platform and it does things and it's like, you know, autonomous like it was wild, right? And this feels like that. I don't know what the intersection looks like, honestly. Like, I'm not sure.

You know, there's something to be said for like agents using Rails, right? Like crypto. That has to be the great intersection. I think that. You know, you can't agents can't go and make a Wells Fargo account or a Morgan Stanley account, right? Be on these payment rails. And in reality, it should out that. I mean, that to me has to be where I think the next bull run is going to come.

I was talking to somebody else about it similar to like I think your point Kane but I was you know I was kind of frustrated as to like where's the crypto bit because stocks are just been vertical for like almost a year now. I mean, yeah, last time we saw stocks like this, crypto was like file quarters at 50 billion, you know, so trying to figure out like what's going on.

The truth of the matter is just like from you, we you take the speculation side of it, it's way more advantageous to go bid a AI on Tradfi that's, you know, building some sort of silicone wafer than it is uh, you know, you know, Filecoin. even though you'd think Filecoin would be biddable in something like this, but nonetheless, it's I think I think the bull run at least for crypto is going to come around this narrative around AI using crypto rails like that I think is going to be the convergence for hopefully our alt season that everyone's been awaiting for for now about five years.

I think we just need to convince the cloud bots that Pangu is the Best medium of exchange. Exactly. Ever. Actually interesting. I I actually I'm going to write that down because I should probably make an effort to like go find one of these guys and be like, "Dude, let's let's make this the transactional currency. I'm going to write it down."

Well, you know, it's funny, right? Because in in the game that I made, Pangu is the currency. I was like, I'm gonna I'm gonna use this as the currency. So, we don't have AI agents running around inside there yet. But soon soon. You could you could.

All right. Before we continue, here's a short word from our sponsors that make the show possible.

The world is about to see one of the largest infrastructure shifts of the century. New technologies are using more energy than ever before, but our legacy grids can't supply the demand, and we are barreling towards a global bottleneck. So, Fuse is rebuilding it.

The energy network is an intelligent decentralized grid that coordinates smart devices to balance supply and demand. The network harmonizes existing infrastructure, increases grid capacity, and unlocks lowcost clean energy. Energy dollar is the native token of the network. The more electricity the world needs, the higher the demand for the energy network, the value of energy dollars may fluctuate. from one of Europe's fastest growing energy startups. Follow at fuse energy onx to find out more.

Multi-chain advisers is an emerging technology growth firm that has helped create over $50 billion in enterprise value for more than 80 clients like Pith, Moon Commerce, and Wormhole. They've worked with some of the largest and most impactful companies in the space. They're the partner you want when you're navigating markets and trying to break out from the noise. They help navigate TGES, go to market, BD and partnerships, capital markets, advisory, PR, media placements, KOA activations, and more. Driving execution from launch to scale. Their results are measurable. To learn more and start building real traction today, visit multi-chainadv.com.

All right. We're back. Tay, if you can introduce this segment, I would appreciate it.

Yeah. Yeah. Yeah. All right. So, this has been like the highlight of my week. So, where do we start?

As most people are aware, the government seizes crypto for all sorts of different reasons. When bad guys do bad things, they seize the crypto. It's a long, long, long process, right? There's like various stages and steps to the seizure process. I think a lot of people don't realize that. You have to like you have to freeze it, you have to like coordinate with the people who have the crypto, you have to like move it. There's all these different things. There's like an equivalent like there's like the onchain processes and there's the off-chain processes.

The offchain processes take forever because it's the government and it's like courts and a bunch of paperwork. So in this case that we're about to talk about, we're talking about funds that were