21 in 21: Calle on Cashu, Clawi, and AI Agents

1 - > Kali, welcome to Presidio Bitcoin and the 21 and 21 show. 2 - > Great to be here. 3 - > This is your first time at Presidio Bitcoin. 4 - > Yes.

5 - > What exactly brings you to Presidio Bitcoin? 6 - > Well, Presidio Bitcoin is probably the best Bitcoin co-working space on the planet. 7 - > So I just wanted to check it out and say hi. 8 - > That is some very nice praise.

9 - > And I feel the same way. 10 - > There are some other nice co-working spaces too. 11 - > But yes, Presidio Bitcoin is special and unique. 12 - > I agree.

13 - > Well, we appreciate you stopping by. 14 - > Okay, so for our listeners that maybe don't know too much about what you're working on, 15 - > although you have done so much in the Bitcoin space, 16 - > I first kind of learned about you when the Cashew DevKit emerged. 17 - > I'm curious, before we kind of get more into the Cashew side, 18 - > what is your kind of Bitcoin journey? 19 - > Well, my Bitcoin journey, I learned about Bitcoin very early and it took me quite a while until I got back to it.

20 - > But so I think the first time I heard of it was probably in 2009 or 2010, the latest, where someone, a friend of mine, explained it to me. 21 - > And I always thought of it as something similar to the Torrent network. 22 - > But instead of sending files, you send around money. 23 - > I didn't quite grasp it and it took me quite a while until I understood how it works.

24 - > So I was just passively observing it and I really dug in deeper once I started being a developer in the space. 25 - > So I started with Lightning, working a lot on Lightning, and that just forced me to understand how Lightning works, 26 - > which forced me to understand how everything on-chain works. 27 - > And so my personal rabbit hole was definitely driven by the development and technical rabbit hole. And only after that, I started to, you know, grasp more and more on economics and just the overall, let's say, monetary revolution that's going on in Bitcoin.

So since then, it's been quite a journey. 28 - > So were you doing too much in Bitcoin before Cashew? I know you were on the kind of, I guess 29 - > you were researching and tinkering, but was that kind of your first big project? 30 - > Not really.

So before Cashew, I spent almost two years working on Lightning itself. So I was 31 - > working primarily on LMBits, which is software that you can run on top of your own Lightning 32 - > node to make it more useful and build applications that use lightning and so that was the biggest 33 - > project that i started working on i had a bunch of like smaller projects myself that i was working 34 - > on one of the things that i started with was a firewall thing a security feature for lnd for 35 - > lightning nodes to add permissions and policies for connecting to other nodes opening channels 36 - > and so on.

So it was mostly based around Lightning. And the reason for that is that, and I think that 37 - > is an experience that many developers have, is that Lightning opened up this door of having a low 38 - > barrier of entry as a developer, because you can experiment with small payments and quick payments. 39 - > That makes it just way easier to get into it. I always thought kind of working on the blockchain 40 - > itself is a bit scary.

So that's what brought me into the space. So from Lightning on, then I 41 - > I started working on cashew maybe like two years after getting really into the weeds. 42 - > It's a good point that you bring about Lightning maybe feeling a little bit more approachable 43 - > because it's not, you know, Coincore itself or the actual blockchain layer. 44 - > I had not really thought about that before.

45 - > Interesting. 46 - > So yeah, so now talking about cashew. 47 - > So you have like a very layman's way to break down cashew. 48 - > Also, the difference between cashew and eCash.

49 - > Well, first of all, let's start with the second question. 50 - > Cashew is eCash. 51 - > eCash, what we typically call eCash, is something called Charmin eCash. 52 - > And Charmin eCash is the first way of really making a cryptocurrency.

53 - > And that is around 25 years older than Bitcoin itself. 54 - > So invented by a guy called David Charm. 55 - > And originally it was intended to be for the fiat system. 56 - > So it was meant to be built on top of the banking system.

57 - > And what it does is pretty easy to understand. 58 - > So basically, the simplest way to understand this is think of how an ATM works in a bank. 59 - > So you have a bank account, and you go to an ATM, you take out cash, and you can use this cash. 60 - > You hold it in your hands.

61 - > It's a bare asset. 62 - > You can use this cash to pay for things. 63 - > And then the cash receiver, the merchant, for example, can go back to the bank and then deposit the cash back into their bank account. 64 - > So that was the idea in the 80s, in the 90s.

65 - > And that's what we call eCash. 66 - > And Satoshi then, basically, Satoshi's big revolutionary idea in an academic sense was being able to decentralize eCash. 67 - > Because previously it was built on top of decentralized systems in the banking sector. 68 - > So Satoshi gave us a new financial system that is decentralized.

69 - > And what we have done in Bitcoin with Cashew and also other eCash projects that built on top of it is to use this new financial system Bitcoin instead of the banking system to build eCash systems on top of it So what Cashew is today is software that anyone can run and it is composed of two components So there Mint that is the server component that you can run on top of Lightning Node 70 - > for example, or on an on-chain wallet. 71 - > And then there are Cashew wallets that can use this Mint to basically withdraw eCash, 72 - > deposit Bitcoin, withdraw eCash, use it in a privacy-preserving manner, very fast payments, 73 - > cheap payments.

74 - > So basically creates a very private, super fast, low fee payment systems for Bitcoin. 75 - > And yeah, that's what we've been working on. 76 - > And Cashew, what's the relationship with Cashew and Lightning? 77 - > So I would call Lightning a layer of Bitcoin.

78 - > So we have on-chain, we have Lightning on top of that. 79 - > And Cashew is, practically speaking, it's a separate system by itself. 80 - > So it's not Bitcoin, but we use it to build applications on top of Bitcoin. 81 - > So it's not a layer of Bitcoin because it doesn't give you unilateral exit properties like a proper layer should do.

82 - > But it makes it way more useful to use a Lightning node underneath it, plug a cache unit on top of it, and then issue eCache, which can be used for applications, for wallets, and so on. 83 - > Interesting. 84 - > Curious, with eCash and Lightning and AI, are you seeing any type of use cases with agentic payments? 85 - > Or kind of what are your thoughts on what the best system is right now for that?

86 - > So, yeah, there's been lots of buzz around agentic payments. 87 - > I've been playing with it myself. 88 - > I've used my agents to make payments. 89 - > I think it is going to, you know, it is part of our future.

90 - > Agents will be making purchases for us or for themselves 91 - > to buy in, for example, services for the agent. 92 - > And we have been experimenting with cashew and agents 93 - > from very, very early days. 94 - > And quite frankly, it was very surprising how easy it went. 95 - > So eCash itself is a bare asset system.

96 - > That means the balance that you have 97 - > is literally stored on the device that you hold in your hand. 98 - > So it could be your phone, for example. 99 - > It's unlike in a blockchain system 100 - > where you have keys 101 - > and they represent some kind of a state 102 - > on a blockchain, 103 - > somewhere on the internet, for example. 104 - > For eCash, you hold the eCash 105 - > on the device itself.

106 - > So for agents, 107 - > it's very easy for agents to understand 108 - > how to use eCash 109 - > because you basically set up an agent 110 - > and what you say is, 111 - > get a cash wallet. 112 - > You wait half a minute 113 - > and then it says, 114 - > I have a cash wallet now. 115 - > What do you want me to do? 116 - > and what you can then do is to send eCash to the agent 117 - > by basically just copying a large string from your cash wallet 118 - > and just dropping it into the messenger 119 - > that you're talking to the agent with 120 - > and that's how the agent receives the money.

121 - > So again, think of how cash works. 122 - > You just hand it over to the agent. 123 - > The agent puts it in their pocket 124 - > and then if they want to make payments online, 125 - > they could either do a direct cash payment. 126 - > There are a couple services that just directly accept eCash 127 - > or make a Lightning payment.

128 - > So as I was saying, 129 - > Cashew lives on top of Lightning. 130 - > So any Cashew wallet is also a Lightning wallet. 131 - > So an agent can go out there 132 - > and use the eCash similar to any Lightning balance 133 - > that they might have. 134 - > Interesting.

135 - > Yeah, we just had a hackathon, 136 - > Spiral just sponsored and we hosted that PP 137 - > and it was all about how, 138 - > thinking about agents as your customer 139 - > and how they can earn money. 140 - > And it's interesting. 141 - > Yeah, it's a kind of universal problem right now 142 - > with stablecoins and lightning, 143 - > and it's all just hard agents transacting money. 144 - > So it's very interesting to see.

145 - > We'll be excited about it. 146 - > On that note of AI, 147 - > how are you thinking about the intersection of Bitcoin and AI? 148 - > And how are you currently using AI and your work? 149 - > So most of my work is development of Bitcoin software, 150 - > Bitcoin protocols, and so on.

151 - > And so as like almost every single developer out there, 152 - > I've been using AI very heavily in my workflows. 153 - > So in that sense, it has a massive impact 154 - > on my day-to-day development work. 155 - > And I'm pretty sure that is almost like the case 156 - > for almost every Bitcoin developer out there. 157 - > And I think the other part is the authentic payment world 158 - > or that whole category of agents being able to pay each other.

159 - > I think Bitcoin lends itself for especially that world where you have lots of agents being interested in many services out there. 160 - > And therefore, they need some kind of a micropayment system in order to use those services. 161 - > So think of it how it works today with subscriptions, right? 162 - > So you can have subscriptions for Netflix, Spotify, and so on and so forth.

163 - > And that limits the number of services that you can use as a person because you're not going to have like 500 subscriptions, right? 164 - > But there's a natural limit of how much you can use. 165 - > And I think that's a very limiting factor, especially for agents. 166 - > So agents will want to use a service just maybe once, maybe twice, and never use it again.

167 - > So the subscription model doesn't really work. 168 - > And for an alternative, you need micropayments. 169 - > You need something that just works with low value very fast and no commitments basically And I think for that Bitcoin is ideal because it universal internet money that anyone is interested in receiving Agents can use it very easily And you can construct a world where you have a multitude of services that you can use online without having to make a subscription and keep using it 170 - > You can just have a single shot service, like a search, like an email service, maybe bypassing a captcha.

171 - > It could be anything that the agent is interested in in that moment. 172 - > So I think Bitcoin is internet money. 173 - > It's useful for humans. 174 - > It looks way more useful in that sense, 175 - > like more directly useful in the payment sense for agents.

176 - > Interesting. 177 - > Yeah, that is a good point too, 178 - > that the way humans maybe think about subscriptions, 179 - > it's not as useful or interesting for agents too. 180 - > That's right. 181 - > Yeah.

182 - > Well, quickly changing gears. 183 - > So another thing that I know you work on and you're known for 184 - > is BitChat on Android. 185 - > Curious what drew you to that project 186 - > beyond Jack starting it 187 - > and also kind of like 188 - > what are you working on right now 189 - > in terms of it? 190 - > Yeah, I learned of BitChat 191 - > at the same time, 192 - > like in the same moment as everyone else 193 - > when Jack released the iOS version.

194 - > I thought it was very interesting. 195 - > BitChat is a Bluetooth-based mesh messenger 196 - > that works without the internet. 197 - > So messages travel from one phone 198 - > to another, and then it builds this mesh network. 199 - > That means messages can hop from one phone to another, traversing wider ranges.

200 - > And for example, in a campus like this, if we had a couple few dozen people 201 - > running it on their phones, we could send around messages across the campus. 202 - > So that's it's incredible, interesting technology by itself. 203 - > Jack, about the iOS version. 204 - > And this will sound very simple.

205 - > The reason why I basically built the Android version is because I'm an Android user. 206 - > I wanted to use it. 207 - > There was no one else working on an Android version. 208 - > So I jumped on it, built it.

209 - > It took me a couple of days to release a version that was working, heavily relying on AI. 210 - > And that's how I basically started working on it in tandem with Jack. 211 - > And since then, we've been working on the protocol itself, adding more and more features, 212 - > making it more advanced, being able to send voice messages, images, adding gossip, 213 - > features to it or sync features. So if you have been like off the mesh for a while, then you'll be 214 - > able to receive all the messages that were sent while you were offline or offline, not in a mesh.

215 - > And recently, what I've been working on is source-based routing. So it's similar to how 216 - > Lightning works. You want to send a message from one BitChat device to another BitChat device. 217 - > There is a gossip system where all the BitChat devices tell their neighbors who they're connected 218 - > it with.

And you can use this information to then route a message one hop after the other to reach 219 - > someone specifically, for example, you know, across the campus out there without having to flood the 220 - > entire network. So similar to what we do in Lightning or similar to what we do in Tor, 221 - > we improve the efficiency of the protocol. And so, yeah, that's basically what's on my mind right now. 222 - > The things that we've been doing recently on it is mostly security focused.

We've had a couple of 223 - > reviews of security researchers. 224 - > Very happy and lucky about that, 225 - > that lots of people are interested in helping 226 - > us improve the system 227 - > because over the last 228 - > couple months, there's been lots of 229 - > attention on BitChad, especially 230 - > in the political context and 231 - > the context of activism 232 - > where people 233 - > rely on it in case of 234 - > internet outages or surveillance 235 - > on the internet, where 236 - > a mesh network like BitChad offers 237 - > an alternative system 238 - > where the entire hardware stack is owned by the people.

239 - > So because we want to make sure that even in those cases, 240 - > it is secure and safe to use, 241 - > that's been our focus in the last few months. 242 - > Interesting. 243 - > That's very cool that you were able to build it so fast. 244 - > And I remember seeing it.

245 - > It was like, yeah, a few days later, like you said, 246 - > and that is interesting that AI let you build it so quickly. 247 - > Yeah. 248 - > Amazing that now it's on Android too. 249 - > besides BitChat and Cashew 250 - > are there any other rabbit holes 251 - > that you've been going down recently?

252 - > So Cashew is my main focus 253 - > lots of time and work 254 - > also on BitChat 255 - > but I have one more bigger project 256 - > that I work on 257 - > is Chloe.ai 258 - > C-L-A-W-I 259 - > and it's a system 260 - > to make it very easy 261 - > to run and deploy AI agents 262 - > for personal use 263 - > or for organizational use 264 - > So Chloe is a hosted agent platform where you can sign up and with a few clicks of a button, you get an OpenClaw agent or a Hermes agent that you can connect to your Telegram, Slack, Discord, whatnot, and then start using these agents.

265 - > I find this new generation of personal agents extremely fascinating. For me, it's a step, a new step in the AI arch that we've been on for several years now. 266 - > I think of it as, you know, I like to say ChatGPT was level one, 267 - > where you interact with an AI, not an agent, 268 - > but with an AI chatbot basically through the browser. 269 - > And then what I call level two is when we developers started using 270 - > real agents now inside our own computers to help us develop And those are like on demand and they do the work for you when you ask for them But once you close your laptop they gone 271 - > And when I first discovered OpenClaw 272 - > and started using it, 273 - > it really felt like a new chapter 274 - > in that what I call level three 275 - > is the agent that is always available, 276 - > always online, 277 - > can do tasks for you 278 - > while you're not babysitting it.

279 - > And also be proactive 280 - > about things that you tell. 281 - > For example, once there are updates about the weather in San Francisco, 282 - > send me a DM with the most interesting events happening in downtown, for example. 283 - > So that is something that just runs in the background always for you 284 - > and can really make your life easier and help you achieve tasks that you weren't able to. 285 - > And so this was for me a very significant moment, I must say, 286 - > since I think of it as significant as the introduction of ChatGPT.

287 - > and so because I myself found it extremely useful 288 - > but at the same time super hard to set up 289 - > we started building a platform 290 - > that makes it easy to run 291 - > and secure to run for ordinary people 292 - > and that's basically has been my goal 293 - > across all my different projects 294 - > is to make these very interesting 295 - > but still very complicated technologies 296 - > accessible to ordinary people 297 - > because I think they really only shine 298 - > once they're in the hands of as many people as possible.

299 - > And yeah, that's what I've been working on 300 - > in the AI space directly. 301 - > I totally forgot that you did Calli 2. 302 - > I set one up like right after you came out with it. 303 - > And it was super easy.

304 - > And I did use it in like a hackathon day 305 - > that we had here. 306 - > And I would say that I felt that it was better. 307 - > I mean, I know it's totally model dependent, 308 - > but I felt that it was better at prompting 309 - > versus using like ChatGPT 310 - > or getting kind of the code the way 311 - > that I want it to be. 312 - > So thank you for making that much more 313 - > accessible to set up an open Klawi agent.

314 - > I appreciate it. 315 - > I'm kind of curious, 316 - > how much adoption have you had 317 - > with Klawi and have you kind of seen 318 - > as the user base, 319 - > mostly kind of people within the Bitcoin ecosystem 320 - > or has it expanded, do you think? 321 - > No, it's very diverse, actually. 322 - > It's not so much focused on a Bitcoin ecosystem itself.

323 - > I think that also makes sense. 324 - > There's a small overlap of these spaces of the AI space and the Bitcoin space. 325 - > That overlap is growing as you can see that in the Bitcoin space, 326 - > basically everyone is thinking or talking about AI. 327 - > That is great.

328 - > For Claw itself, we have a very diverse set of users. 329 - > That's not so much Bitcoiners. 330 - > And there has been incredible initial hype in the beginning 331 - > due to the popularity of OpenClaw. 332 - > and Hermes is another very popular agent that we offer.

333 - > And so there's personal AI agent use. 334 - > But in the last weeks and months, 335 - > I've been getting more and more feedback from organizations themselves 336 - > because right now, every company, every organization out there 337 - > asks the question, what is their AI strategy for the future? 338 - > Because it vastly improves productivity 339 - > and the ability to communicate with your team 340 - > and either make everyone in the organization more efficient 341 - > or help groups of people coordinate on topics.

342 - > So one example is an agent for your organization's knowledge base. 343 - > So you could have lots of documentation about the processes in your organization and so on. 344 - > And being able to drop an agent into that knowledge base 345 - > and then giving everyone in your team access to that agent 346 - > makes it very easy for people to gently find information 347 - > and discover it in a conversational style. 348 - > So it's been really interesting to observe that.

349 - > And the second point I want to make real quick 350 - > is using agents within your organization 351 - > as employees, basically. 352 - > And that is especially useful for computer-related tasks. 353 - > I've been having incredible results with DevOps agents 354 - > that can manage servers 355 - > or customer support agents 356 - > that can read through emails 357 - > and create answer with drafts. 358 - > And so there's, I think, 359 - > especially in the organizational 360 - > and for companies, 361 - > I think we're going to see 362 - > a lot more uptake in the next year.

363 - > I definitely agree, 364 - > especially being a block employee 365 - > and seeing Jack be so visionary 366 - > about all that too. 367 - > It's all changing. 368 - > I have lots of more questions, 369 - > but we're out of time. 370 - > so I'll ask some to you out there.

371 - > But yeah, really quick, 372 - > if people want to follow more of your work or you, 373 - > where should they go? 374 - > Well, you can find me on X, Kali BTC, 375 - > or you can find me on Nosler. 376 - > It's Kali at Hashi.me, my NEP5.

377 - > And if you want to reach out, please do so. 378 - > I'm a big advocate of open source development. 379 - > I advocate as much as I can for people to join us, 380 - > work together with us on Bitcoin. 381 - > We need every single developer that we can find out there.

382 - > And if you find anything interesting that I just said 383 - > or anything else in the Bitcoin space 384 - > where you might need any guidance 385 - > or just want to exchange some ideas, 386 - > hit me up. 387 - > I'm always happy to talk and connect. 388 - > Awesome. 389 - > Well, thanks so much for coming on the show 390 - > and I hope you enjoy the rest of your day 391 - > at Procedure Bitcoin.

392 - > Thank you very much for having me. 393 - > Of course.