Why AI Agents Want Bitcoin | Bitcoin 2026

Añadido: 2026-05-10

[00:00:00]Um, all right. So, we're here to talk about uh one of the most exciting emerging use cases for Bitcoin, which is do you want to give it to your AI agents? Uh, potentially give them sovereign spending authority. Uh, so we're going to talk about three rounds uh with some open exchange at the end.

[00:00:15]Uh, got a great panel here. We'll all introduce ourselves and then just kind of uh dive in. Uh, so my name is Eric Hadley. I run a website called Hyperdope. It does lightning gated video. I also do some things with some agent services. I run a site called l402.directory. directory that lists out different health checked uh services that agents can actually buy today. Um, so that's me. Uh, Christopher, do you want to introduce yourself? What you got going on? Hey guys, Chris. I run Open Agents. We are an open-source AI lab building on the Bitcoin stack, proudly based in the Bitcoin capital of the world, which we all know is Austin, Texas. That's right. Thank you very much. And uh one unique thing about open agents is um we know that everyone in this room has compute that you're not using that we'd like to pay you Bitcoin for because we figured out a way both to collect it from you and to have agents use it for fine-tuning inference and stuff. So um our goal at this conference is to run the largest decentralized model training run in the world. As soon as we cross the 200 node threshold, we will steal the world record from shitcoin company BitTensor. So, um, help us out with that. Openagents.com, click the button, pays it to your agent, it'll spin up a pylon, and hopefully if you've got a Mac or computer with you, we can get you some Bitcoin earned by the end of this talk. So, thanks.

[00:01:40]>> Yeah. Cool.

[00:01:42]>> Eric.

[00:01:43]>> Hey, my name's Eric. I'm a design engineer. I work on Cashew which is a bearer asset uh mechanism on on Bitcoin where if you hold the the string of text or the data you have Bitcoin. Um lately we've been seeing kind of an rise of interest in Cashew from some unexpected places from Bitcoin companies that want to use it to administer payroll to non- Bitcoin companies that want to use it internally to transfer money uh and increasingly uh people that want to use it for agentic payments and that's what we're here to talk about today.

[00:02:13]Hey everyone, I'm Roland from Albi. We build products, tools, and services to make it easier for individuals and businesses to use Bitcoin, which we think is the global currency both for the web and for everyday usage. Uh so recently we've been building a lot of AI tools. Uh we have three different AI skills. one for payments to give your agent uh the ability to connect to a wallet and make and receive payments.

[00:02:44]One for building uh applications that have Bitcoin in them. So just with a single human language prompt, you can f you can build a fully enabled Bitcoin application. And finally, we have a a skill that enables you to your agent to run uh a fully self-custodial manage a self-custodial node. Uh and apart from that, we've also been looking a lot at a agentic payments. So, our payments skill for agents has has the ability to make aic payments for the three main protocols which is L42, X42 and MPPP.

[00:03:26]Gotcha. Awesome. Thanks, Ron. Um, all right. So, we're going to start with kind of round one. Uh, we're going to pop up. Uh, let's just ask kind of a a general question. We'll hear from everybody. Uh, Christopher, we'll start with you. Uh, do agents even want money?

[00:03:39]Uh, do they find it useful? And then among the options, like is is Bitcoin the right option?

[00:03:46]>> I'll start with like a pedantic philosophical point, which is that like agents don't want anything like this is a piece of software. There's certainly like a movement among certain big labs anthropic to anthropomorphize the AIs and make you build like an emotional connection with it. What is the agent?

[00:04:03]What does the AI want? Oh, the AI doesn't want to be sunset and let's like have model therapy like some like psychotic things in my humble opinion.

[00:04:11]So, so, so the agents don't want anything. At the same time, you know, if you look at the models and the the the data that they've been trained on, you know, the Bitcoin Policy Institute had a great paper where they ran some studies about like if you put certain prompts in front of certain models, turns out that they prefer or it expresses a preference for Bitcoin relative to other things.

[00:04:32]And largely that's a function of the data that it's trained on. So we've got like 15 years of people trolling and saying stuff on Reddit and like we've got more content because we're just older. So there's it's weighted more heavily. But that's all a function of what data gets into the models. You can envision models that are trained on not that. There was a project a couple years ago called Spirit of Satoshi that trained specifically on just Bitcoin stuff. And they did fine-tuning. So they kind of started with the base model then tried to make it more of it a bitcoiner.

[00:04:58]But if it's all a function of the data that's trained on, this is kind of our argument for why we want to start training our own models from scratch is what if we don't feed it the like midwit Reddit tier like shitcoin analysis in the first place. What if we teach it like Austrian economics from day one?

[00:05:12]But I'll stop there. Go ahead.

[00:05:14]>> All right. Yeah, Eric, I guess your thoughts on the topic and maybe you can say a little bit about like what is the UX like today maybe without Bitcoin?

[00:05:21]>> Yeah. So, do agents want Bitcoin? I think uh starting there is agents don't want anything people humans want to use an agent to pay for something, right?

[00:05:31]So, that's kind of the agent just wants to complete that instruction and execute the payments. And right now, what does that look like? So if you want to make a payment with an agent right now, the current user experience is tied to human identity in the human world. You either have to feed it a credit card uh or link it to an API which you've already kind of authenticated your human identity to and then it's topping up and that's kind of the current experience and then you can have the agent perform the the call or the function or whatever the case uh may be. And what I think is more novel or more interesting or more unexplored is the permissionless aspect of having an agent perform a payment without an identity because these things are not tied to the human world or a permissionless agent that you spun up, you know, 5 minutes ago doesn't have a human identity, right? It's just this piece of software. So, how can this kind of in the current landscape that exists now go on and execute these payments without a human identity, without KYCing, without having to need permission? And that's where Bitcoin I think solves that problem. And there are many different ways to pay with Bitcoin as we know. There's onchain, there's lightning, and there's cashew, which is what I work on. And I'm kind of especially kind of interested in the cashew use case because it reduces a lot of that UX friction. If you want your agent to pay with lightning, for example, and if it's self-custodial lightning, you need to have inbound liquidity, outbound liquidity, you need to have an open channel. You need to trust the agent's logic to not spend your entire Lightning wallet balance. If you have $200 in there, you're going to tell the agent, "Okay, perform this, you know, for $20 or whatever the case may be. You're going to trust the LLM not to hallucinate and spend your entire balance." Cashew kind of flips that on its head because with Cashew, it's a custodial system with great privacy. You don't have to deal with the uh the onboarding friction of custodial lightning and the the spend is topped off by how much e-cash you give it as a ver asset. It has $20 worth of ecash. It can only perform $20 worth of payments, >> right? Yeah, that makes sense. Uh Roland, I guess your thoughts on that same question and then why would Bitcoin make more sense than some of the competitor Rails like X42?

[00:07:44]>> Welcome to predict. The world is a market.

[00:07:48]Everything is a market.

[00:07:52]Get a 100% cash back up to $100 on your first predict bet if it loses.

[00:08:05]Predict where everything is a market.

[00:08:12]>> Yeah. I' I'd actually like to build a lot on uh what Eric said. So I'd like to ask a question. How many people have run like an open claw or any sort of agent before?

[00:08:22]>> Cool.

[00:08:22]>> Lot of hands.

[00:08:23]>> Um so the first thing I did when I when I set up an agent was I I wanted it to be my personal assistant and I asked it, okay, set up your own email address. And I I thought this was a very simple question, right? But then it thought about it for a while and it came back to me and said, "Okay, you need to go through these steps to give me an email address." And me, it's my personal assistant. It should do the work. It shouldn't be giving me the work, right?

[00:08:51]So this is one of the things that Bitcoin is the permissionless money solves. So instead we can use something like len email or there's another one called agent mail where basically making a micro payment you can give your agent the ability to send and receive emails and it's not just limited to emails but any service that requires KYC or captures these are completely agent unfriendly right um so I think this emerging agent economy that we've seen this year has shown actually how closed down the world actually is the internet actually is it's locked down that users have to provide a phone number they have to KYC to access these different services so why I think Bitcoin is different than the other stable coins or other options for agentic payments is that Bitcoin is truly permissionless no one can stop it uh stable coins in the near term may be more popular but we see every day that stable coins are being frozen just a few days ago over 200 million worth of stable co coins were frozen. So if your agent's money gets frozen, then it cannot actually complete the task that it needs for the human.

[00:10:05]Therefore, it becomes basically useless.

[00:10:08]This doesn't happen in Bitcoin.

[00:10:10]>> Yeah, it makes total sense. All right, thanks. So the the second round, let's get a little more practical. So how do you actually give an agent uh Bitcoin?

[00:10:18]How do you set them up with Lightning?

[00:10:20]Uh do you like type in your seedwords?

[00:10:22]Uh Eric, what do you what's the what's the first step so that people could take?

[00:10:25]>> Yeah. So, this is where I'm I'm going to shamelessly show cashew and eCash because in order to receive eCash, you have to exist because that's the only requirement. It's literally a piece of data or a string of text that you can push through any rail. It's rail agnostic. You can do it through Bluetooth, through email, it doesn't matter. You're just moving a piece of data from here to here.

[00:10:49]>> String of text.

[00:10:50]>> It's a string of text. That's all it is.

[00:10:52]Right? So when you think about now just extrapolate that further to how agents might interact. Agents no longer have to spin up a lightning node. They don't have to know anything about Bitcoin but they all they have to do is communicate from A to B and just push a string of text to each other. And that works at scale and also it works for subscent payments which is what a lot of API calls and kind of um you know a lot of payments that you might be doing are fractions of a cent. Cashew has no problem executing that with with SATs.

[00:11:22]You could do this with with Lightning, but then there's routing fees and you know that it depends on on all these factors. So I would say the the receiving aspect of uh that that Cashew unlocks is uh very perfect for agents.

[00:11:36]>> Yeah, it makes makes total sense. So So that's the Cashew piece. Obviously there's a little bit more of a trust layer there, but let's say if we're going to go to like Lightning Roland, maybe you can jump in with what's the the the Noster uh wallet connect or uh how does that all work? Yeah. So, Nostra Wallet Connect is an open protocol. You don't have to understand Nostra at all.

[00:11:53]It's it's kind of just a technical term, but what it means is it gives seamless access for an application or an agent to access a lightning wallet. And that lightning wallet can be custodial, it could be a cashew wallet, or it could be a self-custodial wallet. So, at Albi, we have two options. We have one where you can give an agent uh permissioned access budgeted access actually to the human self-custodial wallet. So say you have a million sats on your self-custodial wallet but you can give your agent 10 million uh 10,000 sats budget per month or per day uh so that you stay in control and the other option is actually you can get your agent to spin up its own self-custodial lightning node and yeah you have to you have to give trust in your agent that it's not going to spend it all. Uh but on the other hand this agent has the keys to its own money. It does not have to trust anyone. It's in full control.

[00:12:53]>> Awesome. And uh Christopher, how do you do your setup? What do you think about do you do budget controls or how how do you look at it?

[00:13:00]>> First day, it's definitely a beautiful part about Bitcoin because it is permissionless. It's very easy for companies building on the Bitcoin stack.

[00:13:07]Um Albi, Cashew, the money dev kit guys, they can all put out these like little toolkits that are MCP tools or skills, like the kinds of things that you can just literally feed to an agent. there's generally no account to sign up for or if it is like you know get an API key very easily and then all of the the many different trade-offs that you would want between privacy between security between custodianship like that's pretty much covered like we have that covered like a bunch of those different kind of trade-offs and all of that can and should be I think largely seamless to the user at least the kind of like more mainstream user base that I'm target trying to target so we're trying to try to solve this kind of at the product level which is you have a piece of software that you install from open agents called autopilot. It sits on your desktop. It has a built-in Bitcoin wallet that we're kind of shipping using Spark wallet to start, but that should be able and can be able to if you want to interact with your friends or your neighbors or your business partners or someone you're buying from's agent and their agent really prefers Cashew, that should just be a negotiation between agents. You shouldn't have to care about that. um we can have all of like we don't need to choose between these different providers basically like it can be an all of the above kind of strategy.

[00:14:20]>> Awesome. Um all right so we're going to go into uh the the third round of this.

[00:14:23]So so this is about the agent economy the emerging agent economy. What what do agents actually want to spend sats on?

[00:14:31]What are they looking for out there? Um Roland, why don't you start us off?

[00:14:35]>> So I I think agents still act on behalf of a human, right? So a human is trying to build something or trying to achieve some goal. Uh so normally the the agent can can pay for that with Bitcoin instead of having the user have to manually spend time, right? So it's it's basically an automation to save time. Um so that's one thing. There's there's a huge u there's a lot of different indexes like lightning. directory and also forward to index um which has a whole bunch of these uh APIs that agents can use and rather than signing up paying for subscriptions using credit cards they can pay per use uh for these different services. So one I really like is PPQ which is basically you can pay for AI inference. So what I gave my agent was that it pays actually for its own VPS. So it it lives in the cloud on its own machine. It pays for its own hosting once per month and this is actually uh automatic uh subscription payments powered by Nostra Wallet connect. And the other thing is it pays for its own AI inference usage. So every single request uh it can pay to PPQ uh to be able to actually operate and sustain itself. And it can also do uh request to PP ppq uh pay per request to generate images or to request data from the Twitter API and other things. And I I think the Twitter API is actually quite a good uh example because normally you would need a subscription to be able to be able to crawl this data if if you want to get in insights of what's happening um in in different topic areas when you're when you're building or you're doing marketing. Um, but instead your agent can just pay just for the data it needs. It doesn't have to pay for a subscription. I I think this is really cool.

[00:16:35]>> It makes total sense. Uh, Eric, what's your thought on the economy? Any good stories?

[00:16:39]>> Yeah, I would say the the two things that I'm seeing the most of, if I'm being objective, are 402 uh, and using uh, buying LLM compute. And in the 402 space, I will say that the the X42 protocol that the Coinbase and the base guys are pushing, um, I I don't ever want to badmouth any project. I think it's like genuinely impressive and from everything that I've seen the developer experience of that is is quite good. Uh but at the end of the day like like Roland said uh it's dependent on stable coins and stable coins can be frozen. Uh they're not uncensorable. So for me I'm interested in agents paying for things that are in a little bit of a more adversarial permissionless identityless environment. And that is kind of an un a lesser explored design space. And that's something that I'm kind of more interested in is 402 but uh permissionless and uncensorable. The other thing the LLM compute um I think PPP uh is a is a great example. Uh another service that I'm seeing that I also think is really cool is Routster which allows you to purchase uh um LLM credits or compute very similar to open router if anyone knows that but the authentication is a a cashew string or a cashew token. So instead of an API key that you have to authorize with open router, you just generate uh cashew uh and then you use that string, that piece of data as the key to buy the compute.

[00:18:00]So if you generate $50 worth of cashew or bitcoin that is now your API, it's your access and it is your credit. So you now have $50 worth of compute uh that you can access all these LLMs, no no identity, no permission. Um, and that to me is uh is is very cool, is very novel, and that gets me pretty excited.

[00:18:22]>> Amazing. Yeah, Christopher, uh, your thoughts on the economy, and we've talked mostly about agents spending on things. Can can agents earn?

[00:18:30]>> For sure. And, um, this ties also into my answer to the same question, which is like what's the most exciting thing that an agent can purchase? And I I think it has to be compute. Compute is the limiting factor to why we don't have a billion or 10 billion Bitcoin agents tomorrow. There's no compute. If they had the compute, then you can run everything else easily on top of that.

[00:18:54]Um, if we're needing to compete against the hyperscalers and all the VC investment and all of the soon to be public capital money going towards helping OpenAI buy uh GPUs from Nvidia and prices keep going up, then that may limit our ability. But if you take a look at the compute mix that OpenAI has, they said that at the end of 2025, they have about 2 gawatt of compute in their network at the end of 2025. Well, if you total up all of the spare, what we call stranded compute just on people's desktops times everyone in the world, that's 20 gawatt. Can we use that compute for edge inference to train models to do embeddings the kinds of things that are needed in agentic workflow? Well, we've proven that yes that you can. How are you going to do that? You're going to run a piece of software. It's going to run a Noster NIP90 service provider.

[00:19:50]That is you're going to say this is our software called Pylon. You turn you you you install it. You turn it on. You click the button. It's going to say, "Okay, you've got a subset of your GPU that you're willing to sell to the highest bidder at a certain price." Now, the bidder right now, the only buyer is open agents. We've distributed like we've paid like for a million and a half sats worth of compute over the last two weeks. We want this compute to train our models. But you can envision months down the road if there are agents that need compute, they're going to be bidding against us. So, we want your compute to be bid for by a lot of AI agents.

[00:20:30]>> All right, sounds good. So, we're actually uh coming short to the the close of time here. So, um why don't we think about like what's a takeaway uh for the for the audience if they wanted to jump into this world today, later this week? Um what's the what's the first thing they should do? Rowan, >> definitely if if you're not running an agent yet, I I definitely recommend you to to run it. Uh I think there's there's very uh simple options. Uh Kelly, the Chlo guy, right? He he's Yeah, he launched a service called Chlo.ai and just with a credit card, you can set up an agent. So, uh super easy. If you already have an agent, then try and get it to do something without your help.

[00:21:13]And if it gets stuck, then see if there's a Bitcoin service that it can pay instead to get unblocked without you having to do the work for your agent.

[00:21:24]>> Cool, Eric. Yeah, I would say I want you guys to experience what it's like to receive Bitcoin without a wallet. So, come find me. I'm going to go backstage.

[00:21:32]If anyone here has Bit Chat, I'm going to start dropping Cashew uh Cashew strings of text on Bitchat so you can receive them in Bluetooth. I want you to experience what it feels like to be able to receive Bitcoin uh through Bluetooth, through Signal, through WhatsApp without a wallet and then just think further about that for agents because that to me is the big unlock. That's how it should feel.

[00:21:53]>> Cool. Christopher, >> the first thing you should do is sell your spare compute to Open Agents. Let us pay you Bitcoin for it. Um, openagents.com. Whether you have a computer with you today or when you get back home, openagents.com. It's literally a button you click and paste into your agent. If you don't have an agent, uh, I'd recommend the Hermes agent. Uh, open call is okay. We're going to interoperate with a bunch of those. Uh, and it will set everything up for you. Uh, we've got a few people from our team. Open agents guys. Raise your hands. Uh, if you can want help troubleshooting car in the back. Um, you have compute. We want to buy it from you and then do cool stuff with it.

[00:22:32]Okay. Take our Bitcoin. We want your compute. The end. All right. I think that's probably where we should stop.

[00:22:37]We're just about out of time. Thanks so much, everybody. This was awesome.

[00:22:46]Every year this community comes together to celebrate, to debate, to build what comes next.

[00:22:58]And every year the stage gets bigger.

[00:23:05]Sound money center stage. So, where do you go to celebrate the next chapter in Bitcoin history?

[00:23:15]You come home.

[00:23:18]Nashville, July 2027.