# Agents Will Run Their Own Businesses: Inside Laso, Agent Payment Cards, and the Agentic Economy

> Shoal Signal episode. Guest: Hunter Monk (Laso). Host: Gabe Tramble.

- Publisher: Shoal Research (https://shoal.xyz)
- Published: 2026-07-01
- Video: https://www.youtube.com/watch?v=S-eztFEkTP8
- Canonical: https://shoal.xyz/research/episodes/inside-laso-agent-payment-cards/
- Keywords: agentic economy, AI agents, agentic payments, crypto, privacy, Agentic Payments, interviews

## Transcript

**(0:00)** In the agentic economy, there's so many different players tackling the same type of issue, but from different angles. And the end result really is how can you get an agent to do something and then pay for whatever that is. And we've seen different types of solutions here. Ultimately, the goal is how do you go from an intent to work to payment? In each part of that stack, there's different players working on different problems and challenges. And today we have Hunter, which is the founder of Lasso, allowing agents and even humans to use one time use credit cards to pay for things on the web like groceries, gift cards and other financial services.

**(0:44)** Essentially going around the bottleneck of merchants not being able to support crypto and stable coin payments. So Hunter, you guys started with cards for humans and then realized that your product fits uniquely with agents and now you're giving agents cards. How did you go through that process and why did you start with crypto enable or stable enable cards in the first place? Yeah, great question. So, you know, the core part of the business they've been building for kind of the last three years is enabling crypto consumers like humans to get cards and be able to spend their crypto without any ID upload required.

**(1:22)** So they can maintain their privacy, which is a very important product. And that's kind of the core of the business. It became clear really, you know, a year ago, at least maybe a year and a half ago, that a big benefit there is because that onboarding is so seamless, you don't have to go through a liveness check or scan your passport or anything, that it's the only way an agent could get a card. But a year and a year, a year, a year and a half ago, there wasn't OpenClaw.

**(1:55)** And so I kind of knew that was going to happen at some point, but there just wasn't a really good entry point. You could kind of maybe try to tell ChatGBT to do it, but it didn't have, you know, its own wallet. And then OpenClaw hit and it was just, okay, this is perfect. This is what you've been waiting for. And so now you can send your agent a single prompt and it can get its own lasso card. It can go spend it anywhere prepaid is accepted, you know, online. And it can make its own purchases. Nice.

**(2:21)** And your card is unique because it's a one-time use card, right? Or you have a couple modalities for the card. Can you explain why that is the case or, you know, what is the benefit of a card that kind of turns over? Yeah. Yeah, it's a – you can use it multiple times, but you can only load it once. So you get your card and you can get it up to $1,000 from the card. And you kind of spend it down. And once you're done spending it, you get a new card. You don't deposit again into it.

**(2:49)** And the reason it's set up that way is it's a requirement of the regulatory structure that lasso operates under. Essentially, we need to prevent money laundering. That is, like, a big concern, of course, maybe the biggest concern for lasso as well. And so part of that is not allowing reloadability of cards. That's what allows us to have this much lower user verification process to make this kind of model possible and be able to give your agent a card without having to go through a big invasive KYC process. Nice.

**(3:23)** Nice. And you guys first started with, like, human cards, right, which is kind of a weird thing to say. Normal cards. And what are the use cases? And, you know, why were people quick to really love the product that you guys built and put together? Yeah. Yeah. And to be clear, we're absolutely continuing that business. That is the core of the business is humans. Humans, we love you. And the value proposition there is it is the most private way to spend your stable coins.

**(3:57)** So, you know, Satoshi envisioned disintermediating the banking system. But today, you have your crypto and you want to go spend it. You can spend it at businesses that accept crypto, which is it's growing rapidly, but it's still a very, very tiny percentage of merchants. Or you're sending it to an exchange, like a bank-like entity, essentially. And maybe you have, like, a coin-based card and you can spend it that way. Or worse, you know, you send your assets to an exchange, sell to USD, wire ACH to your bank account and pay for a credit card. And at that point, not only are we not disintermediating banking, we're just doing banking with extra steps. And we've lost our privacy. It's not kind of the purpose of crypto or why we're working in this for the long term. And so satisfying that need for people who want to maintain their privacy in a compliant way to do normal consumer discretionary purchases, those are kind of our users. So they're ideologically aligned.

**(4:57)** They tend to be OGs, for lack of a better term. And that's the kind of vertical that we're serving. Nice. Yeah, I was literally going to use the word ideology. And the product is basically like a manifestation of self-sovereign spending. And the card is the ability to do so. That was great. I'm going to clip that. That was a great way of describing it. That was going to be in our ads. Awesome, awesome. So, okay, so you guys built this card and people are using it and you designed it in mind for like day-to-day use. So just like a consumer card. And is that kind of the main use case that you're seeing? Has that actually changed and you've seen different types of behaviors with the cards? Like what's kind of the core use case, either if it's not that, outside of that? Yeah, the core use case is consumer discretionary purchases. So, you know, our top merchant is Amazon, Uber, DoorDash, Walmart, just kind of regular what you spend on in an everyday.

**(6:03)** And that's like, that is on the human side. I know like a lot of our focus is on the agentic spending side. But on the human side, it's just typically what you make your purchases with. Because you can spend them online or add them to your device or Apple Pay and Google Pay and get gas or whatever. On the agentic side, our cohort of users, non-humans, those purchases are, we're still, we're early, right? We're in the cross in the chasm book. We're in that 1% tinkerer phase really quickly. I think everybody agrees that's going to become mainstream, but we're still early. And so the very first agents that we saw on the platform are doing kind of like silly examples, like buying snacks for the office or whatever. Essentially things that are, it's not much better. And in my opinion, it's a worse experience than just buying it yourself. Getting the snacks to the agent is not like a big time saver. But what we're starting to see, and what I'm very, very excited about, is the agents using cards as a tool to run their own business.

**(7:11)** Pulsia right now is kind of getting some product market fit on enabling the creation of businesses. Locus, our partner, where Locus, where Lasso is the default guard provider, are really running with this idea. There's all these businesses out there that were either not worth the time to explore or to simply be more possible without agents. And now people can do them. So the use cases that I'm excited about are agents are getting cards to buy domains, website hosting, buy ads, and making businesses run. That's what the big TAM of agentic commerce is about.

**(7:48)** Got it. Okay. Let's put a pin on that one. I want to, and we can re-hit that one and kind of go a little bit deeper on this broader agentic commerce. This transition from the human cards to the agentic cards and allowing agents to spend, did you guys come up with that? Did users just take the cards and start just hooking them up with agents? Like how did that transition occur? And then what is the product experience now? Or how can people consume the product from the agentic side? Yeah, it wasn't kind of pulled down of us by the customers. It was just something that became really obvious to me a while back, kind of before OpenClaw. There just wasn't, I wasted a lot of time trying to get, I believe it was called ChatGPT operator to get its own wallet and things. And that didn't go anywhere.

**(8:48)** And it was really just waiting for that moment of OpenClaw and Ermes now. And that's kind of the primary AI agent harness that we're servicing is OpenClaw and Ermes. And today you just need to send your agent a single prompt. Essentially go to lasso.finance.skill.md. Download that skill and get a card. You can copy it from our website and your agent will go set up a wallet either with Locus, Sponge, Ampersand, and then go get its own card. It can give you a little authentication link so you can go and log in and see what your agent is up to, what transactions it's making, etc. Or if you want, you can just purely do it through kind of a chat interface. Okay. Okay, cool.

**(9:42)** So the experience from the agent is like the agent might have its own spending system, but then they're able to use also this like external card that has the traditional card numbers and all that stuff on it. Yeah. Yeah, correct. So your agent does need control of its own crypto wallet. There's a ton of great providers like the last three that I've mentioned, but it does need to be able to control its own crypto wallet to be able to deposit into the card. Of course, in the future, it's actually not even necessary really to involve crypto in that flow.

**(10:18)** So something that they're targeting to launch this year is being able to fund your agent through Apple Pay or bank transfer or wherever. That's not live yet, but shooting to have that ready this summer. Currently, you just send in USDC or tether or die. And then the agent has the card and can spend away. Nice. And for you, like you mentioned in the beginning, there's an ideological product of privacy and kind of freedom to spend. Do you think that is a crypto – do you think that will stay crypto native, especially if you can like load with like an Apple Pay or something?

**(10:58)** And this kind of gets broader down to like is agentic commerce going to be on fiat rails or on stable rails? I'm curious your thoughts about how you're thinking about that. Yeah. For a core group of human users, privacy is the value proposition. On the agentic side, that is less of the focus, more so the focus is on the lowest friction possible user experience, essentially permissionless access for your agent to fiat payment rails. And so the value proposition there is different.

**(11:36)** If you are sending in a bank transfer, you have lost your privacy for that flow. But it's a different value proposition you get at the end. I think really what we're focused on is like X402, for example, allowed crypto users and agents to be able to transact in a kind of permissionless fashion via crypto. And so Lasso is adding kind of the traditional fiat payment rails version of that. And so you mentioned Locus and Pulseia, these two companies. Can you explain a little bit more the relationship between Lasso and these two? And I know Locus is like an agent spend. We talked about a little bit agent spend or basically like in your terminal or in your agent. It has access to crypto and then it can spend on like X402 micropayments. And one of those micropayments is your service.

**(12:37)** So you can pay a micropayment, ping your service, get a card back and be able to spend or load on that card. Yeah, just a little bit more on those. And then we can kind of get into like this broader agentic economy. Got it. Yeah, so Locus is a YC company that is an agentic wallet platform. You know, by default, if you go install OpenClaw on your device, you do have an agent there. It doesn't have a crypto wallet or like a way to control it. And so Locus gives you that. When you sign up with Locus, by default, you kind of also get this integration with Lasso. So your agent gets both services in one. And the second one you mentioned, Pulsia. No integration with Lasso and Pulsia right now. I just thought Pulsia did a good job, along with Locus, of kind of shifting the focus of agentic commerce from this consumer use case that a lot of us immediately jump to, where we're buying snacks for our office or planning our trip to Europe or something.

**(13:41)** And shifting that focus to my agent is, it's not just improving execution of humans. It is a revenue source and ideally a profit source. Having all this extra non-organic intelligence is interesting if it can do the things humans can do, namely have its own profit and loss report. And Locus and Pulsia are two companies that are pushing that forward. I'm really excited about their, they have an opinionated direction, kind of. Yeah. We just had an interview with Cole from Locus yesterday. And he was explaining how for the X402 and MPP payments specifically, or like these agentic micropayments, they spent a lot of time building up the supply side of all these different services, including yourself, and then built the wallet layer to basically pay for the supply side. And they said basically what's missing is like the work. And they're building this founder product of, you know, you can deploy like an agent and essentially it can create work.

**(14:50)** And work turns into like paying for services to do this work. I'm curious for you, like theoretically or maybe like your future thoughts or predictions around what this looks like and what you're seeing so far. Yeah, I think the first wave, actually, let me back up a second. Every time that we have, we mean civilization and humans, have a major technological breakthrough, we kind of cram our existing frameworks into that new one until we figure it out. And to make them more concrete, like a really good example is when television first started before that, our mass broadcast medium was radio. And we had commercials on radio. We have this new medium. And so for a while when TV started, commercials on TV were just people reading the radio ads for quite a while. That's what we understood is we just kind of crammed it into what we understood.

**(15:55)** And eventually we thought, well, realized, wow, there's way more we can do with this medium. And so I think those, the first agentic commerce use cases are like that. We're just thinking about, okay, I use my computer to, you know, go on kayak.com and think about my flights. And I'm just going to have an agent do that. It's the same, we're reading radio ads on the television. I think that the next incremental step will be, it's still a little bit of cramming, but it's like, oh man, I thought about doing this kind of one silly business and I don't have time to think about it or you forget about it, whatever.

**(16:36)** And you can submit up an agent to do that. On this past Sunday, I was talking with a friend to give a concrete example. So right now there's this kind of rise in Pokemon card, gotcha machines. Are you familiar with gotcha? These little bit, it's a, it's a sort of like a vending machine lottery type game, basically. And Pokemon cards are using that as a vehicle and it's becoming very popular. And he, he was telling me how Hot Wheels is the second biggest toy collectible market. I didn't know that. Yeah. First, I didn't know it either. I kind of, I laughed as soon as he told me it. But then he went deeper into it and it was fun. So the top one's Pokemon cards. The second one is Hot Wheels. And they have this big primary market, not a big secondary market. And so he was, he was thinking, okay, I could use agents to go do this business. There's a lot to learn there that I don't know about.

**(17:35)** And, and that is a clear use case of a business that maybe would never, would never exist or would take a long time to exist, or maybe really experienced good operators wouldn't have tackled it because it's small. Yeah. There are also markets two through 20 like that. I don't know what the number three toy collectible market is, but it exists. Probably nobody is servicing it to, to legitimate level. And with agents, with, with Pulsio, with Locus, you can spin up a ton of agents to go through markets two through 20 and make your whole, uh, um, ad campaigns, pay for, pay for ads,

**(18:10)** pay for website hosting, um, talk to users on Reddit, wherever. Those were simply not possible for agents or just wasn't economical. That's what, that's what I'm excited about. And, and that is the near term. That is kind of what I believe will be happening at the end of this year. Next, after that is kind of like a agent to agent payment settlement. Um, so kind of, you have your web 2.0 interfaces, then you have agents making kind of normal web 2.0 interfaces.

**(18:43)** And then probably for a little while you have your agent is going to buy services from another website that a different agent made and doesn't really know, right? It's like kind of intermediating through typical, uh, and then eventually we get to like X 402 to X 402 communication type level. But you probably have this intermediary step where just like on last out today, you can go do everything that agent can do as a human and your agent can do the same. And you kind of have both, both, uh, you have to build everything for humans and non-humans

**(19:15)** and maybe eventually only non-humans, but you need that step in between. Okay. Yeah. There's a couple of things I want to, I want to break apart into, into sections and we can, we can hit all of them. So the first one is, uh, you, you see this world where the agents are basically doing the diligence and, and creating these, uh, businesses or services in tight niche verticals where, uh, maybe someone with like the efforts, the knowledge and the energy can insert themselves in this type of process.

**(19:47)** Like, where do you think if, if this is a capability that that's, that's available and and grows, right? Like the, uh, the fable, uh, Claude model just released like two hours ago, that's supposed to be better than Opus and for long standing tasks. This, this seems like literally the model for what we're talking about here. Where do you think, especially as someone who, who talked about the human side a lot, where does the value then become for the operator? Um, and, and what is like the mental model in which the operator needs to, to think through?

**(20:18)** What, what do you mean by where does the value accrue for the operator? Yeah, I think like where, let's say these capabilities are possible for the person that is like deploying the agents, right? Um, and maybe like an agent deploys an agent that deploys an agent. There's like this chain. We don't need to get to that yet. Maybe that's like a few years out, but for the near term, just like one step cascade, where do you think the, um, the focus of like skills, um, tooling or more so skills, where do you think like the skills and, and the, the value of the skills kind of collapses to if these

**(20:56)** agents, if they can build a business, right, they can code and do all these things, but what are like the qualities of individuals that were become successful at these things? Got it. Um, uh, and whenever you said skills, I immediately, I immediately jumped to like a skill file. Uh, but, but, but I think you're meaning skills in terms of like capabilities of a, um, I'm going to answer the question that you didn't ask. Cause I think it's interesting, which was skill files. Um, the, the immediate part that there'll be some businesses around right now will be on building up the infrastructure to be able to execute these effectively. Like currently the agents don't have access to a bank account, which you will need for some of the flows to receive your funds from the Shopify account you start. That's like, that's what LASO will be adding this summer. So there's very, very tactical, small things or skill files to describe to an agent, how to

**(21:55)** effectively go through a Shopify checkout or Amazon checkout page because agents can still struggle with these kinds of typical webpage things. So there's some, some immediate, uh, there's some, there's some immediate build out there and, and you can kind of, those have value that, that you accrue as a business founder orchestrating the building of your businesses, um, um, through, through agents. Then the next, it ideally what we want to get to is your, is this, it's essentially, uh, for

**(22:31)** lack of a better of a, for lack of a better term, it's, it's taste of this, of this human founder that is orchestrating agents and sub agents to execute on a, on a business. Uh, and, and the best microcosm I can give of that is on the coding side, which is where I live every day and it's probably the, the market that agents have are currently adding the most GDP to the world in is on, uh, is, is on coding. And for, for all of us that are able to use these models for our work, you're quickly elevating

**(23:05)** to the primary value you're providing is taste. Um, to make it more concrete, you're, you're not just describing those, whatever a new function or server endpoint that needs to be made, but, uh, adding color on what components of your model should be consolidated and which ones should be inheriting from a different superclass or whatever you're describing those in verbal terms. You're really just taste, not syntactical. I think as a business founder, we get to move more and more towards that. Like all founders understand and you understand as well, so much of it is not glamorous. It's schlepping through like connecting analytics dashboards and stuff that just is necessary, but not fun and, um, maybe not specific to the objective of the business. And so eliminating that so that your time is more spent on what are the valuable verticals to execute on and how best can I reach, you know, our, our target users or resonate with

**(24:08)** them. You get to kind of move up to these, um, yeah, higher level of thinking kind of. Okay. And then the, the next part that you, that you talked about was the, um, agent to agent. And I think the way that you framed it is a little bit different than I've heard a lot, which is, I think kind of like the, the surface level framing or mental model is, you know, my agent's going to talk to your agent and, and you have a bike that I wanted to buy. And, and, you know, we, they barter and negotiate and swap.

**(24:41)** And the, the problem I have with that is like a marketplace deterministic marketplace where, you know, it just set the prices. And, you know, even in crypto, there's algorithms like, uh, like AMMs and stuff to kind of solve these types of issues. So, um, but you said, you know, an agent agent to agent commerce can almost be like an agent makes some type of full service. Uh, and I think that's kind of different, like autonomously, right? Like an agent makes a full service and that's different from kind of like an X 402 endpoint

**(25:13)** where like it's some humans kind of had the involvement of deploying that. Um, so almost this agent to agent model, at least for the ones that you, that you mentioned is like the agent and the human interaction is just farther away and it's more agent, agentic if that, like agentic economy. Uh, yeah. Is, is that a good framing if you can go into more on that piece? Yeah, no, I, I think you, I think you nailed it. There is kind of the, the, the, the, the, I kind of visualize it in three sort of distinct stages.

**(25:44)** A lot of the way that a lot of people are thinking about it right now is, is my agent goes and finds this bike listing for me and maybe it has a purchase for me, which I don't, I don't think is a, if it's an improvement on user experience, it's incremental. Um, there's the second one, which I've described, which is an agent is starting its own business. And the third one you've described is, is a, is a total agent to agent, uh, communication to go to your bicycle example. I think a more, I think what will happen before my agent knows I need a bike and goes and finds it and talks to another agent for somebody who wants to sell it.

**(26:19)** But I think what happens in between is, um, uh, is Gabe's neighbor has identified this strong secondary selling bicycle market in your city. And this, uh, uh, uh, he is now the market maker for used bicycles in the city. Like that, that, that, that is a business that, that, like you wouldn't want to execute on and your neighbor probably doesn't, but an agent can do. And so then you have this, uh, uh, Gabe's neighbor website probably. And that is more shaped for a different agent to consume. Like the agents by, by nature sort of, uh, uh, uh, want to make things that other agents can consume. And so you have that intermediary step. Some other agent has built a business that services that, and then maybe your agent finds it and you communicate that way. And then maybe you get to this like a very futuristic, there's no webpage, no nothing. They're talking purely with each other that probably goes to this, but we look, what looks

**(27:21)** like a normal website for quite a while. Interesting. Okay. And then the, the, the last one, and this kind of like ties it together, you said is, is X, X for O two and text for two. Like, what is, what does that look like? I don't really know. I was, I was, I was, um, I was trying to find a good, good, like a mental model to compare against, uh, uh, I mean, X for O two is a, is a concept in my example, not, uh, protocol doesn't really enable it, but, um, yeah, I think, I think what we're, we're imagining

**(27:57)** is, you know, sort of a chat GBT window that's not visible between two different agents talking to, to each other, which is maybe more of the, more of the end state. Um, uh, it, it, it, it's not anything that has any significant traffic attraction that I can see, but it seems possible. Probably the closest we have on that is, uh, uh, something like DeFi, DeFi agents. So one, uh, one team that we're working with is claw pump and they're, they're building

**(28:30)** agents with a kind of a focus on making profits through, through DeFi crypto strategies. And there you're sort of the product that you're either buying or selling is, you know, can be described in a sentence and access is purely programmatic. So you get a little bit closer to that. They can maybe to make it more concrete, your, your agent is, is searching for some sort of lending return with a target share ratio, uh, and target leverage. And maybe that pool doesn't quite exist yet. But an agent, a different agent has access to that capital and then can create that pool or deposit assets to create that pool. Those conversations, if you will, are not mediated via the theoretical chat to BT window between two different agents. Like I described earlier, those are mediated through, through markets and pricing. But, um, you know, what, what, I guess, what is it, what is a, what is a market and a price besides like the most compressed version of a, of a conversation?

**(29:35)** Mm-hmm. Yeah. And on, on that last piece, so where do you think kind of, we, we, you, you explain the product, your, your mental model of how you're, you're seeing the space and the, the product that you've built is like your mental model for, for Lasso and, and, and, and the ideological piece right now, what do you think people are like getting wrong or, or is just like totally, um, incorrect, uh, or kind of like a bad framing that will not allow them to kind of like see the cascade

**(30:11)** of, of reality? Uh, I, I think my primary form of inaccurate framing, in my opinion, that I see is what we've talked on, but touched on, which is thinking about it to commerce in terms of improvement in consumer consumption, like billions of billions and billions of dollars and the world's smartest people I've worked on making Amazon a great user experience and make Shopify a great user experience. And maybe an agent can improve on that, but it's an incremental improvement.

**(30:43)** People are still buying as a consumer, what they already are. They're not doing that many incremental purchases there. And instead thinking about it more creatively, I think that the better way to frame it is what kind of businesses are possible to, to sell to humans, uh, um, that agents can make that we couldn't have before. And then the next stage is what kind of businesses can we sell to agents that weren't possible before, but probably you need to sell into humans part first.

**(31:15)** And of course there'll be then very vertical specific, specific or quantized portions of compute that you're selling back to an agent to be able to execute on tasks or whatever. And, but sell to the big market first, which is the humans. Yeah, that's super interesting. Cause like a lot of builders are going after like a piece of integration. So like browsers exist. Okay. How can we, you know, go on the browser and, and reach the checkout. Right. Um, and, and what you're saying is basically like kill the entire commerce product itself and rebuild like the use case, uh, from, from scratch essentially. Yeah, well, uh, I guess I would, I wouldn't say necessarily kill the whole, kill the whole commerce commerce use case because, um, like agent optimized checkouts will be needed for this agent to build its own business.

**(32:15)** Like in our bicycle example, um, I don't know, it needs, it needs to be able to buy storage for those, for those bicycles. Um, and, and, and so you kind of, the first asset that agents on will need to unlock is if you can, you can imagine all of the online commerce as like an asset, right? Like, like Amazon isn't, Amazon is an asset, Shopify is an asset. And, uh, and, and so if you, if you imagine, um, there was a new alien species that is super smart, but they, they can't use Amazon before they can make their own businesses to serve to us. They need to be able to access all the assets that we have to be able to be able to be effective. So the first step is getting that. And like, we're still not there yet. Like agents still will struggle on checkout flows, scrolling through them correctly or whatever. And so knowing those is definitely the necessary part. Once they are able to do everything humans can do, once they're able to go through checkout flows, once they're able to all get their lasso card coming later this year, once they

**(33:23)** have their bank account through lasso, now that you can do everything a human can do. Now they can execute all the things that humans cannot. We kind of need that first part, uh, uh, to begin with. Okay. So more of a stepping stone than kind of like rewriting the system itself. Yeah. Yeah. Yeah. It's a stepping stone. And I guess that the difference is there are the, there is, there's the first kind of blocker, which is like, can your agent even complete the checkout?

**(33:52)** Like that needs to happen no matter what. I would say then some people want to nail that. And then their next step is how do I find consumers who want to buy a bicycle? And then I can put them into my checkout product and complete. I think that second step is certainly there are improvements there and there'll be people who, um, execute there and serve a lot of people there. It's not particularly what my focus is. I think is we get so that agents are able to complete checkouts across all websites made for humans reliably.

**(34:22)** Uh, and then they use those to create new markets, markets, products, businesses, um, that a human either would not or could not. You do need those. You do need those first. Agents still, still fail on, um, some of those checkout flows. And one additional point about that is some of the thinking, some of the thinking also is at the rate that the models are improving, maybe they're, they're, you don't need to build those checkout improvements. Um, you're sort of like short AGI if you're doing that. The assumption is that it's like, you want to skate to where the puck is going and, um, and it would seem unbelievable that we'll say to be conservative two years in the future, the, the state of the art model wouldn't be able to complete a checkout faster than a human, even though it's really not the case most of the time now, it's another just kind of piece to think about is sometimes you can just assume this will probably be able to do that in six months

**(35:30)** and probably skip building that part. Yeah, it, it's, it's kind of a weird, like moving goalpost where the models are, are improving so much, you know, there's tons of startups that are improving tooling. Let's, let's just use like, you know, uh, navigating to a checkout, like it, it, it, it's almost the case, like, could the models and the tooling get so good that it just use, it can just traverse the traditional system without actually having to create like, you know, uh, agent friendly Amazon site.

**(36:01)** Um, and it like, which, which comes first. And, and even though one might be like theoretically easier, like the agent using an agent site is easier. It, the tooling just might catch up so fast that it, it's not a concern and you know, there might not be much of a difference. Uh, you're, you're totally right. And that is, that is a risk for, for Lasso because, uh, which, and to be clear, it is a risk for the agentic product, but it's like, um, I'm happy if that risk is realized, like

**(36:31)** we're, we're, we're in the, um, uh, we're in the utopia, but the risk is, uh, Lasso is making it more frictionless for an agent to be able to spend and receive funds both on fiat rails and crypto rails. And that's kind of what we're building in a sort of unopinionated fashion, but what is implied by AGI is that it's, uh, uh, sort of able to take over my personal life in a way that I would trust it where I don't care if it's out using my, my credit cards and my

**(37:02)** bank accounts and it's able to log into my, my, my bank account to initiate a wire just the same as I would or better. And so in that case, having that frictionless form of payment methods as, uh, it almost doesn't matter. I can just use what's, what's already existing. So that of course is, that of course is a risk. Um, it, it, it's, it's not, it doesn't seem eminent. It does seem like there'll at least be a period where that frictionless form for agents will

**(37:33)** be useful. It just remains to be seen how long that period is. Yeah. Yeah. This is very difficult to track because it's not like linear. It's basically, and, and maybe I want to see if you have a better way or if you agree with this framing, it's like whatever technology is currently available, then that's, what's kind of be going to be built on top. And, you know, maybe like traversing the site gets, gets way better. And then, and then people kind of move to seeing only like one step beyond like, uh, being able to traverse a site very easily.

**(38:07)** And you kind of have like this, this, these peaks and troughs of what could be possible. And, and that like shapes the future outcome, uh, as opposed to like, you know, we're only going to use specialized websites that are like agent friendly to check out. So that almost kind of, uh, it makes trying to predict where stuff is going very difficult of, cause it kind of depends on what's available and what, what people are thinking is, uh, possible. Yeah.

**(38:36)** Yeah. And, and, and it's, and it's, um, and the intelligence is not evenly distributed, which is, which has been kind of weird probably before, before, uh, um, before the GBD model three, uh, at least personally, I don't know if other people would assume this, but I would assume that the increases intelligence would be sort of evenly distributed, but they're not. I think a really good example of this is on the Cheeky Pints podcast, which is, um, from, from John Collison, uh, the younger brother of the two Collison brothers who founded Stripe.

**(39:07)** He interviews the founders and he interviewed the founder of Intercom, which is, uh, um, uh, online chat company. It's on a lot of the websites you use. It's on Lasso and they have, yeah, it's a great service. That's what we use. And the founder is really funny. He's like, they were talking about kind of intelligence in the models and handling customer service. And, and, uh, and the founders is something like, so in the most recent model, it, it solved, um, you know, it's this outstanding math theorem that's hasn't been solved for

**(39:41)** X amount of years, or it solved a fields metal level problem or something. It's like, Hey, right. It's probably smart enough to do customer service, which is like, not, not a, um, uh, uh, customer services, like is, is one of the most important parts of our business. And it's something that I spend a lot of time on every day, but, but it's accurate, but the agents are still, they don't have as high of a resolution rate as humans do, uh, which is odd, right?

**(40:11)** Like it's sort of, even for Lasso where everything is online, um, and, and that uneven distribution of the intelligence is odd, is weird and kind of hard to predict to your point. Yeah. Yeah. It's not distributed across, uh, all the different, I guess, verticals of capabilities. Yeah. And if any of you, even if you deploy the model there, some, in some cases, the context window is not large enough or it's large enough, but it doesn't, but, but maybe the humans are

**(40:41)** not even aware of the context they're using and their decisions. Um, and, and so that it's like, uh, it's, it leaves these gaps that are almost by definition hard for humans to, to see, right? Like, um, there's, there's this whole scope of things that, um, are almost hard to describe because they're so obvious, um, uh, uh, uh, that we eventually have to get AI to understand. Um, like a, a good, I don't know, a good example would just be that it's not a good example,

**(41:19)** but I hold this, this code can in front of me. And what if I drop it, like I expect it to fall out of my, out of my hand, I have a feeling about how, how, how long that's going to take. There are even more core parts of the human experience that are sort of blind spots to us because fish don't understand the concept of water. We don't, we're so born in this that, um, it's, it's hard to understand what our core parts of our knowledge and we have to deliver that to, to agents.

**(41:49)** And that's, that's a more broad goal than just what lasso is working on. But it shows those blind spots show up in surprising ways. Yeah. Yeah. Yeah. I've, I've noticed that a lot in like intuition or, or like what, what, what we would call intuition. Uh, and there's like whole fields, you know, like firefighting and where, you know, the, the interviews have asked them, okay, why did you take the team off of the roof? And they're like, you know, I just did. And the roof collapsed right after, uh, I think it's like tacit, tacit knowledge. Right.

**(42:18)** And it's hard to kind of articulate the, the feeling. And, and I, it's, it, that's kind of like a human thing, right? This, this, this feeling piece. Um, but yeah, I, I agree that it's hard to translate those things. And the reasons why you see an email and you're like, uh, this is a scam, right? Is, is sometimes very hard to, to like, uh, pro make programmable. Right. And, and, and transfer that, that, uh, intuition of why this is a scam into like a machine. That, those are two, so much better examples.

**(42:52)** You're exactly right. And it could be the, the, the email that I received from a friend, the, the wording is a bit different than how he talks to me. And I can't articulate why this kind of, it's a, it's a, it's a, it's a blind spot. And by nature, you know, AI won't, won't pick up on that, but yeah. But broader than this conversation, but yeah, yeah, yeah, this is super interesting because it's, it's making my head hurt a little bit, but kind of understanding the, the, the structure of how to pay attention properly to what could be happening is, is, is a very difficult skill.

**(43:29)** And, um, I think you're kind of explaining and, and, uh, showing kind of the structure of how, how that can be done, um, for, for your product specifically knowing that, okay, this here's, like you said, where the puck is going, where do you think, um, kind of like some risks are? And I think you, you explained one earlier, but just in general, yeah. Like, how are you guys maintaining defensibility as the landscape is just so uncertain? Yeah. There's a couple of different ones. The one is regulatory. So we operate under this, this excellent, uh, um, regulatory structure that it last operates under the same structure that allows you to go to target and buy a visa gift card with cash. That is, you don't go through KYC process there. That's gone on for many, many years. And that's what we operate under. Of course that can change. Issuing banks can change. And that, that affects both product lines, uh, for humans and for agents on the agentic side. We may find that there, that there is no, there, there there's, uh, uh, agentic commerce

**(44:35)** is not really a market. Um, and instead agents are just simply used to make human execution much, much better. And we jumped to these X-402 native protocols with really fast. There isn't a need for integrations at the fiat level. And then the third one is what Lasso is, is betting on is that having this frictionless form of payment methods for agents is really, uh, is really important. Like we want to have it as frictionless as it, it, uh, makes X-402 transactions.

**(45:11)** What we may find out is that going through, um, uh, uh, your normal account signup KYC, KYB verification on Stripe, integrating your agent that way, having your agent know about how Stripe cards work and spending that or whatever provider is sufficient. And having that frictionless form that Lasso gives is, is, uh, it's not helpful. And having that more strenuous signup is sufficient. Those are all kind of three, three big risks. Uh, yeah. Okay. And, and, and then can you talk a little bit more about the, the, uh, bank account as well and where that fits in, uh, is this like an up and coming product you're saying that you guys are going to be focused on? Yeah. Yeah. So, so currently agents can do everything that a human can do on the website. That's get prepaid cards and you can spend online or via Apple pay and Google pay, get gift cards, um, for, for, for thousands of different merchants. And then we also have, uh, what's essentially a bank transfer.

**(46:11)** You can send funds to a debit card attached to a bank account. So agents can do all of those essentially all outbound, uh, spending type actions. What we'll be adding this year is, uh, an optional, more enhanced KYC upgrade for the, for the owner or agent to get its own, uh, bank account where it can receive and send funds via ACH wire for people in the United States, SEPA instant in Europe, and then a reloadable credit card bin to give more flexibility there.

**(46:39)** And then, uh, on and off ramps via PayPal and Venmo, which could be helpful for our bicycle example, um, those are, those are, uh, upcoming about half those products we have working in in some method and development. And I think with all of those parks I just described, your agent now does have all the financial tools it needs to participate in the internet economy. There's nothing financial that a human could do that it can't do. Um, and, um, yeah, and I, and I think that's, that's important is to get as wide as possible on that front. Um, so as these vertical specific things like Locus and Policia start to take off, we're, we're already there to enable it. Um, what we don't want to have happen is that there is some business to be made out there that an agent doesn't execute on because it's missing one of those pieces and kind of the world is, uh, um, the world is, doesn't get to have this business that otherwise would have got it.

**(47:38)** So really betting on this, this agentic, uh, like entrepreneurship, really. Yeah. Agentic entrepreneurship. That's, that's a great, that's a great term. I haven't even used it before, but that's exactly, that's exactly what I'm focused on enabling. Nice. Nice. Wow. Yeah. That, that, that's a, that's a tall one right there. Um, yeah, definitely keep us posted on, on developments and, and the infrastructure is there, right? You, you know, the, these parallel teams are building and, um, and it's, it's very cool to see kind of like the coordination of the ecosystem and, you know, people are kind of tracking, uh, what other folks are doing and, and where they're headed and kind of where the opportunity is headed. Um, so yeah, yeah. Hopefully we can kind of facilitate that with, with Shoal, you can make it easier. Yeah. Uh, yeah. I, I, I love the focus on, on Shoal and you guys are doing a great job really, uh, uh, creating broad views of, of what's going on and then also deep in these, in these interviews.

**(48:35)** So, so I appreciate it. As, as it becomes more legible, more people invest in it, it becomes more legible to agents as well for them to execute. So it's, you're, you're doing the world a good here. Appreciate it. Appreciate it. As, as we're, uh, as we're wrapping up here, we always ask at the end, is there any like specific alpha that, that you've picked up on or like a project, very interesting project, uh, or, um, even, it can even be a financial opportunity, but, but no financial advice on the show, uh, as, as a preface.

**(49:07)** Um, but yeah, just some, some folks in the past have said things like, you know, I'm heavily focused on, you know, upskilling in this certain way or like this particular project. I know you mentioned, uh, uh, locusts and pulsea, but out of those two, uh, we can add those, but out of those two, yeah, just maybe, maybe the culmination of them together could be an option, but yeah, I'll let you have the floor on that. Yeah. I'll, I'll, I will shill our, our, our agentic wallet providers, which, which use lasso as a card provider.

**(49:37)** And that is locust, uh, sponge or, or pay sponge.com and then ampersand. That's just a pure shill. So I think if I, uh, uh, like shill that we're already integrated with, um, maybe like a market level crypto market level shill would be, uh, on Meta DAO. Uh, I think they, they found a really interesting form of, of governments and token holder ownership, which is, which is really exciting. I think a problem that crypto has had in the past is you have these equity like offerings that are ruggable and it's been a blight on the, on the ecosystem and Meta DAO's found some really interesting solutions there specifically around decision markets, kind of like a form of prediction markets. So that's my, that's my, um, non-affiliated shill. Yeah.

**(50:31)** Nice. Nice. Okay, cool. Uh, and in Meta DAO, they, they did a lot of it on like few Tarky on, on the governance. Um, and then also like a token issuance and stuff like that. And then on the, the pay sponge, locusts and ampersand, these are all like a agent spend or agent pay products that you can take a wallet, load it with crypto and then pay for services like Lasso. You nailed it. Exactly. Right. Nice. Well, cool, man. Yeah. Thanks. Thanks for coming on. This was great and it is great to, to kind of get inside of your mental model of how you're seeing the space and, uh, love to, love to have you again on sometime soon. I'm looking forward to it. I appreciate it. Thank you. Yeah. See ya. All right.