AI Agents Will Manage Your Money Better Than You Ever Could

Transcript

All right guys, welcome to AI Supercycle episode number 24. We've got Crossmint co-founder in the building here live in the nest opening up the Agoric finance boom. Alfonso, welcome to the tokenization tower, man. Thank you. Thank you for having me. I'm super excited to be here at the, you know, core Empire State Building middle of Manhattan. Thank you so much for having me. First time here in the Empire State? >> First time, yeah. What's the vibes like coming up the elevator? Very very cool. The only thing is I didn't tell my mother, she's in Spain. She's like, you know, terrorist alert everywhere, whatever. What are you going to do in the top building? So that's >> [laughter] >> I'm going to tell her after this goes live. But no, super excited. I I never knew I'd be here for, you know, to talk about Agoric payments of all the things.

So, yeah. Beautiful. Well, as I said, this is our AI weekly show presented by our friends at Near. They have great near.ai products, [cheering] Iron Claw, more. Check them out at near.ai. And let's talk about Agoric payments. So Crossmint is an API, a back-end provider for stablecoins payments wallets and more. Let's just hear a bit about your your background, Alfonso, how you got into this space, what this looks like, and what you're passionate about. Yeah, um So I've been in I've been in tech for around 13, 14 years.

And I've I've been working on some of the things that maybe lead to what we're starting to see now in in Agoric finance, although it had different names. But I actually started a on the other side of blocking bots, how we call them back in the day. I was the the product manager that created the infamous I'm not a robot checkbox that everyone hates. I I didn't add the traffic lights, so that was someone else. But since then, also was the first PM doing anti-abuse in WhatsApp and a few other platforms. And I've been grappling with this problem of there's clearly a use for automated systems, for bots, for things like this.

Uh Current internet businesses are trying to block it. And there's going to be some tension. This needs to be resolved. And at some point, you know, I think what what around when we're starting Crossmint, the thought was like probably the only way to solve this is actually to have a new foundation of the economy, a programmable layer uh that, you know, happens to use stablecoins, wallets. Although there's there's already things with Visa and Mastercard coming in that are also very interesting. And I think really what primitive is in all of this. And a lot of companies are going to need to migrate. So, we started with that vision in mind, without clarity of when was the time going to be, or the use cases that we're going to be solving every single year.

Uh but that vision started building a suite. And here here we are, more or less. All right. Well, Alfonso, that is really cool that the I am not a robot shot and an idea that that button that you click is something that you guys have pioneered, or you personally. Um I think that's that's that's fantastic. So, I wanted you to talk about agentic kind of finance, payments, commerce today. We've seen a lot of hype over the last several months about this happening, right? We We've heard we need billion TPS blockchains. We need this idea of infinite scalability to provide for agents. They're definitely going to use stablecoins. They're They're not going to use credit cards. All this doesn't

really make sense. There's a lot of kind of hocus pocus. Break us down here. What is agentic finance? And what is the real, separated from the fluff? How does this actually work in practice? Uh yeah. Um the first thing I want to say is like it's it's early. And someone that kind of presents to you that they know 100% how this is going to go, or things like that, is probably more of a smoke uh seller than than, you know, an actual builder. I think the reality is everyone who's here at the edge knows that things are complex, and it's still up in the air how these things are going to work in, you know, 2 3 years.

Um having said that, I can tell you what is real today and what Ortesis is. Um the first thing defining agentic finance, uh you know, different definitions, but I think generally it's if you delegate to your AI agent your money or finances, whatever use case there may be and I can cover some, that is agentic finance. It's basically the idea in the same way that you may have a uh executive assistant and you delegate them some of your finances. It could be managing your investments uh or it could be doing the groceries.

Those things are included. Think of that, but now it's actually Open Claw or uh the next version of ChatGPT, maybe that's more agentic or whatever it is. That's the general idea, no? And so, I can dive into use cases, into payment methods. Let me know where you want to dive, but but I would start with that as the as the foundation. Yeah, okay. And so, when we break it down into like the separate categories, what are the components of the agentic finance stack? Uh you've got APIs, wallets, uh payment processors, card networks, the actual card itself. Maybe just take us through a bit of a components of the stack here. And then what changes when you implement AI agents, agentic finance as part of that stack? Yeah, um maybe I'll start

actually with a bit more like the latter but um I think before even talking about primitives, it's important to think about this from, you know, I think like first principles point of view. Um To be I always go back to the you have an executive assistant, for example, that you delegate some finances to. Um how how how does that work, no? And maybe not everyone in the audience has this, but maybe you have a you know, significant other or your parent that you delegate some finances to and that's the analogy you can think about. Um what changes when someone else manages some of your your money uh uh and not you?

The fundamental thing that changes is there's these uh trust problem, no? Or a principal agency to some extent. You're giving someone your funds and then now you need to trust them uh through different ways, no? And so, every primitive and standard that's coming out is all trying to solve the trust problem in one way or another. Yeah. There are issues that come up around, for example, uh well, I think the main one is like how do you make sure you don't lose all your money uh because it get hacked or things like that. There's another problem you need to fix, which is how is how do you make sure your money is applied in the right places, uh how do you make sure it's legal? That one is going to be the toughest one cuz regulators are still catching up. Uh and what not. And then [clears throat] when

you deconstruct this problem and try to you know, I think provide solutions to this, the primitives that are emerging are on one hand, and I would be very direct about this and I know, you know, this more blockchain topic than not in this podcast, but I think uh credit cards take uh a really like big and important piece of all of this. Credit cards today cannot be delegated to an agent. There's a lot a bunch of reasons. PCI compliance, in in general, agent can leak the entire card, terms of service don't allow it. And in general, the whole chargeback liability framework around cards is not prepared for what if your agent hallucinated that instead of buying a potato, you wanted to buy a tomato and buys the wrong thing.

>> [laughter] >> Th- Those things aren't ready, but Visa and Mastercard and a few other companies have started to build the foundations of this new thing that are I I like to say I like to call them agent scope cards. They really the names they have is more of a mouthful and every company names it the different ways, like Visa Intelligent Commerce and Mastercard Agent Pay. But it's generally imagine a similar to a Ramp card, and we can talk about the Ramp announcement later as well, which is the same technology I'm going to talk about but uh you can Let's you can go with your Capital One, give it to an agent, and if it has this new technology, it can create virtual credit cards with their own number, like um actually more similar to like Apple Pay or Google Chrome.

>> Yeah. Yeah. And then these virtual numbers, the agent can use them. They cannot use the your Capital One, you know, Venture X card, but it can use the the virtual card derived from it, and it gets limits applied at the network level. So, for example, it cannot spend more than $200 or whatever you set. There's going to be limits around merchants, intent, and other things. And so, this is a primitive that's emerging. It's basically secure credit cards for agents that are not only have the guardrails, but they're also changing how the liability shifts uh if there's a dispute and things like that. I can dive more into those later, but that's one of the primitives. And then on the other hand, you have uh stablecoins and methods [clears throat] to send them,

X402 and a couple of the protocols that are coming up, that allow you to pay instead of cards directly with an API first method, which is basically sending a uh a stablecoin to someone else. To enable stablecoins, I don't expect my mother's using an agent, and she will give it stablecoins directly. Yeah. You need uh you need wallets, and then you need on and off-ramps um ideally that make the entire experience uh seamless, that it doesn't feel like you're topping up Solana or something in a wallet. It has to feel more like you're topping up your agent's balance or something like that that is authorized to use. So, I think those are the main ones. There's a third component for, you know, for those who are very like want to nerd out in the audience,

which is the identity part. There's a credential aspect, uh KYA, know your agent. >> Mhm. That one, the only challenge with it is is the farthest away from being standardized. So, there's a few things coming online, but I couldn't tell you there is nothing really that has wide adoption. Uh but it will be the next one. It's more Imagine you can delegate you get method to your agent. Now, your agent may need to buy tobacco and needs to know the buyer the seller needs to know that you're authorized to buy it. So it's over 18 or there's things around, you know, KYC and things like this and how do you solve that problem?

We'll come next with identity as well with agentic credentials or KY A. Yeah. Yeah. Okay, so I think you kind of painted the landscape here in a really thorough way and I think the main pushback that I want to have to like this whole you've argued that stable coins will be a big part of this stack as kind of like this native payment structure for agents. But then in the initial part of your conversation of your comment there, you kind of explained how credit cards enable these virtual cards perhaps with some spending limits or some kind of guardrails for agents.

I'm having a hard time understanding how and why agents truly need stable coins outside of just this virtual card idea. And perhaps it's just an evolution thing. Perhaps it's just we start with credit cards and all that sort and then we move to stable coins. But help me understand why stable coins are so important here because I think the the stemming of all the hype is stable coins are going to be huge agentic commerce. But it's like wait, I can just give my agent my credit card set it you know, some very solid guardrails and that would be easier for me than giving it access to a whole wallet and and you know, private keys seed phrases whatever that is.

Why stable coins? Yeah. >> [clears throat] >> So maybe what I can do is tell you the bull case Yeah. for cards and then the bull case for stable coins separately Got it. and let the audience pick which one they want to they want to stand behind. Uh uh so May I start with the one for stable coins and then come back with the one for cards? >> Sure. So I think the bull case for stable coins is there is a world where uh one uh these new card technologies don't get implemented fast enough. Mhm.

Or let's say the path of the speed of adoption of these new credit card technologies is slower than the demand for agentic finance. Um in that scenario what could happen is you may start seeing a lot of merchants say I'm going to accept stable coins because open clause suddenly are a trillion-dollar market, imagine, making this up. Uh or it could be a big market in certain verticals. And uh half of their users don't yet have cards that are compatible with this or some issue or blocks it or things like this, okay? So, there's a scenario where there's an impedance issue here where these scope cards cannot catch catch up fast enough.

>> But is that realistic? Like is there actually going to be impediments from stable or from credit cards or Visa or these networks to agents? Like or aren't they just going to adapt or or basically that's what you're saying, they may not be able to do it fast enough. So, uh here I'm trying to be as neutral as possible, no? And we are [laughter] to be totally fair, we're neutral and we're trying to help uh card companies as well as stable coin. Like we're trying to help ultimately the agent developers figure this out and how can they get their agent to pay. And it is in our interest to make sure that everything works. Now, uh on the card side, I mean, I've seen I think Visa and Mastercard have been moving as fast as they could ever have.

Uh or as they may have in like the last few years on developing these new standards. You can see announcements from them like every week. So, they're definitely moving fast. At the same time, the card network is complex. Uh there are things like there are issuers that may have a voice. Uh uh there are um some special requirements like uh step-up authentication and things like this, some frameworks around it. For example, in Europe impacts on regulation that may or may not cause friction. Yeah, and and then finally adoption of companies actually like AI agents tokenizing these cards and using this new technology. All these things need to happen. Is it complicated? No. Are company Are these

companies rushing to get it ready? Of course. But I think that everything I think is also going exponential. So, it's hard to say >> Of course. in 2 months what's happening, no? But that is one of the scenarios. The other one for stable coins, there's two other buckets. So, this one is the more generic. Like in general to buy shoes I think it's the least It's the least likely I think, but let's say to pay against a that a shoe merchant decides to stay start accepting stable coins because literally there's so much demand and cards don't work. I think this is very unlikely but plausible that could happen.

The next up in likelihood scenario that gives more weight to stable coins is micro payment based meter systems. So, this idea for anyone who's done agent development or you know, looked a little bit into how these LLMs work. People are already paying for these services metered a bit like electricity or water. You're paying for inference for ChatGPT or cloud or or what have you. Um >> [clears throat] >> A and not only the agent developers are used to this but also they need more tools beyond inference. They may need the computer that are hosted on. They may need maybe an image generation APIs, a bunch of

other things that they're all kind of being paid metered by consumption, by request. And today once again cards maybe are not the best way to pay for this especially in the scenario where your agent needs to select many different vendors or tools and they don't You cannot have a pre-existing relationship with all of them and get an API key and things like this. It becomes easier to just say, "Hey, I'll send you 1 cent every request that I need every, you know, to pay for every hour of compute or to get the weather or things like this." And this is where XYO 2 is starting to develop. It's solving this meter.

And there's a third one that I think this is the wild card. We'll see what happens. Uh we've talked about existing merchants for some reason accepting, well, needing to accept stable coins just because cards ended up not working for other people. That is a low probability but high impact if it happened. The middle bucket is uh uh um micros these new microservices or or sorry, meter services that require micro payments that have it more convenient to pay with stablecoin. And the third one is a whole new category. I think this one is doesn't have a name yet, but I I call it like let's say vibe merchants or something like that.

And it's like it's this idea that um like who created like starting the the hardest part of starting a business um had always been building the let's say a software business had always been building the product, incorporating it and getting uh Stripe or whatever it was the easiest thing. Uh compared to, you know, building a SaaS that you're selling. But that's inverting. Like now I can vibe code uh some service people are vibe coding services from the toilet, no? Like they go to sleep, they tell up in cloth make uh you know, HR software make no mistakes and wake up to it kind of done.

And Uh I see. When you do that, like are you now going to, you know, incorporate a business and deal with all that? Like suddenly looks looks very complicated. How are you going to like open a an account with a PSP? Is there an underwriting process going to allow >> are agents? Right. And yeah, and you say, "Yeah, people buy me are agents and I wrote this thing in the toilet. I didn't test it and like but you know, I got like $1,000 of volume a day and this new segment of buy businesses I think it's real. I don't know how big it will be and at least I mean I do expect if a company start making a million dollars, they definitely will incorporate there's lots of reasons to do that like you'll pay less taxes,

you'll have you know, limited liability etc. But as they start they may start accepting stable coin because there's no underwriting product like you can just start taking them. So this is the third bucket that's on the bull case for for stable coin, no? Yeah. Okay, so just to kind of put this together. We are going to have Vibe coded businesses which are going to have front ends and purchase terminals meant for agents, right? Like they're going to quite literally be created for agentic commerce, for agents to interact to buy and sell them, right?

That is like kind of like one of the pillars. Um the other two pillars kind of frame stable coins as a necessary secondary option when you have, you know, interference from credit card companies, from PSPs, etc. And so this brings us to like okay, at least some of the market is going to be segmented into using stable coins. Like that's for sure. Now I think a lot of the hype and fluff is like it's all going to be stable coins. That's obviously, you know, going to take time and likely never the case. It could eventually get there.

And so what do we need on the blockchain back end to facilitate, let's say, whatever this small amount of initial agentic finance and commerce is going to happen on chain? Is the current set set of chains? Polygon did 500 million or 500 million transactions of stable coin, you know, transactions and payments last month. Near, Tempo's coming out, Arc, plasma, Ethereum itself, and any kind of fast chain. Do we have enough block space? Is this agential finance happening going to solve the block space dilemma right now that is like very very low demand? Or like what is your

perspective on the amount of transaction volume demand that's going to stem from this agential finance sector that uses stable coins not use credit cards? Yeah, yeah. That's a good question. Um Your head is I don't know. No, but uh A but the framework to think about it is the is the is is the right one as you mentioned. What I would tell you is the standards today, like the current version of X102, for example, doesn't scale that well for for a future where you assume there's going to be a significant amount of transactions.

Yeah. Uh now there are potential updates to the protocol. Things like a little bit like in lightning network you have I think it's called channels or things like that or uh in general there there are ways that you don't you you could accept micro payments but don't need to submit a single transaction on every single request. You may be able to batch settle down the line without anyone taking major liability. That I think will help us scale 10 100x uh easily. And then batching transactions and things like that, we may be able to scale three four orders of magnitude where we are today. And we're probably

three or four orders of magnitude to hit the bottleneck in blockchains today. So there's a lot of upside. I don't think the technology and protocols are there. I think everyone working on let's say assume X102 is the protocol, everyone working on the protocols there is going to have a very busy year if things go well in constantly improving uh the protocol for for for this and that. But I think my my guess is there's There's to there's going to be enough demand with high-throughput ones and some L twos. Yeah. But remains to be seen. And maybe that's more of the worst of base. I wouldn't say that's a that's a bull case for Visa because if we're hitting those limits, it's that L twos doing very well. But nonetheless, it is a possible scenario.

We have seen blockchains get collapsed with meme coins and stuff with the Trump token and stuff. So definitely here it could happen if the protocols don't catch up. But I but I think the chains themselves will be able to hold up. And it's just the protocols. Yeah. I think it's a good so segue into what Ramp just launched. They launched this agent card idea here where it's like a a basically card meant for agents. It's has special capabilities, some of which you spoke about earlier. Could you just explain to the audience what this is? Like what is this agent card that Ramp has launched? And is this like a happy medium between the credit card stable coin landscape or like how do you think about agent cards and you

know, what Ramp just launched? Yeah. So yeah, what Ramp with Ramp launched is uh exactly that. It's basically if you're a Ramp customer and you're already using uh for your business to pay pay to pay for things, Ramp issues the cards. Now you can actually delegate those cards to an agent with a Visa intelligent commerce technology, the the one I was mentioning earlier. So basically To to be honest with you, the interesting thing to me is like Ramp already kind of had 90% of it because they already had scope cards.

Uh which is their basically the main thing they sell. Uh the additional which already had limits. So the additional thing is more like there may be higher acceptance or legally the legal framework uh terms of service and whatnot is going to work better. Uh but nonetheless, the the technology they created is you can give an agent a card. you can set a limit uh for spend, I think merchants as well. Yeah. And then it will fail if it doesn't go through. I'm going to put a plug there, which is that on the Crossmint side, we have the API for any company that wants to build the same thing that Ramp built.

>> Yeah. So, hit me up if you wanted to. There's a few agent platforms already building the same thing as well. Yeah. Okay, so that's the Ramp card. That's what Crossmint also kind of has intersection on. Let's talk about the So, we covered the blockchain side, the agentic payment side with stablecoins, and the credit card side. Let's talk about what you guys are doing with Western Union, cuz Western Union it it is also interested in in the stablecoin. There's a Western Union stablecoin. Um I think it's on Solana now, a a couple other chains. What are you guys doing with Western Union, and and what are they doing with agentic payments?

Yeah yeah. Uh it may seem a bit schizophrenic that we're going from agents to Western Union. The reality is the under lying forces that are hap- that are going on in the in in the world and in fintech is that the economic layer is getting upgraded to programmable rails, and it helps Western Union be more efficient and capture more value from their existing network. And [clears throat] it may also enable an AI agent to pay. Uh in general, I would say we are seeing a race between networks, Western Union, MoneyGram, Visa, Yeah. and technology, where I think the smartest ones are adopting the new technology instead of fighting it.

That's what we see with Western Union, we see with Visa. Uh to not get disrupted by it. Kind of assuming that, hey, even if they keep to the hopefully they keep the business because they have the big network, they have the hundreds of millions of people that know their brand, trust it day-to-day for in the case of Western Union receiving remittances. Uh but with the efficiencies of blockchain and whatnot. And so, for Western Union specifically, what I can talk about uh as public announcements is one, they created their own stablecoin in Solana, a USDPT. Yeah.

Uh and second, that they are creating this this network that basically will will add connectivity of that stablecoin to their endpoints, to their cash locations. They have I forgot what it is, but I think they have over 180 countries where they have physical locations from their own stores to the supermarket, uh to like stalls in the street in emerging markets. Yeah. Where you they have their own agents that are not AI, they're they're uh shops or or or a guy with a cart.

Uh that today they allow you to receive a remittance from, you know, your family member in the US in, you know, Morocco. And tomorrow, not only that remittance will go through stablecoin rails, but second, all those endpoints will be places where you can get stablecoin in and out. So, you'll be able to go from cash to the guy you know that's been there 20 years, bakery that happens to have a Western Union logo. And they will be on and off-ramps for stablecoins as well. Uh so, I think that's fascinating. Is the other part the photo of this? No, it's not the agent, it's the kind of the other way around for people who work in the who live in the cash economy, they're they're creating, you know, these on and off-ramps uh for them. And

Crossmint is helping with a few of those, the uh technology underneath. Yeah, well, then let's talk about this kind of more human side because I think the big stablecoin transaction and payment volumes are coming from global payments. And you know, uh foreign workers sending remittances back home to their families, uh typically South America, Africa, Asia. These regions have adopted stablecoins as well as a means of kind of holding value. And so, how will Algorand foresee or AI agents impact the cross-border payment and remittance market, right? Cuz when you think about it, it's very human by design.

Um what does this look like? How will agents and agentic commerce impact these two markets specifically? Maybe in the more developing parts of the world. Yeah um everything I want to say here is a speculative. Uh if there were any concrete plans uh from these companies, actually I wouldn't even know uh not not working with them on this part yet. But uh what I do think is the interesting thing where these things interconnect is let's go back to network, no? In the same case that um in US and, you know, like uh Europe and whatnot, Visa and Mastercard have high adoption. Yep.

Yeah. And they are trying to interoperate with all these things, have these virtual cards that work for AI agents, and and on-ramps to stablecoins. They're also doing stablecoin settlement and whatnot. I do think that brands like Western Union and MoneyGram have a chance of they already have that network in global south, and have a chance of being that entry point to the agentic economy. Maybe I don't know exactly how the primitive would look like, but because they already have the stablecoin balance, the wallet, they already have if you're able to connect that with your ChatGPT or with your open cloud on the line, you will be able to pay with your balance in Western Union, for example.

You may be able to fund your agent by going to a Western Union store, and it it it sounds weird, but but it it actually probably may be the most natural thing for someone in in Nigeria, in rural Nigeria or something like this, that they may need their their agent to automatically buy, you know, for farming, their supplies, and the only way they have to fund it is going and depositing cash in a store, and then the agent picks it up and does stuff, and this this sounds maybe strange to us, but it may be the most natural way to adopt agentic economy for folks in those countries.

So, I do think this is where things overlap. This is not like next month we're going to see a lot of usage of agentic finance by farmers in Nigeria. However, the foundation is getting is getting laid for for this to happen, no? Yeah. Um So, that's the way I see things interconnecting. I'm sure there's other ways that these things interconnect as well. Yeah. Yeah, I mean, what are the other big markets that you would expect disruption for in the more imminent term, right? So, perhaps more of these kind of cash economies will be on the later half of the adoption cycle. More imminently, who are the customers? Who are the businesses? Who is adopting agentic you know, agentic finance or agentic commerce at speed, at scale right now?

Like, who is who is doing this? What types of companies? What industries? Yeah. I actually want to say one thing that in the same way, you know, I think I'm not sure if many folks in the audience know about like mobile adoption in, let's say, Africa and Southeast Asia was was much faster, like smartphone adoption, even lower quality phones, but but the adoption there as percentage of the population and so on was much faster than it was in the emerging world. Um depends on what metric to pick, but in general that adoption of phones there was super fast because people kind of leapfrog the computer.

And I think with agentic tasks, we may see this happening as well. This is a bit of a wild theory, but if you think about it, agents are helping you in some cases skip things that required some kind of white white-collar professions or some kind of education that maybe here we take for granted, but maybe in those countries didn't exist. And so, there are things here that we're used to doing, you know, paying some paying some accountant or things like this that suddenly, you know, someone in rural Africa may get that power when they realize if if as long as they know how to, you know, that they can ask an agent to do it. So, I wouldn't I wouldn't assume that they're going to be

too much lagging behind. I think we could get surprised that a lot of things there may move faster because actually they're not they will get access to, in a way, like intelligence services that wouldn't that they couldn't pay or didn't exist in those countries before. So, anyway, with that aside, where is agentic happening today? Agentic finance happening today? Um I would say there are Really it also how you define it, but if you look at money being moved automatically, I'm going to say that number one that no one wants to talk about cuz it's tainted, but the majority of the volume that's moved automatically is still like DeFi and agents and things like that. Yeah.

It's not that intelligent. A lot of things are like kind of hard-coded algorithms or things like that, but that's clearly, I mean, the majority of money in crypto is being moved automatically. It's less interesting. It's not what people talk about agentic finance. It's not the main thing you think about. But I would say depending on the definition, but that's there, right? And that has been there for a few years. Now, if you think about really LLMs being used to make the decisions, so more intelligence, uh everything is really adjacent use cases to people who use agents today, which are mostly developers and some cases people using agents for growth, uh like marketing and um and things like

this. So, what we do see is the scenario where an uh someone is saying, "Hey, make me, you know, an app that sends an SMS when the weather goes below 30." And the agent builds it and there's like, "I need a I need a service to send an SMS. Yeah. >> Uh I need a Twilio API key or what have you." And suddenly it either asks you for your card or it finds an export or choose service and uses it. So, you see I see some traction in s- services to send email that are built for agents and that you pay programmatically, SMS, to pay for compute, to pay for inference. So, we're starting with

basically what what do the agents for developers need to develop something? So, that's the first set of use cases. There is some commerce. We across me also have this API that allows you to buy through an API with a stable coins like things on on Amazon and Shopify. And we have seen that one grow significantly the last couple months and it's people with open claw asking for toilet paper or or, you know, those kind of things. And open claw realizing >> [clears throat] >> Amazon website didn't work, but they found this API that worked. Um and that is one side. The other side of where we see a lot of interest, I

wouldn't say that much volume yet, is this community of hustlers that has always been in, you know, YouTube and TikTok like uh watching these videos of how I made $10,000 with an A10 or web development or whatever whatever. They're all now how I made $10,000 with open claw. And so, there is a lot of indie devs that are stepping their foot in this, creating these like vibe coded businesses that I was talking about, and some making some money. It's so early that I cannot tell you the biggest things I've seen are things that reduce the token for open claw or things that are still very like within within the the ecosystem, but there's so

many people starting to build these vibe coded businesses that I think I would expect there will be some there that start blowing up down the line. Yeah. Well, I'm glad that you brought that up because I wanted to think about this from like more of an investor's perspective, right? Like if you're looking at this agentic finance landscape here, how do you think about investing in this space? Right? And I'm not asking for like what token to buy or what stock to buy or like which deal to try to get on on the venture side, but more about like which companies or which types of companies are poised to really capture a lot of value and growth from this happening, right? So like in the second and third order effects of agentic

finance happening at scale, right? Like what are the the the specific parts of the stack or or the components that you think are primed to be non-commoditizable and have a an ability to actually charge a premium price with a monetizable moat? What are these companies? And then you you can shill Kausman here or just kind of talk about other areas here, but like which companies have a monetizable moat in the agentic finance world? Yeah. I mean, I'm a little bit obviously so biased that I I think in Kausman we have a thesis. We've been playing in the space for a year and a half.

And things are having playing out as we thought. Um and you know, everyone should we're not we're not looking for investment, but but anyway, I think on obviously I I do think Kausman will do well if we continue executing. Uh having said this in a more generic way, I think what I would say is first I would warn uh I think the right strategy to some extent if you want to uh invest on this is to some extent a hedge. I think going all in on card networks or going short on card networks with happened the other day with that the the reality report >> Yeah. Yeah. is honestly like we don't have enough information. It's going to be very volatile.

Uh the other way around, going all in on on stable coins, it's honestly it's too early. Uh so I think to some extent mm either things that are in between or or or hedging if you really wanted to invest I would I would look at both today. Uh now uh there are things that will always be hard. So there's bank relations, there's licenses, there are things like that that cannot be bytecode, let's say, and that you cannot expect let's say bytecode business to have MTLs. Yeah. So I think if you if you follow that, I think you you will find places, you will find you know, things that will be hard that will still be

hard to build that more people will demand. So I think that's the way I would look at it. Yeah. And so some of the moats in this are like money transmitter licensing, regulation, like regulatory moat. Is there much of a tech moat? Like is there a difference between the adoption of one agentic protocol or or type of agent versus another or is the like intelligence pretty pretty kind of flatline across the board?

Yeah, no, I think uh so I leave the agents outside of this like obviously like the biggest investment is if you pick the right, you know, if Anthropic or OpenAI do well. That always continues to be I mean I don't know if they're overvalued or not. I'm not going to get into that, but but they are the biggest winner from all of this so far seems to be that. Um or Nvidia, you could say. But leaving that aside, um between between LLMs, we see very similar task tool execution rates. So I think both will probably be doing fine.

I think they it is unclear if the API UI, let's say ChatGPT UI and whatnot, will be able to capture value in this case or make the experience better. They did some things around agentic commerce that I don't think have that much product market fit today. And a lot of the economic activity is actually things more like open claw and whatnot where is using the open AI API but not the Yeah. >> not the user profile and and whatnot although they will try to control that I assume. Um so no significant differences there. Um and I would say on the technical realm though, I mean for sure if like let's say X402 ends up being the technology

I don't think there would be big winners in people who are building infra for that, no? >> Yeah. Um if it ends up being cards, it will be the card network. So this is not to say that there's no no good businesses there. Uh it's just like if you want a sure bet, I'm not I'm not sure. I I I would bet on on multiple. Yeah. Last thing I would say is I what I am actually pretty bullish this is this is an investor but if you're an entrepreneur is I I could be wrong here but I think the what I maybe call by businesses or by merchants but in general applications created for agents uh are different than regular websites, are different than mobile apps. And I do

think we may be at the early beginnings of a new app store like opportunity where thousands if not millions of developers may be able to make their own small businesses obviously with some giants emerging. Thinking about this of what if what if you make an app for an agent, no? There's no app store, agents don't need an app store. They install skills and maybe MCP servers. Mhm. Uh but if you start thinking about what are the necessities of agents, what and you think of it from first principle, not abstract but actually what are people using agents for today? Coding, whatever. Who's the person behind it?

What are their needs? Where are they getting stuck? I can you know every time I I I can think of hundreds of things that are needed. Nothing is a billion-dollar company today, but if we're if you extrapolate that there's going to be a thousand times more agentic usage in 3 years, which could be uh suddenly a lot of opportunities become really big. And so this one not ready to invest, but if you're starting something, be it venture-backed business, not sure, but at least like you want to start a bootstrap business, I would definitely look at this space.

There's almost no one yet who has figured it out and many more opportunities than than than than, you know, competent builders. So I would look at that. Yeah. >> Yeah. So kind of looking ahead here as we come to the last couple of minutes, do you see a future where most online transactions are actually initiated by AI agents? If so, you know, what does the stablecoin component look like of that? What is the cross-mint component look like of that? And what is the kind of change in structure of the internet look like as a result of the initiation of a lot of transactions by agents? You mentioned the micro-payment side and the metering taking over the traditional

software like uh a monthly billing model, for example. Is this something that you expect? And then what are the ramifications of that? Yeah. So go [clears throat] even bolder. I'm I'm make a prediction which is like kids born these days, when they become adults, I think money for them will be a let's say a back-end technology. Uh I don't think in the world in 10 or 15 years you're going to see prices on things. You're going to be uh thinking about money. You may know how much how much your net worth, how much it went up and down, but I think every decision about whether to pay, how much to pay for one thing or another, how to manage your finances,

absolutely it be done better by agents. So it will be all delegated. Like today we're figuring out how to trust them, but I think in five years we won't trust ourselves doing it better than an agent. Yeah. Like we won't trust myself buying the right pizza, the right pants, or investing the right strategies. So, I do think we're moving towards a world where money disappears, is a back-end technology, you your agent takes care of it. And of course that requires a lot of trust, but I think that will happen. In that transition, everything's going to change. Uh again, there could still be card networks, stable coins, and and whatnot, but that's definitely going to happen. What is Crossmint playing all of this? We're an infrastructure company, we help businesses don't stay behind. We're at the forefront, we're the first

integrators of uh hopefully live going live with this intelligent commerce. Also have on-ramps, uh wallets, and have been building uh for agent developers for a year and a half, and are col- collaborating in all the standards that are coming up. So, I I think the way we help it depends on every month, but in general, if a business is thinking is building their own agent platform, or is thinking how to accept agent payments or things like that, either we build it or we're connected with the best who build that part and can connect you. So, that's kind of the box to put us in if you're in the agentic side. Outside of that, uh any other company moving to the programmable economy, be it using stable coins for remittances like MoneyGram and

Western Union, who we work with, uh cross-border payroll, OEMs, there's a lot of other companies that are looking at uh integrating stable coins in this stack. Our solution is white label, works kind of for every kind of use case, um and we're always happy to help. Beautiful. Alfonso, great great great conversation. Thank you for joining us. Uh agentic finance is a boom right now, it it will keep happening. I think it's yet to be seen what the percentage split is between payment volumes in in credit cards or traditional payments and stable coins. I'm on team stable coins. I want it to happen. I want to see 24/7 global payments. I I want to see actual businesses not not have to suffer from

failed card payments, you know, lowering their LTVs, having all different types of of issues that will be solved with with stable coins, better APIs, better wallments or sorry, better wallets and better payments kind of like structures overall. Alfonso, great work, thank you for joining us for AI Supercycle episode number 24. Thank you for having me. I hope everyone got at least one useful learning and if you want to follow along, follow me on Twitter, search for Alfonso Cosman and you'll find me there.

Guys, that's AI Supercycle episode number 24 powered by Near, the AI Supercycle.