Agentic Payments - the Builder’s Perspective

Mason Nystrom sits down with Cuy Sheffield (Visa Crypto) and Jay Yu (Pantera Capital) to explore the future of agentic payments at the intersection of AI, crypto, and cards.

Stablecoin supply grew 40% in 2025. Adjusted transaction volume grew 90%. X402 and the Machine Payment Protocol are live. Visa CLI is in beta. Agentic commerce is no longer a concept. It is happening right now.

Key Topics:

• Stablecoin growth in 2025: 40% supply expansion, 90% growth in adjusted transaction volume
• Stablecoin-linked cards: why wallets need card credentials to reach 150 million merchants on day one
• Vibe coders as the most exciting customer segment on the planet and why headless merchants serve them best
• The three generations of small business: brick and mortar, e-commerce, and the headless merchant endpoint you can build in an hour from bed
• One human companies: why the barrier to starting a business has never been lower
• Killing the API key: why wallet-based identity over X402 and MPP is a 10x better developer experience
• Visa CLI: how Visa built a card-based agent payment tool for vibe coders, starting by solving their own problems
• Why the value of software is trending to zero and trust plus distribution is the only remaining moat

Transcript

SPEAKER_00 0:00

All of the arguments that crypto people make that cards don't work for agents, frankly, are incredibly shallow. This is a once-in-a-generation shift of how money is going to be moved. The real utility of Staples is that it expands the payable surface.

SPEAKER_02 0:13

And if you could build products that help a vibe coder enable their agent to pay and be paid, I think that's an enormous market.

SPEAKER_01 0:23

Welcome to Staple, a Pantera Podcast. I'm Mason Neistrom, your host, and today we have an amazing episode for you where we're going to cover everything at the intersections of payments, AI, and crypto. And for that, we have two very special guests. One is Pantera's very own, JU, who is an investor at Pantera Capital and spends all his free time uh tinkering and playing around with all things, uh agents and payments. Uh, and then we also have perhaps one of uh the best person to tackle the intersection of payments, crypto, and AI, who sits at a very unique seat, Kai Sheffield, who leads all things uh at Visa Crypto Labs. Great to be here. Thanks for having me. But before we begin, a quick disclaimer. This content is for educational and entertainment purposes only and does not constitute financial investment or legal advice. Please do your own research before making any investments. So I think a good place to start, Kai, is level setting from your perspective. So if we look at stable coins over the past year, they've obviously expanded enormously. Stablecoin supply has grown 40% in 2025. Uh stablecoin adjusted transaction volume, which is a stat that Visa tracks for organic volume, has also exploded 90% growth in 2025 and continues to grow into 2026. And so from your perspective at Visa Crypto, what does that strategy look like today? And in particular, what are the problems that you see being tackled that are ready for production now versus the more experimental elements of you know crypto uh and stable coin adoption?

SPEAKER_02 1:50

Yeah, so I think for context, you know, we've been working on you know crypto and stable coins at Visa for about seven years now. Uh and so we set up the team in in late 2018, early 2019. And I think stable coins were pretty obvious to us early on. Yeah, it made a lot of sense that if blockchains were going to be useful for payments, it would probably be for transferring dollars or fiat currencies, not transferring new volatile cryptocurrencies that most people don't have an interest in holding or dealing with. So we've been pretty deeply involved in in the space for a long time. I think our our first approach was just how can we enable better on-ramps and off-ramps? Uh, and so we've worked with you know crypto exchanges and wallets and orchestration platforms, and they're they're merchants. And so being able to use a card, you know, to be able to buy a stable coin, move dollars from a bank account into a wallet, that's been something we worked on for a long time. And you know, there's real volume that that's moving between fiat and stable coins over our network. And then we saw really the growth of uh stablecoin wallets looking to issue stablecoin link cards. Uh, and so that's become a really exciting business for us that you know we like to think of it's it's in hyper growth mode. There's just incredible demand of all these wallets that are popping up all over the world. And I think what they're realizing is if you want to build a stablecoin native Neo Bank global Venmo, a remittance app, a business neo bank, like they're all these new products that people are building. And if you want to get real people to use it, you know, it needs to have utility and people need to know that they can actually spend from balance of a stablecoin. And what cards do is enable a credential to be in a stablecoin wallet that works at 150 million merchants across the world on day one. And so I think that there was a time when stable coins were emerging when people thought, you know, maybe the stablecoin was going to be a payment method in itself, and you know, people are just gonna go and scan a QR code and you know, pay a merchant. And we have not seen merchant adoption anywhere really at scale. And so I think instead we're just seeing a massive opportunity of clients that want to issue cards and want to link a card to a stablecoin wallet. And so that's been a huge focus for us. We think it's a major use case, it's gonna be a big category that continues to grow. And then I think we've also been very focused on how do we use stable coins inside of our network for use cases like settlement of how do we actually move money behind the scenes? And we're seeing a lot of demand and and growth there. And so I think it's still incredibly early. I think with regulatory clarity in the Genius Act, uh, it's clear that stable coins aren't going away. You know, they have real utility across a range of use cases. And now we're in this kind of decade-long march of how do you integrate them into the existing payments and financial system? And we think that there's a major role for Visa to do that, but it doesn't happen overnight. Uh, and so we make a little bit of progress every week, and like we have more conviction than we've ever had that stablecoins will play a role, but it depends on the use case, yeah, it depends on the market. Uh, and so it's not an all or nothing. And we think in many ways, stablecoins can accelerate Visa's growth, which is why we're so excited about them.

SPEAKER_01 5:01

Yeah, you mentioned this idea of stablecoin settlement, which obviously Visa plays a role in, and as well as kind of meeting users and you know, you know, kind of merchants where they are. Uh, and simultaneously, like as the stablecoin opportunity has grown, you've also seen a lot more emergent players try to capture that settlement flows. And so we kind of have this new dichotomy between players that are building their own settlement networks, their own kind of integrated, vertically integrated chains, whether it's L2s or L1s, players like Stripe and Tempo or Circle and Arc. But some then you also have on the other end the players like yourself who are also using permissionless networks like Ethereum and Solana. And so, how do you and Visa think about the difference between leveraging these permissionless networks versus maybe some of these more emergent verticalized ones?

SPEAKER_02 5:49

Our general view is that like we're gonna live in a multi-chain world for a long time, and we want to support all of the chains that our clients demand and want to use. Uh, and I think that there are gonna be only more chains that exist in the future. And I think that it's exciting to see a lot of competition and innovation around certain chains that are optimized or focused on certain use cases. Uh, and so I think it's it almost feels like we're on like the third generation of chains. You know, you had Bitcoin and ETH as the first generation, then you had Solana and Polygon and Base as like the second generation, and you know, they got faster and cheaper. And now you've got this third generation of Tempo, Arc, and Canton, arguably, which are just focused on different use cases. They're focused on institutional capital markets, they're focused on payments, uh, they're not focused on you know crypto, they're not focused on meme coins. And so like we don't really have a strong perspective other than we know that everything that we build needs to be multi-chain, and we want to be able to give our clients the flexibility to say, you know, if you want to build a stablecoin native neobank on Solana, great. We'll help you issue a card and accept settlement on Solana. Yeah, if you want to build a uh AI native merchant endpoint on tempo, great. We'll work with you there. We'll settle with you and you know this a stablecoin on tempo. And so it's I think that the chain ecosystem is vibrant and competitive and that's awesome. And we're not gonna pick winners or losers, we're gonna support a bunch of them and and see where it goes.

SPEAKER_01 7:15

We're clearly moving to a multi-chain world as the opportunity set presents itself. Stablecoins and you know, building like a payment chain is clearly massive. And so I think that's why people are obviously tackling that path. The other kind of flip side of what's happening in payments outside of crypto and stablecoin adoption is agents, in particular as it relates to like agentic payments. Uh, and so I think it'd be great to hear how you view the intersection between crypto and AI today. What do you view as viable versus not viable? What do you think is real versus maybe less interesting and you know how Visa broadly positions itself for the intersection of these two technologies?

SPEAKER_02 7:56

Yeah, it's a fascinating space, and and we've been following it closely for you know a couple years now. Uh, and it's funny thinking back to you had a class of AI crypto agents uh with Truth Terminal and AI 16Z, Eliza. So it was like there was this early you know wave of okay, how could these two worlds come together? And you know, there were a bunch of meme coins and a speculative bubble, you had BitTensor, and so you had kind of crypto and AI as like this like emerging category of crypto. But I think that at the time there wasn't a lot of utility that came out of that. You know, there were a bunch of experiments, but there weren't really businesses, there weren't really viable products with fundamental value, they were more kind of chatbots with meme coins. Uh, and so that was kind of the the first iteration of it. And it was fun and interesting, and like we we followed it closely, but like it we didn't see, I would argue there wasn't really agentic commerce. There was like agentic like crypto meme coins, like that was kind of what what it was about a year ago. And then I think the biggest inflection point was really the launch of X402, which was you know the first time that we started to see a path where you could actually have agentic payments in commerce. Uh, and I think that the story and the vision of X402 is is a great one. Like the the early version of the internet and 402 of payment required, and like bringing that back 20 years later. And I love the idea of killing the API key and being able to use the wallet. And so I think X402 was like a major step function uh that moved the space forward. And so I think that was what about a year ago, or we have lost track of time.

SPEAKER_00 9:42

It's actually like really six months ago that things really started to take. Six months ago when it went live.

SPEAKER_02 9:47

September, October, at some point, yeah. Yeah, X402 went live. Yeah, and I feel like that was an inflection point. Then you had a bunch of activity in X402, and it was very long tail, you know, looked like a toy, differential. I remember doing, I think there was shirt.sh that like you could generate an image, they'd mail a shirt to you. There were a bunch of like meme coins in the games. And so I feel like X402 had this like promise, and the concept made sense. You just didn't really have commerce that was happening through it, but there was a developer ecosystem that you could see start to emerge. And it was starting to blend like AI people, web 2 people, crypto people kind of coming together, hackathon. It was great. It was just like great like developer energy. And like we're very focused on you follow the developers. What are the developers excited about? Arguably, what made stablecoins successful is developers got excited about stablecoins, and you know, they built a bunch of different you know wallets and you know products on top of them, and and we've seen developers get excited about X42. Uh, and so I think that was a an inflection point of you know, the momentum is starting to build there, albeit like very, very early, not meaningful volume, but pretty cool concepts and exciting innovation. And then I think the next inflection point was you know the launch of MPP of the machine payments protocol uh that we worked on with Stripe and Tempo. And I think that was just another step function to say now you've got some of the largest payment companies in the world with Visa and Stripe coming together with a new next generation chain in Tempo, saying, How do we continue to build on top of and really you know iterative what should the protocol look like that enables this communication between agents and merchants? And I think along with that, you know, if you just look at the some of the endpoints in the services that you know you can use over MPP today, there's OpenAI and Anthropic and OpenRouter and like Suno. And so you're starting to see like real merchants that are emerging in many different forms in terms of how these these endpoints are being run. But I think that's incredibly exciting. And then I'd say the last piece is I remember when I first started experimenting and using X402, there was the question of what's the interface look like? Like, how do I actually ping an X402 endpoint? And at the time, the only thing that I knew how to use was X402 scan. And Merit Systems, I think, has done a great job and they've provided a ton of value to the ecosystem. And so they had a tool called Composer that you could go through and it was kind of it looked like a chat GPT interface, but you could call an API and it was like you could do a demo there. But it wasn't something that was like a part of my workflow that I would like use on a regular basis. And so I'd say it wasn't really until December or January that you started to see this stuff come into Cloud Code. And I think you know, the the rise of Cloud Code and of OpenClaw, which although they're not crypto, they kind of feel like crypto because you like run it locally, they're like at least OpenClaw is permissionless and open source. And I feel like that was like a major step where you have a bunch of people that are using these tools, they're creating value and vibe coding and you know, creating all sorts of things. And now these protocols like X402 and MPP can now be embedded into these tools, you know, through MCPs and CLIs. And so like you could see these like puzzle pieces start to fit together. And I think we're at the point today, just to wrap with where we are right now, I am making agentic payments every single day. And I'm having a great time doing it. And I couldn't say that a month ago, I couldn't say that three months ago. I think I made my first like natural in a flow of work agenc payment probably like a month ago, where it wasn't like I was just doing it to test. It was like it actually solved the problem for me. Uh, and now I'm doing it every day. And I plan to continue doing it every day. And I think 2026 will be this inflection point that you look back and you say, okay, before 2026, it was a concept. By the end of 2026, most people, at least in our early adopting corner of crypto, will be doing some type of agentic payment. And that's really exciting that we're we're at that point.

SPEAKER_01 13:54

Yeah. Jay, maybe you can provide some of the investor context. Kai just mentioned these kind of inflection points, X402, MPP, you know, actual usage and adoption of agentic payments. How are you looking at this broader agentic payment landscape and what's interesting from you know the investor perspective?

SPEAKER_00 14:13

Yeah, I mean, I first started really paying attention to like X402 and kind of like the broad vision of what this represented. Maybe I would say also six months to a year ago, around like when it first came out. But back in the day, I looked at it more from a developer talent perspective as well. There wasn't that many people that were really like um building like novel, really interesting products out in the space. And generally it was like confined to a lot of like meme coin corners of the world. But I think really in the last, I would say three to five months, you've really seen this interesting wave of um new builders entering the space from more like Web2 or more like SaaS type backgrounds, alumni from let's say Stripe or Coinbase or some of these other bigger fintechs really coming in and building use cases that are and workflows that are really, really interesting here. And so I think there's been this sort of like meaningful shift, especially like around the time that like Claude Code and OpenClaw has been coming along. Like I personally myself, I come from more of a developer background as well. I've been to going to a bunch of these sorts of like XRO2 and like MPP and et cetera hackathons. And it's just been really amazing to see like the energy that's in those rooms of like developers actually building like zero-to-one tools, for example, surrounding, let's say, ZKID or surrounding um how do you actually buy actual things in actual life, or how do you orchestrate together many different APIs into a single chain? And so you see a lot of like developer grassroots energy come around. I think that's been really exciting. I'd I think one of the biggest debates though on like where crypto kind of fits into this broader vision of like agentic payments is also is basically if we're bank if we're building out this sort of like banking layer for crypto, is that going to be using like stables or using cards or using some combination of both? And I know, for example, like Kai, like Visa is a card network. So I'd be curious to get your perspective, having played with both like stables and being very familiar with cards, like how you plan to sit on that sort of like debate.

SPEAKER_02 16:18

Yeah, I I think it's a silly debate. The answer is both. Yeah, like it's it's just like anytime I see this like very zero sum thinking of is it gonna be cards or is it gonna be stable coins? And it's like it's like coming, it's not, it's gonna be both. Like the overall market, if the Jet Tick payments and commerce happens, it's gonna be a massive market. And there's gonna be massive volume on cards and on stables, and it'll depend on the use case, it'll depend on the market. There are a bunch of different things, like in the middle, but it's definitely not a one or the other. And my general view, and I think what we've been you know working towards you know inside Visa is I've been in crypto close to a decade now. I love stable coins. I have stable coin wallets, I've got USTC in my wallet right now. I would still rather give my agent a card than give my agent a stablecoin wallet for many reasons. There are more controls on cards than there are on wallets. With a stable coin wallet, it's just the controls that exist inside the agent. With a card, you can have a CLI card wallet that the agent has with controls, but then you also have controls that you can put on the card itself that sit outside of the wallet. And so you get this double set of controls and protections, which I think is very important. I think you get the benefit of like in the US in developed markets, like I like to spend on credit. You know, I think there's a big benefit of spending on credit versus debit. There's a reason most people in the US are not just doing 100% of their purchases on debit. And like, that's what a stablecoin is. It's like you're using cash. I like to get rewards. Like, I like to have protections if something goes wrong. And so if you think about you know this vibe coder market, which by the way, I think is the most exciting customer segment on the planet. Like anyone who could build products that are sticky and get real product market fit with vibe coders, it's like this interesting space because it's like somewhat of a niche today that like most people don't even know what a vibe coder is, but it's just like growing every day. And I think they're gonna be hundreds of millions, if not billions, of vibe coders, and everybody creates their own personal software. It's like, and if you could build products that help a vibe coder enable their agent to pay and be paid, I think that's an enormous market. And so, how do you build the best products for the vibe coder? And like, yeah, I'm a vibe coder myself, and you know, it's like what product do I want to use and what how do we solve problems for ourselves? And then if we can solve problems for ourselves, we can solve it for others. I want to give my agent a card. And if I was building an agent company and you know, doing you know all sorts of entrepreneurial things, I would probably get a ramp card uh and I'd probably give that to my agent, and I'd track all the expenses. Maybe I'd have many agents and I'd give them many virtual cards. It's like that is like the best experience for the consumer. And there's a lot of work that we have to do to make it work in all the ways that we want, but I think as a starting point, consumers dictate how they do payments. You know, merchants don't kind of dictate how a consumer pays, it's they want to sell a product to consumer.

SPEAKER_01 19:18

And I think cards are the best consumer experience. Well, so exploring that a bit deeper, I you know you say it's a silly debate, and to an extent, I agree, but I also think you just made an excellent case for why agents are going to use cards. And so, what is the the flip side to that of you know where you think companies just using you know crypto more directly for payments make sense?

SPEAKER_02 19:36

So there we talked about this term, Simon and I on tokenized for going back many years. And we used to call it like the there was the DeFi mullet. I think the bankless guys talked about it. It was like fintech in the front and like crypto in the back. And I feel like as we go into this AI and agency space, we're starting to see like the best outcome could end up being the AI mullet. And it's cards in the front and stables in the back. And I think that's where like stables are really good settlement infrastructure. They're really good for actually moving the money, they're not as good as a front-end consumer payment method. And so I think that's where you can have both working together, where you give a card to an agent, and the agent then has, you know, there's a bunch of controls in space in a bunch of controls in place, and the agent uses the card to spend, but the merchant could be setting up a merchant endpoint on MPP Tempo or on USD or on uh X42, and you could have funds settled to the merchant in real time in a stable coin. So the consumer pays with the card and the merchant receives a stable coin. And I think that is almost like the best of both worlds. Because I think there's a real argument to say that if I'm setting up a merchant endpoint, uh, and the other problem I think we run into a lot is when people think about Egypt e commerce, they think about the existing e-commerce merchants of today, they think about buying socks on Amazon. And I just don't think that that's the use case that really matters. I I think it's going to be these like merchant endpoints that are, you know, Noah Levine talks about this kind of like headless merchants that are selling an API that does image generation, music generation, domains, can be all sorts of developer things. And I think it's hard if you're running an image generation merchant and you have agents that are pinging you to generate images, you don't really want to wait four days to get paid for those images. You got GPU bills to pay. And so I don't think we can have a world where like you have all this cross-border international agent commerce, but the merchants are waiting to get paid to then reinvest to that. So I think you need to have this like real time settlement that the moment someone generates an image, boom, the the merchant gets the four cents of USDC for that image. That's really powerful. So then they can pay their GPU bills, then they can invest and run that business as late as possible. And so that's where it's like you've got to create the best experience for a consumer, and I think that's with a card, and the best experience for a And I think that's with the stablecoin. And so those are some of the things that we're building towards right now. That it doesn't have to be okay, every consumer has got to download a stablecoin wallet and figure out how to give it to Claude and not have to worry about the private key and where that's being held and what are the controls. And then I got to go on ramp and buy USDC for like that. That's silly. Just put your card on file. But on the other hand, the merchant should be able to receive a stablecoin in real time. And so I think that there are ways that both of these can work together. And it just builds on top of some of the work we've done on stablecoin settlement the past few years, where these two systems are merging. And I think that there's a massive opportunity for both of them working together rather than competing with each other.

SPEAKER_00 22:38

Yeah, totally. I think to add on to that, right, like my personal framework for where stables kind of like fit into the picture here is that stable, the real utility of stables is that it expands the payable surface of like different things. So going back to the idea of building for the vibe coder economy, am I really going to register for a full-on like company if I just have, let's say I vibe code like a dashboard or something? Am I really going to register for the full like company with a Delaware C Corp and all of these certificates and whatever just so that people can pay to use my dashboard? Probably not. But I what I can easily do is set up this sort of very lightweight, sort of like, let's say like an XRO2 endpoint or an MPP endpoint that gives me this sort of stablecoin wallet to actually allow me to actually commercialize on my actual vibe coded product. So you can imagine like this hugely expands the surf, like the surface of payable things that I can buy. Like every like different um dashboard that I create or any sort of small service ad hoc thing that I create can all be monetized. And that just accelerates this sort of growth of this general e-commerce market beyond just the traditional sense of e-commerce, but also for like APIs and creating the sort of like next generation of what agenti commerce might look like. And so that's where I think stables are like super, super valuable. Another aspect is also, how you touched on the point of like security guardrails. Um I think another interesting angle that I've been thinking about a lot is, for example, like we've spent years thinking about how do we design the security guardrails for secret keys. Like we've got like NPC, we've got TEs, we've got FHE, and all of these cryptographic techniques. But if you think about it, uh a secret key is really just a very secure form of an API key. You've seen all of these hacks out there with like OpenClaw leaking all of the secrets and whatever. It's so another question that I've been at least like noodling through is how should we actually repurpose that entire like stack used to protect secret keys for API keys in general and to use a bunch of this like cryptographic techniques to create like a more secure like vibe coders playground, so to speak?

SPEAKER_02 24:49

It's a great question. I want to go back to your first point first because I think it's just it's so important that I don't think a lot of people have like really grasped like the barrier to entry to creating a merchant that can sell some product over the internet is going to zero. And if you think about what does the next generation of a small business look like, I would argue that we're entering into the third generation. The first generation was it's a physical like mom and pop shop. And that has a pretty high barrier to entry. You need physical space, you know, that that's not cheap. You might need you know, a permit, you might need a building, you might you gotta have a physical employee. It's like there are a lot of small businesses, they're physical small businesses, but you can't start one of those overnight. Uh like that takes some like real work to stand up a physical brick and mortar shop. Then we saw when the internet emerged, you had the second generation of small businesses where now you could build an e-commerce shop, you could build your own t-shirt shop. And you know, with companies like Shopify and Stripe and and others, and we have Visa kind of enabling a lot of the payments, it's that became so much easier than starting a brick and mortar location and so much lower cost to create your online t-shirt shop than it is to like get physical space. However, it still required some work. You still probably needed some employees. Yeah, you would have to design a website, which is kind of intimidating and yeah, it takes work to do that. You've got to figure out fulfillment. You know, there's there's still some barrier that you're probably not gonna set up a Shopify store, an e-commerce shop overnight. It's probably gonna take some days, weeks, like months of work to be able to do that. Uh, and you're probably gonna have one or like a small number of them. And I think now this third generation, which I would think about as this like headless merchant of like a merchant endpoint, doesn't even have a website. It's literally just an API. Like that's it. And you could vibe code an API while you're laying in bed, and you could do that in an hour from your phone. And so I think we will see the emergence of what I like to think about as like one human companies or OHCs. And some people talk about zero human companies, which I think is kind of hard because it's actually pretty scary of a world of like you've got agents running out around there, there's like nobody that like they're actually like accountable to. So I like about thinking about it as like a one human company where think how much barrier it is to find a co-founder, find someone that you're willing to get into business with. I think you're gonna see a whole new class of entrepreneurs say, I don't I don't need a co-founder, I've got an idea, and I'm sitting in bed and I'm gonna, in my open claw, vibe code some service that I want to sell, and maybe I don't even need a website. Maybe it's literally just an endpoint that I'm gonna make available and I'm gonna sell that product to people's agents. I think that is the lowest barrier to entry to starting a business that has ever existed that you could do in an hour. And I think that it's not just gonna be like someone does one, it's like you could have an individual that has hundreds of these. And so it's like, when was it possible that like you would meet an individual who's running hundreds of businesses like from their phone? There's just no, there's no way that you could do that. And so I think that this barrier to entry is getting lower and lower, and being able to set up endpoints, receiving stable coins, even just like testing an idea. And you could throw something out there, and if it doesn't work, who cares? You spend a half an hour at it, and you go on to the next idea. Uh, it's I think there's this this opportunity for a renaissance in entrepreneurship where you know there could be millions, hundreds of millions of people that are building these agent native businesses, using agents as their employees, and then as they grow, then like you hire other employees. Uh, but I think that's great for people. I think that's great for the economy. And I think that the combination of vibe coding, agentic tools, merchant endpoints, stable coins, like it's all kind of coming together to enable that third generation of small businesses. And that's something like we're super excited about and we want to you know help enable.

SPEAKER_01 28:46

Just to highlight something I think that's important there is the reason that this idea of a headless merchant becomes more possible is not only because it becomes easier for uh you as an individual human to set up this endpoint uh or you kind of your own you know service, but rather than the historical demand side of humans having to like comb through every Shopify store to see what you want to buy or every service to see what you want to buy, an agent can go through that process much faster. You know, it can use one service for this specific type of token compute task, another service for a different one. And so because you have agents that can you know iterate through more services faster, like it it is part of the enabling flywheel of why this like headless merchant concept can can really take off.

SPEAKER_02 29:31

For sure. And I think it is a flywheel that's you have more agents, and then as more agents, then there's a need for more merchants, and then as there are more merchants, and there are more agents, and like you have this kind of two-sided you know market that that starts to emerge. But I think to to your your question, Jay, you talk about API keys for a second, just in in general, and it's so interesting to me because I'm I'm not an engineer at all. I have zero technical ability. I had never written a line of code before 2026, and so I'm coming into this like brand new. I have no idea what I'm doing. And I set up Claude Code in a beginning of January, and I went down the rabbit hole. I've opened up the terminal, now I'm like obsessed with the terminal and never want to leave it. I'm like, it's my favorite thing to do. I don't want to go talk to anybody. I'm like, I've got my headphones on, I just want to vibe code. Oh my dogs. That's that's all I want to do now. And as I build things, I'm like running into like this, all of these like challenges and commerce moments that I'll build a website, I'll build a dashboard, and then I want to be able to publish it, deploy it, share it with people. Uh, I need a domain name. Oh, I can't buy a domain name inside the command line. So then I gotta jump out, I gotta go to GoDaddy, I gotta create an account and go through, I got all these pop-ups. I got I don't know where I'm I don't know what to do. I'm going back and forth. I got three windows open. Then I just get, I'm like, all right, I'm over this. I don't want to do it anymore. And so you lose the flow when you're like hopping out of your inner rhythm vibe coding, then you gotta go and like you know navigate something else. I have the same thing with API keys. Like, there's some service. You you need data, you need inference. Okay, I gotta go sign up for this account. I gotta get which subscription do I do? Where do I find the API key? API keys are in different spots on every single website. It takes me like 10 minutes. I'm literally asking Claude, like, where do I go on this website? Like, it is insane. It's a terrible experience when you're building and you run into a wall and you have to then go find an API key and then copy and paste the API key, and then you're giving the API key to the agent, and you're like, okay, well, if this becomes valuable, like how secure is that? I just put the API key in a chat. So I really, really want to kill the API key. Like, I think the API key should be a relic of the past. I think that the ability to have a wallet that has identity that you can then go and communicate over a protocol like X402 or MPP and say, here's my identity. I want to buy an image for four cents. I don't want to have to go to your website and sign up and create a subscription and grab an API key. I just want one image for four cents. That's what I'd like. And to send that request, get the response to say, okay, yeah, where's your four cents? You send the four cents and you get the image. Like it is a 10x better experience for anyone who's gone through side by side, built something, had to leave to grab an API key and came back, versus built something, stayed in the flow, and just was able to buy what they need as they go. And arguably it's much more secure because then you don't have all these API keys that are floating around that you're trying to figure out where do I put them and what if someone else gets access to them. And so it's it is fascinating as a non-developer that like you developers have been putting up with this for so long, and like I'm just I'm too impatient. I'm like, how do you deal with this? Let's kill these API keys. Like, there's there's no reason that we still have to go through this pain. And I think that's one of the biggest problems that X42 and MPP solves. And and I think it's so much easier for vibe coders to just do commerce in the command line and not have to jump around and go to other places, which is kind of what the default of the internet's been.

SPEAKER_00 33:04

Yeah, totally. Maybe uh I know that you guys have been working really hard on prototyping uh Visa CLI, Visa's own like command line solution. Uh I was playing around with uh yesterday. I was able to create like an entire image-based menu for a fictional Italian restaurant in outer space, basically in an hour, just like vibe coding along and paying like 40 cents. And I was just like using my fingerprint to like tap and execute. And then I wake up this morning and I see all my transactions like rack up just like on my normal credit card. So I thought that was a wonderful experience. I'd be curious to also learn more. Where do you think like, how do you see that sort of like prototype fitting in into this broader vision of like agentic commerce that like Visa has? And what was kind of like the inspiration and kind of like the journey in creating this product?

SPEAKER_02 33:53

Yeah, it's super fun. I think that the the starting point that we've had is if you go all the way back to 2018 and 2019, like one of the things that we really focused on at Visa in the early days of Visa Crypto was how do we build a team of people who are true believers in the technology, who like legitimately believe that this technology is is amazing and is gonna win, and it's just a matter of time, and are willing to kind of work towards making that a reality, and are active users of the technology. And so everyone on on the crypto team in 2018, 2019, like when we got started, we were using stable coins personally. We all had wallets, we'd send stable coins back and forth. A number of meetings at Visa would have people set up wallets and you know, they'd be in another country and I'd send them a stable coin. Like, you just if you didn't use a stable coin, if you've never set up a wallet, there was no amount of consultants you could talk to or memos you could write or webinars and be like, here's what's gonna happen with stable coins. Like, you just you don't know until you use the products, and then you can kind of feel the friction, you could see what the opportunities are, what the benefits are. And so that was just that was a core principle that we've had for seven years now that like we use these technologies personally, and then we find ways that Visa can use these technologies in a kind of secure, yeah, compliant you know way uh inside the the organization. And I think with agentic commerce, like one, it's really fun for me because it feels like crypto in 2017. It really is the frontier right now. And and what I've found and observed is a lot of my friends from crypto who were you know early in ICOs and then they were like all about NFTs, like they're now all into agents, and it's like the same group of people. And no offense, but like crypto itself is a little boring right now. Like it's institutional, like that, it's which is a good thing. It's a good thing that like with regulatory clarity, it's you have BlackRock and Visa, and that's it's the 10-year march of how do you drive adoption. It's it's no longer the frontier of just a bunch of crazy experiments. Uh, and so I think that the frontier has shifted from crypto to now agentic payments. And some of the most creative people who just live in the future, like they're spending all their time building these multi-agent systems, giving them money and stable coins and cards and having them go out and build a bunch of things like that. That's where the frontier is. And so our approach has been how do we live on the frontier? And so, how do we use these technologies personally? And that's why like I started vibe coding, and the whole team is much smarter than I am. Like, they're way deeper in all this stuff, and I'm just kind of tagging along, like trying to make my like simple websites. And then we said, okay, well, like once you start vibe coding, then you experience these problems firsthand and you come across these commerce moments, and then you say, Okay, how do we start to build products for ourselves? And if we're in the space and if we're vibe coders and we build these products for ourselves, and if they work for us, then they might work for other people, and that's the best way to learn. And so that's what we do as a team. And Visa CLI came out of that mindset of, okay, we're all vibe coding. How do we make payments? I've tried probably 15 agencies stablecoin wallets, you know, MCP wallets or CLI wallets. And we said, I don't like to keep having to onboard and manage these keys in a weird way and send funds to my Claude's wallet. And I actually had a time when Claude started buying a bunch of domains. I was like, stop it! Like I didn't have controls. It's it bought these like weird domains that like I didn't ask it to buy. And I was just like, this is not like how the experience should be. And so we're like, what if you could put a card on file, use the tokenization infrastructure, and technology we built so that that could be delegated to an agent? What if you had touch ID? Every time your agent made a purchase, you could just touch ID. Like that makes me feel a lot better than when my agent just went off and started buying things on their own. And so we built the product for ourselves. We started using it internally. I use it every single day. I'm doing all types of things with it, having a great time, having more fun than I've had in my entire career. And then as we've done that, we said, okay, if we like it, if it's useful for us, let's let some of our friends use it. So that's where we are right now of you know, a few friends and in a small friends and family beta, they're using it and they seem to be getting value from it. And so we're gonna go from there of just trying to create the best payment experience for agents to pay and be paid, and living in this ecosystem on the frontier, and then building products that work for us. And if they work for us, then maybe they work for other people.

SPEAKER_01 38:25

And as you think about this compression of being able to build other products more easily because of tools like Claude, how does that change how Visa thinks about what to acquire versus what to build internally?

SPEAKER_02 38:37

I think my my personal opinion on this is that the value of software itself is like very quickly trending towards zero because it's just becoming so easy to build anything that you want to build. Like if you can explain it, you can build it. And that's what we found. And if I can do it, if I'm building things, if I anybody can do it, and so it's just like it's this amazing freedom. And so then you say, okay, well, what's left? And in my opinion, it's trust and distribution. And it's kind of the question of okay, you can build it, but will anyone care? And particularly if it involves money, will anybody trust it? And something that one individual built while sitting on the toilet that they published that has money, do you want to trust something? Is that going to be secure? And so it's like the value of technology and software is going to zero. The value of trusted distribution is getting higher and higher. And I think Visa's in an incredible position, you know, with the trusted distribution that we have. And now, like, what a gift for these magical tools to emerge that enable us to build software faster and easier of anything that we want to build. And so we're very focused on how do we build an AI native team and culture and have you know the right people who are much smarter than I am that know how to use these tools and build all sorts of things. And I think that like early on, what we've seen is like we're shipping more code and comits like on a weekly basis than we had like in six months, like combined. Like, it's just like the speed and velocity that you could build, and so it's very hard to look anywhere else in a lot of areas when you could just say, let's try and build it, let's build it as the way that we learn. And maybe it doesn't make sense, but that's okay. It took us a day to build it, and we could just throw it away, or maybe it does make sense, maybe it led to something else. And so, yeah, I think this is the most interesting time for just like the industry from a product perspective that anything that you can describe, you can build, and it's just a matter of then how do you package it up and decide does it make sense to launch it? And then how do you launch it? How do you do it the right way? Like, how do you make sure it's secure? I think that's a major area we're spending a bunch of time on. What does like cybersecurity in the agentic age you know look like? How do you work with legal and risk? And like we have to do things at Visa that like most startups don't think about. And so I think those are like some of the considerations, but it is it is such a feeling of freedom to be able to build anything that you can describe. And until you really experience that, it's it's hard to explain.

SPEAKER_01 41:19

Yeah. Kai, this has been super insightful. I would love to end uh on a quick fire where I ask a quick question and get your immediate uh response or answer. So outside of Visa and then maybe also Stripe, uh, I'd be curious what company do you think has been most adoptive of crypto and agentic rails?

SPEAKER_02 41:36

I've been really impressed with what Merit Systems has done. I think starting with uh X402 scan and now Agent Cash, like I think that they've built some great products and I've used Agent Cash quite a bit and I've found them useful within my my workflow.

SPEAKER_01 41:49

And then a lot of startups obviously want to work with Visa Crypto and Visa. What's one thing they should think about when trying to partner with Visa Crypto?

SPEAKER_02 41:58

Uh I well, it depends what you're trying to do. I I think we want to work with everybody. Like, I think there's anyone who wants to issue stablecoin link cards, reach out to me and I'll I'll get you to the right place. And we've got a bunch of demand there. Right now, anyone who you know wants to test out Visa CLI, like we're in this friends and family, but yeah, want to get all the feedback we can. So if you want to test it, yeah, reach out. And and I think there's also a big opportunity of we want to add a bunch of merchant endpoints into Visa CLI. You know, we've started with a few just to test it out, but over time, I think you could have you know hundreds or thousands of merchants that are able to be accessed through Visa CLI. And so anyone who's building X402 or MPP endpoints, you know, that wants to be able to make those available for Visa CLI users to pay with the card, reach out to me. Like I think there are many opportunities to partner with Visa across the different initiatives.

SPEAKER_01 42:44

And then the last question I'll open to both you and Jay for the quick fire. I'll start with you, Kai. What belief do you hold about the future of money movement that sounds contrarian today that you think will be obvious in a few years in hindsight?

SPEAKER_02 42:57

It's just obvious, but like cards are the best way for agents to pay and be paid. Like it amazes me that that's contrarian. That I show up at crypto meetups and they're like, there's no way cards are gonna work, that agents can't get bank accounts. I'm like, there's a person that set up that agent, and that person has a bank account. And so it's like all of the arguments that crypto people make that cards don't work for agents, frankly, are incredibly shallow and are things that we are fixing on a day-by-day basis. And so it's like, yeah, I I think very quickly people realize cards do work for agents, and that's our job to make that happen.

SPEAKER_00 43:30

Yeah, I think for me, it's just like agenti-commerce is an e-commerce opportunity. This is a once-in-a-generation shift of how money is going to be moved. And I can imagine a world in which on the internet, maybe 10, 20 years from now, agents move much more money than actually humans do. And I think that's a world that we're moving towards. It's a slow moving process, but eventually we will get there. Awesome. Kai, Jay, thank you both.

SPEAKER_01 43:53

This has been incredibly insightful and incredibly fun. Uh, and for the listeners, if you enjoyed this episode, please like, share, and subscribe. And catch us on the next episode.