← All episodes
Shoal Signal

Agents Already Spent $76 Million: Circle's AI Agent Stack, x402, Stablecoins and the Agentic Economy

A Shoal Signal conversation with Corey Cooper (Circle), hosted by Gabe Tramble.

Transcript

Gabe Tramble (00:01) Yeah, so Corey, when Circle says they're building the financial infrastructure for agents, what exactly does that mean? And and could you explain how that ties into X402?

Corey Cooper (00:11) Yeah, yeah. So, you know What that means is we're focused on two layers. You know, one is obviously at the currency layer. So ⁓ USDC being basically a universal digital dollar that can transact on blockchain Rails. And then the other is actually the blockchain layer, right? The settlement layer. How do we make those transactions more performant? ⁓ create experiences where these transactions can be, you know, ⁓ high frequency. low latency as well as also to completely gasless. and so that is that is how we are approaching, you know, building infrastructure around the open protocol, which is X402.

Gabe Tramble (00:59) And and for builders, like why should this matter to them and and and what is the benefit just to just to get us all caught up before we we dive in.

Corey Cooper (01:09) Yeah, for builders, I think it enables builders to build experiences that were never able to do before. And so what I mean by that is, you know, most builders are building experiences for end users. And in order to deliver those experiences, a lot of times they have to basically get API keys to be able to deliver those experiences. And so APIs is is what the internet application layer has been built on to communicate. communicate you know value but also to information for the most part and so as a builder it's important to be able to deliver you know exceptional experiences to your end user and so XWorld 2 unlocks this ability for a developer to basically deliver experience where an agent or an application can actually communicate with an API endpoint without having to sign up get an account and create an API key. So So therefore, you can create these software to software experiences that are pretty pretty much limitless on the internet. ⁓ You can design an end-user experience where maybe that end user wants to access a lot of information around what is happening on social media from a social intelligence perspective. So now you can design basically an application experience where that person can access all of the information that's happening on ⁓ through a prompt and X402 is enabling that end user experience to be unlocked because that agent can go out, communicate with APIs that kind of scan what is happening on X and deliver a response of what that actual end user wanted. Before then, software was limited, you know, it was limited to what integration points the APIs had. Now with X402 developers can you know clear a lot of boundaries of building super apps maybe that can go out with agents and do so many different things as more and more API providers come on board to the protocol ⁓ with allowing for agents to pay and get access.

Gabe Tramble (03:31) Nice. And so agents can basically access these services and the services can be accessed through micropayments. And kind of one of the big pieces is just USDC, right? The s circle product is the asset that is is typically being transferred. And I and I think it's like ninety-nine percent of the transactions, if not a hundred, are USDC or is USDC, right?

Corey Cooper (03:56) Yeah, yeah. I think a l a lot of that, yes, had to do with one, you know, how USDC was designed from a token perspective, you know, ⁓ by default, you know, we wanted to build ⁓ very robust on and off ramps around USDC. because we wanted USDC to operate as a payment rail. And so in order for payments to be sent around the world, there is what we call on and off ramps into bank accounts. And so our team has done an excellent job of going in the front door in all these different regions of the world and abiding by the regulator. to basically build a highly regulated stable coin that follows the rules and that people can trust that they can actually use this as an interoperable, you know, ⁓ programmable money with the actual fiat system. The other thing is there are some things on the initial design of USDC from a token contract perspective that allows people, ⁓ well, developers actually to interact with functions on the contract. And one of the critical functions that enables ⁓ a gasless flow in the XO 402, you know, work workflow is ⁓ basically ⁓ EIP 3009, which is a transfer with authorization call. So think of that as when the Agent is calling that ABI function is basically signing an authorization that hey, this amount of money I authorize to be basically deducted from this wallet, and it's passing it to basically a facilitator, and so that facilitator and handles the broadcasting and the verification of signature and settlement of that X402 payment, and therefore the agent or the end user initiating it does not have to worry. So it's it's sending that authorized transaction as a message to basically a facilitator and facilitator can can actually handle the rest as opposed to a traditional transfer that happens with the USDC token contract where the initiation requires gas. So there are you know a lot of different things, but that was one design ⁓ you know, choice that was made early on on USDC. that enable the team that launched X-402 to actually build a very gasless, seamless experience early on.

Gabe Tramble (06:32) And and so because of the positioning, because you ⁓ like control more or less the the or or issue the the underlying asset, you guys have the ability to do things that just other service providers can't, like this, this, these gasless transactions. And just before we move on, can you can you just share a little bit of context and ⁓ like definition around what the facilitators are and and where that fits up actually in the stack?

Corey Cooper (06:59) Yeah, yeah. ⁓ you know, a facilitator and X four two flow basically handles the verification of the transaction and then broadcasting that transaction to the actual blockchain network of choice that the that the payment scheme is requiring. And so ⁓ you have like Coinbase, they run a facilitator, do an excellent job, and then you have other companies in the space ⁓ that are running facilitator companies and they're handling basically basically the They're like the acquirer, right, of the transaction. ⁓ and then passing that on to the network and ultimately that is being ⁓ you know deposited into the merchant's wallet address. And so in order for that flow to work, like I said, that transfer with authorization, ⁓

Gabe Tramble (07:43) Got it.

Corey Cooper (07:52) ABI function on the USDC token contract enables that flow to occur in a gas-free, gasless experience. And it was one of the ⁓ you know initial you know ⁓ design choices you know early on of the evolution of USDC. How do we do these things? Because we're we do want to have a future where developers or end users can have these experiences where they don't have to have to worry about. that kind of friction point of of like gas. And so so that that's how the facilitator kind of plays a critical w role of like acquiring those transactions, verifying them and then broadcasting them to the network and settling them in the merchant's wallet address.

Gabe Tramble (08:38) And and basically you're saying you guys have like multiple pieces in the stack. So there's the asset, you know, the facilitator component, and then even the chain component, which which if you could chat on in a minute, ⁓ and then also the application. So if you can if you can go through ⁓ like the full stack and and in its entirety and and kind of hit on these these two pieces, because even on the application side with the the the wallet CLI product, you you basically end to end.

Corey Cooper (09:09) Yeah, yeah. You know, ⁓ you know, as we saw the growth happening in the X402 ecosystem, you know, we started to build a lot of technology around it. ⁓ and so ⁓ you know Obviously, USDC became very popular as the payment currency for X402 flows. ⁓ recently we've launched a CLI tool that allows for developers to easily give agents wallets. And so we we we actually launched it a couple weeks ago. And so ⁓ the wallets that you know you can access through the CLI tool also have built-in functionality that is connected to a lot of our on-chain primitives. For interoperability. So for you know, doing cross-chain transactions with CCTP ⁓ is something that you can get in that CLI tool. So if you're a developer building agentic flows that need to actually go from you know base to Arbitrum, or you know, eventually when we launch Maynet on Arc, you know, base to ARC, that that can be done with our circle agent stack CLI. But also too, ⁓ you know, we built a ⁓ Product that was specifically for agentic flows, which is Circle Nano Payments. And so Circle Nano Payments is wrapped in the CLI tool, and what it enables a developer to do is it leverages our ⁓ Circle Gateway product. So Circle Gateway allows for developers to build chain abstracted balances with USDC. So you deposit your your USDC in a gateway wallet currently. contract on whatever supported source chain we support. And what it does is it gives you basically this virtual balance that instantly can be spent or transferred on whatever native destination chain that Gateway supports. Well on top of that we built nano payments. And so nano payments allow for an agent to deposit their funds in in the gateway wallet contract. And then once it was in the gateway wallet contract, They could perform gas-free transactions to any ⁓ X402 endpoint that supports basically our gateway batching ⁓ you know, ⁓ gateway batching x402 ⁓ product. And so if the merchant has added basically gateway batching to their X402 endpoint. And the agent has a gateway balance. It really doesn't matter what the merchant's chain is of choice of where they want their actual USDC to be sent to. ⁓ it can be transacted, you know, from a gas-free chain abstracted experience, and then the merchant can actually basically batch settle those ⁓ transactions onto any supported USDC chain. So it creates this. experience where we're trying to get to a world where chains don't matter ⁓ when you're dealing with usdc and so we want to you know move toward a world where agent just has usdc and the x402 endpoints are chain abstracted and agents can just transact ⁓ with these endpoints without saying I've got to be on this same chain you know as my wallet to actually Transact because that creates a lot of you know friction and fragmentation in the ecosystem.

Gabe Tramble (12:54) For the the transactions that you're seeing that that you're either hosting through the application or just seeing in general as like the entire ecosystem, which is a lot of transactions for for USDC with regard to X402, what are the use cases or the the types of users that you're you're seeing right now?

Corey Cooper (13:14) Yeah, so the types of users we're seeing in the X four two ecosystem is a lot of ⁓ power users who are probably ⁓ technical users who are working in ⁓ a lot of these coding agent terminals, whether it's cloud code or you know, even cursor codecs. ⁓ and they're working in these these terminals, but they're actually not just coding in these terminals. They're doing a lot of research and other things. ⁓ and so those those power users are basically kind of using it as their their gateway to doing research or accessing information on the internet. And they don't want any limits. ⁓ and so using X402 within those those actual products gives them the ability to not only do their work as they're building products and software, but also do a lot of research where they can go out scrape the internet, scrape different social networks, ⁓ or access a collection of AI models ⁓ as well. And it it just gives them the ability to expand what's possible, you know, from their their entry point of how they're they're working, you know, every day. And so so we see a lot of lot of like technical people, but also too we're sort we're starting to see ⁓ ⁓ non-technical people start to take advantage of these these tools as well. ⁓ You know, as you can see, like Claude, Codex, Cloud Code and Codex, they're they're you know designing these UI interfaces to be a little bit more friendlier ⁓ for the non-technical people. And so you know these these non-technical people you know who I see adopting these right now are mainly like people who are highly focused on like ⁓ marketing data. And so they're starting to be power users too in these ⁓ in these different products and they're leveraging now more and more X four two resources because they're seeing how powerful it is to just have access to ⁓ pretty much thousands of APIs through pay as you go model when they have a wallet, you know, in their in their Product.

Gabe Tramble (15:46) Yeah, yeah. Personally, just a as as a user of of the the Circle Agent product in in my experience, it just really just flattens the ⁓ the internet. And and it seems like just from what you're saying and just some folks that we spoke to in the past, like like Merit Systems, we we had them on ⁓ from Agent Cash, like a lot of people are doing search and retrieval type of jobs where maybe it's like prospecting or as you're saying, ⁓ like marketing, you know, finding information from From social platforms. A more deeper question here, like if we have ⁓ a era, you know, let's say two years from now, and you can basically just call anything that's available on the internet due to these tools. ⁓ because what's happening is like someone can be a domain expert in one specific ⁓ retrieval lane or data lane, right? And they can kind of just focus on that. but what do you think kind of the dynamics of of the internet? become and this is more of maybe like a broad question, like what are the impacts if if everything is just accessible and there's really no gates anymore and it it's just monetary gates.

Corey Cooper (16:55) Mean I I think that You know, I I would like to think pretty optimistic. I think ⁓ it would make everybody a little bit more smarter. ⁓ if if they had access to, you know, anything that was on the internet or behind an API, ⁓ and they could access the things that are behind the API at a fraction of a cent, you know, per request, I think that would give people more intelligence to make more informed decisions. so that's kind of what I what I think potentially will happen. ⁓ you know, when we when we went through the Google era, right, ⁓ people now had at their fingertips the ability to just search anything that was published online because Google did excellent job of indexing. And I think this is kind of an iteration on top of that is now ⁓ paywall content and research Now, you know, that type of information can be monetized from a pay-per-use perspective, unlocks more opportunity for people to actually access it. Because before X402, the the economics didn't make sense for that type of content to be accessible from a pay-per-use perspective. because, you know, you were leveraging fixed costs that were built built into ⁓ you No ⁓ car transactions.

Gabe Tramble (18:29) Yeah. At at the the the service provider level, is are you seeing kind of the same reactions and and maybe if you can talk just like as an enterprise talking to other enterprises, what what has been the reaction and and that kind of assumes that you're talking to these these types of companies, right? Like the the X Pro two providers. But yeah, if you if you can touch on anything as if it's like an enterprise consuming these tools or enterprise providing the tools, if you can share some some some details around that.

Corey Cooper (18:59) Yeah, I mean we we've had several conversations building up our marketplace with providers who are on the enterprise side. ⁓ what was interesting is, you know, ⁓ Most of them knew about X402, but the ones that didn't, the aha moment was, you know, this is a a new customer persona that you can target, right? Which is agents. And agents is the hottest thing in AI, right? And also too, it's a new revenue stream. And so at that moment, when we kind of explain that it's a new persona and a new revenue stream, that's when it clicked. to these actual providers that we should try this. And so I think ⁓ you know most companies that start get exposure, they start getting some exposure to these ⁓ these conversations, I think I think it's kind of a no-brainer for them to say we we need to try this out because There's not many opportunities where you're going to a business and you're saying there's a way for you to target a new customer persona that's growing rapidly. And we have data to support that there's been over $76 million worth of transactions that have happened, you know, with this new persona. And it's growing daily. And also, by the way, like this is a whole new revenue stream for you. And I see think when you kind of talk from those perspectives, everybody starts.

Gabe Tramble (20:35) Yeah. for the the the back to the consumers, ⁓ for like the stuff that they're they're transacting, is there any like ⁓ additional places that you're seeing like activity or in interesting workflows? ⁓ maybe it's like a sp certain type of endpoints that are being targeted or or maybe like provider behaviors. Yeah, is there is there like some some more detail just as as a as an operator on the front lines that that you've noticed in trends there?

Corey Cooper (21:04) Yeah, most most of what we're seeing like, you know, ⁓ from the conversations I've had and working with different providers is number one is we're seeing like research on social. ⁓ mainly like on X. number two is ⁓ People in certain markets wanna access certain models and they can access them because ⁓ they can get access to, you know, traditional payment rails to be a customer of those. ⁓ so ⁓ company like Block Run, they're actually providing frontier models to emerging markets of people who don't have the means to, you know, ⁓ pay twenty dollars a month, right? For a frontier model, or they don't have access to a you know, credit or debit card to put on file because there is lack of banking infrastructure in those markets. So ⁓ that's number two and number three financial intelligence from like the US equities market. So ⁓ there's a lot of ⁓ providers that are providing a lot of granular financial data on, you know, ⁓ Nasdaq ⁓ S P five hundred companies and ⁓ they're using agents to actually, you know, help them make financial decisions on, you know, ⁓ strategic investments. So those are those are the things that we're seeing a lot of.

Gabe Tramble (22:32) Okay. And and we've we've kind of like covered the the the use case, right? And in and the areas in which people, you know, you guys on what you're doing and the and the in the use case for this broader agentic commerce piece, ⁓ right now what we're seeing is like agents hitting endpoints, right? Like service endpoints. What do you think is missing for this ⁓ kind of like agent to agent, ⁓ fully automated environment, ⁓ or even like fully automation is a baked in. assumption, right? What is your maybe like two-year ⁓ visual look like of how agents are interacting based on kind of what we've seen and in your experience so far?

Corey Cooper (23:14) Yeah, you know, I think ⁓ it's a great question. You know, obviously I think the reason why I feel like ⁓ agent to API, you know, is working very well is APIs are deterministic, you know, so like ⁓ and APIs kind of like control how integration works on the internet with software. And so ⁓ it's it's a easy ⁓ entry point into it to a customer experience, right? I think with agent to agent, there is there's a lot of risk that a lot of people are building technology around, you know, ⁓ agents can tend to hallucinate. So if there is one agent communicating with another agent, you know, there's that huge hallucination risk. ⁓ and so ⁓ there's a lot of, you know, risk around just kind of prompt injections and things like that. So I think there's a lot of and then you know there's there's the deterministic part, right? Starts to break down and then how, you know, do we include human in the loop and things like that to make sure like the agent to agent commerce stuff actually starts to work well. ⁓ I think ⁓ I I think we we we there is a lot of teams I would say are working on, you know, solving this, but I think right now we're still some ways away for the agent to agent commerce ⁓ story to kind of ⁓ you know, spread as fast as we've seen the agent to API.

Gabe Tramble (24:59) Yeah, yeah. On the agent to agent commerce, it's it's hard for me to, and I have said this before, to basically like conceptualize what the use case is outside of it being connecting to a deterministic like marketplace, or you know, the agent talking to a service, and then maybe that service talks to another agent. ⁓ maybe that nuance doesn't matter. It it it seems like there's something there, but I I do agree with you kind of like my codex talking to your codex for whatever. Reason for for transacting value, it is it I've just I can't see it yet. ⁓ but yeah.

Corey Cooper (25:36) It's yeah, it's it's it's tough right now because we just haven't, you know, seen any products take off, you know, that ⁓ that can showcase that. ⁓ there's a there's like I said, there's a lot of teams working on it. I've seen a lot of like agent to agent, you know, platforms for, you know, doing like task rabbit jobs. It's kind of like a like platform focused on being a similar to what's the name of the Company I'm trying to think of. Those platforms are like freelancer platforms where you can put up like upwork, those types of yeah, fiber and those. And so I've seen I've seen people kind of build systems like that where agents are submitting work and other agents are picking up the work, and then there's like escrow contracts in the middle to facilitate that with USDC. ⁓ and then you know, you have like these different standards that are happening in the ecosystem, like the 81.

Gabe Tramble (26:17) A board or five or

Corey Cooper (26:42) 183 standard for agent representation or reputation. And so those things, you know, are being built, but we we haven't seen them kind of like take off as fast as we've seen, you know, the age of the API stuff. And I think it's just an easier sale right now for the average person to experience the the agent to API because they understand the concept of APIs are usually like subscription based. And now I can, you know, put a dollar in, interact with this API, and not have to sign up and do all this stuff. But I'm still getting like the value and it's just a paper use. And it feels like Apple Pay actually. And I think that moment kind of makes people kind of gravitate more to like, well let me continuously experience this agent to API. The A to A it's still a lot of groundwork to be done to give that visceral like consumer experience where people, you know, are saying, I've got to get more of this. Yeah.

Gabe Tramble (27:48) Yeah. Yeah. See hitting you know, orchestrating like a couple endpoints together and then getting the result through X402, I think is like a magical experience that you have to experience and then you're like, okay, this this makes sense. do you think that's enough? You do you think this this can be, you know, ⁓ billion, multi billion, hundred billion, trillion dollar market just on ⁓ you know, the the paper use?

Corey Cooper (28:01) Yeah. Yeah. yes. ⁓ yeah, because you know every application has APIs. And so when this becomes the standard Every software application can communicate with other software applications through X402. And so if we want to like think about the TAM of software, that's trillions. And so ⁓ and I think that you know you think about it, just think about just like the integration that it takes for software, you know, to integrate APIs. And then you think about how fast, right, agents can adopt APIs and use them as standards, that that could be the new integration point. Software has an agent that goes out and needs to get access to another software application. That software application has a collection of H402 endpoints that they're willing to give the agent access to. and it can go out, retrieve the responses, and then input it back in the source software. We could, you know, see that future pretty soon and it looks like we're trending in that direction based on the behaviors we're seeing with how people are stringing together different X402 endpoints to get results.

Gabe Tramble (29:45) So is SaaS dead? Everyone says SAS is dead, but we're kind of seeing the early indications of, you know, more use and and easier discovery for these products, just in a different format, right? And it it I think there's so much conflicting information if you gather it all together. I'm curious what your take is on how this evolves for like service providers. Do they commoditize or can they kind of stand out and and even overperform?

Corey Cooper (30:15) Yeah, I don't I don't think SAS is dead. And I think ⁓ a lot of people have to think about kind of these SAS models and how beneficial they are to you projecting, you know, your earnings at the end of the year, right? You know, I think that what it will do is I think ⁓ it will complement SaaS sales. So think of it like this. You you you've seen, you know, a lot of companies, SaaS companies, they'll have like a seven day trial, right? Well, but X402, maybe there's no more seven day trials anymore. It's just you can pay per use, right? Or you can opt into the SAS. And the seven day trial period kind of goes away and it's just an X four two endpoint. Maybe there's certain rate limits, right? That you're constricted to if you're using the X four two endpoints and if you need to upgrade to higher level rate limiting, more, you know, more support and things like that because you're gonna be doing like a lot of transactions with this API point endpoint, then there's ability for you to kind of upgrade and just opt into the SaaS. But in the beginning you can kind of run your trial off of X402 endpoints. And now the API provider they they don't have to worry about ever losing money on servicing an API endpoint. You don't have to worry about the churn that happens when someone signs up for a seven-day trial, right? Puts their card on file, and then on the sixth day, right, cancel it. At least at day one of you providing that API, right, to an agent or a human, you can monetize.

Gabe Tramble (32:11) Yeah, that that is something that we've been looking at a lot. And and basically like our thesis is the the value or like a large portion of the value for now, right? This this could change in like two years. Is the discovery layer? And kind of what you described is this threshold from ⁓ a customer, like the initial customer is the agent on behalf of the person that can convert as they hit like a certain, you know, demand, if they need more demand than than than what is present. And that that's kind of like our lens and how we're looking at at the space. And for yeah, actually, first, yeah, what do you do kind of what what is your take on that? kind of the way that We're seeing it.

Corey Cooper (32:54) So as far as like discovery ⁓ is concerned, well actually restate your question to make sure like I understand so I so I answer it correctly.

Gabe Tramble (33:05) Yeah, yeah, yeah. So we we think that The discovery piece of of kind of the the accumulation of using the product is a form of discovery and it's almost better than advertisement. You know, traditional ⁓ advertisement because you're getting the the payment right then and there. ⁓ and kind of what we're seeing right now is that is a huge portion of the value add, right? Is this discovery and kind of early ⁓ planting the flag into a new customer? ⁓ I'm yeah, I'm curious. Are you guys Is leaning onto this kind of discovery piece or primarily focus on okay, you know, t two years from now or a year from now, when we start seeing extr you know, a very meaningful volume. ⁓ it's it's almost like a a chicken and an egg and and for the for the service providers, we're just seeing like the value today is literally on the the discovery. ⁓ primarily. So yeah, curious curious if you if if you agree or disagree, but ⁓ or how that's probably impacting like how you guys are moving forward.

Corey Cooper (34:10) Yeah. So, you know, at Circle we haven't, you know, ⁓ focused much on on the discovery layer. We're focused mainly on the settlement and the transaction experience, right? with USDC. but discovery is huge. ⁓ what I'm seeing a lot of is, you know, a lot of these companies that are playing in the space that are doing very well, they are actually kind of playing both sides of the fence. They're building up the actual technology. ⁓ on the buyer side and they're building the technology on the sell seller side and they're using skills that are built into their agent buyer side SDKs right to make it easy for the actual ⁓ marketplace of sellers, you know, to be more discoverable. And so I think ⁓ right now I I definitely think that discovery is very, very important because one you want to start you know ⁓ driving a lot of traffic to these actual marketplaces ⁓ so that the word of mouth actually can build up where more providers are want to come and and and actually wrap their API endpoints with X402. And so I see ⁓ right now you know Bazaar basically the X four two ⁓ layer that Built for Bazaar to make things more searchable, adding different like metadata to the actual schema. And then I obviously, you know, Temple has theirs with MPP, they have a discovery layer as well ⁓ that you can add to your endpoint to make semantic search better for the agents to find those endpoints. And so I think that is gonna become you know ⁓ more and more important. You'll probably see more sophistication around indexing and ranking, kind of like what we ⁓ And I think I think right now though what we're seeing in the market is, you know, our a lot of our marketplace providers are actually building basically skill.md files ⁓ and they're building the actual SDK as well and trying to acquire consumers so that their endpoints can actually, you know, be discovered a little bit better. And so they're trying to build up both sides of the house at the same time. And so you know, it is it it's gonna be interesting to see how it plays out from a discovery layer perspective for the ⁓ providers or the sellers. but it is gonna be very important to their businesses because they're gonna wanna rank high and I think ⁓ we're gonna start to see way more sophistication around there pretty soon.

Gabe Tramble (37:06) Yeah, yeah. Right now there's ⁓ there's some pretty pretty solid providers on on the the network and ⁓ I've seen some stat that they're coming on like every couple hours. You guys have a marketplace yourself. can can you just chat a little bit about like how you built the marketplace, not from a like technical perspective, but almost like a a methodology and the the types of companies that are there and and and maybe why you chose those. because that that will help us kind of understand, okay, what's available, you know, b all this stuff is like on chain, right? You can't really see it. So you're a discovery layer in your yourself.

Corey Cooper (37:45) Yeah. Yeah, like I think ⁓ when when we started to kind of like build out the curation of the marketplace, I think the first thing we started to look at is okay, where is most of the, you know, activity happening in the X402 ecosystem that we believe is gonna be around, you know, for the long haul, right? And so ⁓ you know, one of the main things we saw were like marketing teams, you know, using Codex, they were using Claud Code, ⁓ and they were using, you know, even ⁓ you know, OpenClaw, right? To to do marketing research, right? And so we were like, okay, these things are being used by by for a lot of marketing research. ⁓ and then we start to talk to some of the actual early X402 endpoint providers, right? To ask them like, where are you getting most of your usage? And there was a common theme, you know, and that common theme was two things. It was social intelligence and then it was financial intelligence. So that's when we were like, okay, we've got to start like curating a marketplace, you know, that that is mapping back to the behavior of the early traction. Because, you know, if we can curate these these these providers and they can provide really good data, we think that the people who are the power users will eventually you know start building very innovative, basically agentic powered applications, you know. Instead of just a one-shot prompt, they're kind of thinking long term, like how can I actually build basically some software where I have agents, you know, embedded in this software accessing this marketplace of endpoints to provide some service to the end user. And so ⁓ just you know one provider in particular, you know, ⁓ I founded a hackathon you know in SF. It was this company called ⁓ AISA. ⁓ and so they provide a lot of financial intelligence and social intelligence endpoints. ⁓ you know, and they joined our marketplace, one of the first ones to join. there's other companies, Block Run, they kind of focus more on offering ⁓ inference from a pay per use. perspective with all these different frontier models and and local models and open source models. They were getting a lot of traction, you know, for for their ⁓ their inference. ⁓ you know and and so we we partnered with them and brought them in into the marketplace. And then ⁓ you know we started to think about okay like what type of ⁓ workflows you know we're These users will be kind of connecting together multiple A X402 endpoints to do something. What what does that look like or whatever? And you know, we we added ⁓ you know, stable enrich ⁓ and and stable email shouts out to the merit systems team. ⁓ we started to add their stuff to the actual marketplace as well because that's important if you kind of want to build workflows that leverage email ⁓ or phone calls. Calls. ⁓ and then we recently added ⁓ Twilio and Bland that a as well to the marketplace because we saw wherever someone may want an agent to make a phone call, capture some information, and then, you know, ⁓ store that in maybe like a spreadsheet or a Google Docs. So we were thinking like, okay, what type of workflows outside of the social intelligence and the actual financial intelligence that people want to use on a daily basis to be more pro but productive and then let's curate these these endpoints. in our marketplace to actually build, you know, ⁓ the best, you know, ⁓ experience from a a a discovery in a marketplace perspective if possible. And we're constantly adding people to the marketplace or companies. So ⁓ you know, any company, you know, that's interested definitely go to AgentStyCircle dot com and you can submit to be added. But, you know, we really want to build that up because that's the most important component to the agent actually being ⁓ you know, ⁓ valuable.

Gabe Tramble (42:25) Yeah. Yeah, this this trend that that that you're you're saying and and and and that we're seeing too is that these very specific workflows, ⁓ social is what you're saying, you know, financial intelligence, ⁓ again, like this search and retrieval of information, synthesizing it. And I think this the the this is kind of like the value add right right now for for X four two, a value proposition for people using it. There's tons of different stuff, right? There's probably airlines and there's There's you know, stuff that's long tail, right? ⁓ but for now I think that that's makes it very clear cut on on what people are doing. I'm curious, and and this is probably a lot of people's questions, ⁓ why do you think this needs to be on stablecoin rails? And and specifically, you know, why can't this be a bank that, you know, issues something? And and why do you think that stable coins are the perfect vehicle for ⁓ these type of ⁓ interactions between an agent and and a service?

Corey Cooper (43:30) Yeah, I think ⁓ well one I think the actual the settlement aspect, you know, of the merchant actually, you know, receiving instant settlement is important. ⁓ I think the other thing because their value is being given out right immediately. So in that X four two flow, you know, you have hit the endpoint it re re returns the four or two required and explains like this is how much you have to pay and then after that is paid then immediately you're exchanging digital information, right? And so if someone wants to exchange something digital, ⁓ they they wanna be, you know, they wanna have instant settlement. So I think that's that's one. But I think, you know, the biggest thing, right, is When we take a step back, right, and think about like Circle's ⁓ mission, right? Our mission of a company is about raising economic prosperity through the frictionless frictionless exchange of value, right? And how we do that is through USDC, right? It's something that is accessible by anybody around the world ⁓ who has an internet connection and a wallet. And so those things in in in pretty much most markets are are almost free, right? You can you can you know access the internet free in a lot of places and then you know for you to create a wallet ⁓ that is that's free as well. And you can actually have based A digital dollar account that's equivalent to you know ⁓ you having basically a dollar in a bank account, and so when you start looking at X402 and information. Information is value transfer, and we want to make sure you know everybody around the world has access to information and also too, most recently, we want to make sure everybody in the world has access to AI and models and inference. So the reason why it's important for stable coins to continue to be a method used in X402 flows is because you are able to enable people in certain worlds. parts of the world who can't get traditional banking accounts or or debit card or credit card to actually pay for intelligence. And and so again I I bring this company up so many times in this this interview, BlockRun, you know, they're servicing inference to emerging markets who can't get access to these models any other way. And so the only way is through a stable coin And so we want everybody to have access to intelligence. And so that's why it's important for stable coins to be a critical component of, you know, the the payment methods of X402 endpoints.

Gabe Tramble (46:37) Yeah, this this is really interesting 'cause it it almost skips a step and and I'll get to that in a second, but even even before saying that, I think in the in the US we forget how ⁓ privilege, you know, people are to have a robust banking system. And really stablecoins just issue this blanket financial rail, right? Globally that that is literally fully global. ⁓ and and for like the ⁓ the the the primitive of basically like having these cer access to these services, ⁓ it makes it really easy for folks that you know traditionally would try to get like a stablecoin maybe to fiat and then create an account and create the service. But it's almost like X402 is is removing ⁓ a a chunk out of that that process where you know people no longer need to like figure out the the on-ramp, right? Because there there's like some crypto cards and services that are like crypto to X to ⁓ fiat or crypto to like a foreign currency. ⁓ but this just goes straight to the service, right? You don't even need to kind of ⁓ intermediate the change between currencies anymore, which usually was what was what people were working on, right? so that's that's pretty interesting. I'm curious how that kind of plays out

Corey Cooper (47:58) Yeah, yeah. And I think also too, you know, another thing is You know, the the on ramp experience in the US for X four two and USDC has just gotten a lot easier. I don't know if you had a chance, but you know, Cash App just turned on USDC payouts within Cash App. And the way the experience works is you go straight from your, you know, dollar balance in your bank account in Cash App to basically on chain with USDC. So there's no there's no like on wrap experience basically your cash balance in cash app is interoperable with USDC on select chain so now if you're trying to fund an agent wallet to do an X402 transaction it's just you know you just going in cash app and putting the wallet address in there and just sending it and boom it goes from fiat to stable coin into your wallet in less than like one second. So so like you know, I think you know over time, you know, when stable coins become more interoperable with people's bank accounts and we're removing that that actual UI experience of on wrapping and it's just kind of like directly interoperable with with stable coins on chain, I think you're gonna see more actual X four two transactions actually proliferate because like the on ramp experience will it'll just kind of be invisible at that point.

Gabe Tramble (49:40) Yeah. Wow, yeah. Cash app to to agent to ⁓ to agent operations is is is a pretty interesting pipeline.

Corey Cooper (49:48) Yeah, yeah.

Gabe Tramble (49:50) That's good. Just just as we're wrapping up here, I'm curious for you, and we and we've been asking this question to folks coming on, where do you see like the ⁓ alpha bubbling up? Whether it's ⁓ something like a like a block run, right? ⁓ but something that is gonna be like a big opportunity within the space, ⁓ just from your experience and what you're seeing. Yeah, like where where should we be paying attention to or just like check out something super interesting that you see some longevity and value accruing.

Corey Cooper (50:23) You know what's interesting? I don't wanna say it's gonna come from one thing, but I'll say it's it's it's it's it's a few things that could happen, you know, that could that could drive massive adoption. One is I think that there's still an opportunity ⁓ for companies to build the most ⁓ you know, frictionless consumer agent experience that gives you access to thousands of X402 endpoints in a way that you really don't know that ⁓ your interact X402 endpoints. ⁓ I think Merit Systems Poncho product, right, is kind of leaning in that direction. I think what we're gonna see though is the these consumer apps that access you know a collection of X402 endpoints, ⁓ the UX experience will be a lot less verbose. I don't know, like you know, when you're doing these X402 endpoints in interacting with them in a coding agent, you know, you see a lot of things like bash scripts being ran and stuff like that. So that can be a little intimidating for the average person. But I think you're gonna see, you know, some breakout consumer mobile app that basically is running a lot of transactions through X four or two flows that is gonna blow up and it's gonna drive basically the snowball effect of, you know, a lot of API providers wanting to come on board. and and and access, you know, ⁓ you know, the community for, you know, the user base. ⁓ I also think that you'll probably, you know, start seeing ⁓ you know, more and more situations where, you know, like the other day Robinhood launched, ⁓ there third party integrations to their APIs. So that was a big deal where you could basically bring your third party agent to connect to Robinhood's APIs. And so if if you see more of that where APIs that were private are being exposed to agents, well what could potentially happen is that agent is like, okay, I have access to these APIs on Robinhood to To make trades and manage portfolios. Well the agent One thing they're gonna want to do is get smarter, right? And so the easiest way to get smarter from that agent that's low friction, right, is through X402 endpoints. So now that agent, imagine that agent getting very smart by calling different X402 endpoints in the circle marketplace that is giving them financial intelligence data, right? And then it's taking that data and information and using it to make informed decisions. against trading on Robin Hood. Those type of, you know, interconnections, they may look like they're isolated, but they're not, and they're gonna continue to grow the community because more and more people are gonna try to figure out how do I make my agent intelligent really fast without having to set up thousands of API keys to make it, you know, intelligent really fast. And the fastest way is through, you know, the X402, you know, ecosystem.

Gabe Tramble (54:11) Yeah. Wow. Corey, yeah, this is this this has been great. ⁓ I think the for for the the consumer use case, the payments piece, even something as simple as like knowing your balance or or estimating how much a a workflow might cost is kind of just hidden in the in the logs. ⁓ so yeah, that that's that's a really big one. Almost like a ⁓ like almost think of Moonshot, right? When people were like trading coins and and Moonshot came out as a whole new, you know, Apple Pay, USDC. you can easily buy ⁓ coins, similar experience that you're seeing with ⁓ merit system poncho and yeah that that that's a good one

Corey Cooper (54:53) Yeah, yeah, yeah. So I think ⁓ you know, there's there's a lot of companies, you know, in the in the space that are building really cool stuff. So I just encourage people to you know who are listening to actually immerse yourself in the community and and try out these products because these builders in this ecosystem are very smart, very passionate. ⁓ and I think that, you know, ⁓ check out different projects that are in the ecosystem now, like, you know, ours, which is Circle Agent Stack, but also to You know, there are other companies like Proceeds that are building basically a lot of the plumbing on Bringing providers online and and managing ⁓ you know, wrapping their endpoints at scale. Block Run is another one as well, and AISA and several other providers. ⁓ you know, Sponge is doing an excellent job in the community. So there are companies that are really, really focused on building up this community on the seller side and the buyer side. ⁓ some are doing both sides. ⁓ you're gonna see more and more companies like just get laser focused on this ecosystem because, you know, if you're looking at the data, you know, it's showing, you know, a trend of going up, not down. And ⁓ in and companies want to open up opportunities where they can make more money. ⁓ they can access different markets that they couldn't access before because of the stablecoin innovation being built into, you know, X402. So ⁓ so yeah, this has been ⁓ a great conversation, but you know, I I think highlighting a lot of these these companies ⁓ you know, and and having people kind of just test them out is just gonna put the push the industry industry forward.

Gabe Tramble (56:42) Yeah, yeah, I agree. The the biggest value is is is trying it out and then you can kind of mentally understand, wow, okay, these things are powerful.

Corey Cooper (56:51) Yeah, yeah. I think yeah, that that is that is the experience. You have to try it out, you know. ⁓ I think it's it's it's still difficult for us to explain it in a way that does better than you experiencing it. Yeah.

Gabe Tramble (57:08) Corey, thanks again man for for coming on. ⁓ hope to have you on soon.

Corey Cooper (57:12) Yes, thank you. Thank you. I really appreciate it, Gabe, and ⁓ look forward to reading all your literature. Okay, thanks.

Gabe Tramble (57:18) Appreciate it. See ya.

More from Shoal Research. Theses, reports, and interviews on the agentic economy. Browse all research →