First and foremost, AI is the new customer. In payments, friction is a real issue. Sometimes it's good, sometimes it's monotonous. Do you have to enter your credit card every single time? What I've seen over the years is there's really two key things. How do you remove friction? And how do you enable more participants? We want to make sure that the systems that send money and receive money are trusted, whether that be through our own algorithms or existing infrastructure or new infrastructure that's being built. Twenty thirty comes. What does the world look like? Are people still manually on Amazon clicking a button? What I can tell you is that. Alright. We're here with Jim Nguyen, co founder and CEO of Inflow. Welcome. Thank you. You have a very deep background in payments. Tell me, with AI hitting the scene, where are the biggest issues and pain points that you're seeing today with payments? Well, we have limited time on this podcast, so I wanted to share that, you know, AI is, first and foremost, AI is the new customer. Just like the human is the new customer on the Internet. Kids and our parents became the next buyer. So, agent is another buyer on a new platform, Web3AI. That's how we define the top level of where our thesis is at, is how do we enable buyers and sellers to transact, to send money, and really to be able to participate in commerce in general. So the idea here is we have AI agents out in the world right now. It's being adopted. They're going and running tasks, but they're running into roadblocks. Yes. When they have to pay for something on behalf of the person they're doing the job for, they're stuck. Because everything we've built today is to prevent computers from going through with a payment. Is that right? Absolutely. So, agents, LLMs are doing meaningful work. But what happens here is, there's a roadblock when an agent has to go and access an external paid service. So, for example, if a web developer is sitting in a web development environment, they say, hey, please create me the next great shopping site. And the workspace creates a site. Then the next step is, you need to host this on another service, which we don't provide, which the Workspace doesn't provide, then who does that? The developer has to go from that web developer platform, open a database account, pay for the account, then link it, right? So, we are okay with that because that's how we are trained. But when you look out two to three to four to five years, when you really want the agent to be helping you with all the things that you need to do, it needs to be able to do all these things, but on its own, right? As I shared a little earlier when we spoke, you know, AI agents is leverage on human output. So how do you enable them to do the things that you do, but with guardrails. Right? So, going back to the example, now a web development environment, they have to get their own database, they have to cut their own, create their own domain name, QA service, all these things have to happen differently. And you're saying like if I had it, if I wanted to build a website and I went to a site and said code it for me, that can be done, but in order for me to get a site live, I've got to go get with GoDaddy or whatever my site was, and I've to go buy a domain. Exactly. And that's where the That's where they're stuck. Exactly. That's where they're getting stuck. They're doing good output, good work, but how, in our world, are multiple services tied together for a total output. Right. And that output is a whole website and there's three to four to five steps that needs to be done. So you've had a deep history with payments. You've seen it go from the online boom to mobile to crypto, now AI. Yeah. Tell us a little bit first about how your background is shaped out to to where you are today. So, in those transitions, what I saw really is two phases. The first phase is to really, what's possible in this new paradigm shift. And then, once it's possible, the second generation is, how do you optimize and improve on those possibilities? So, that's really general in the last two or three shifts. What we've seen here is that every platform shift, there's a different interaction model between the buyer and the seller. With online, you have the human with the website. With mobile, you're on the mobile device, and you're interacting through a mobile device. The interaction model is different. But the difference here is the interaction model in the previous two shifts were human based. We're now in the third shift, which is, hey, it is a similar interaction model between buyer and seller, but now is the agent. The main difference is that all the systems that really are secure trust, UI, and all those things are designed on humans. You need to capture, you need to confirm email. Those things were built around human interaction. Ages don't do that. So you have a headless interaction versus the headed interaction where there's UI involved. Right. Headless is API only. Okay. So you've seen each of these kind of waves come. Yeah. As far as an adoption curve, if you had to map it out today, which is gonna have the steepest adoption curve between them? The adoption curve comes where you can think Well, it comes to the payment side. Yes. On the payment side. In payments, friction is a real issue. There are two things, don't Yeah. It's good. Sometimes it's monotonous, right? So do you have to enter your credit card every single time? Right. Right? That's the friction that will reduce a checkout, right? What I've seen over the years is there's really two key things. How do you remove friction? And how do you enable more participants? And more meaning global, right? Can you transact only the US to US or US to Europe or US Latin America or Asia? So you enable more, the commerce becomes more attractive, right? So two things: remove friction, enable more people to do that, more buyers or sellers. Those are the two real big parameters, right? So in this next platform shift, again, there's a friction problem. The friction is: can agents onboard themselves onto a service? And then can they pay? What's happening now is we're seeing a lot of solutions that focus on the wallet. But if the agent can pay, the human still has to onboard with the service, right? Still, there's a friction. You need to be able to make the agent do the same thing that you do, but within the policy that you define. For example, going back to the web example, today I would go to the database providers and say, hey, I'll spend fifty bucks a month and then keep my credit card go. I want to be able to give my agent who's the coder and say, hey, I want to give this agent that spend ability fifty dollars a month on these three websites. Then the agent can go and execute those transactions. If it goes over, then they ask for my permission to say, Hey, it's ten bucks over. Will you allow me to do this? For example. Does this unlock other capabilities when it comes to like digital marketing or, you know, so now you want to say, Hey, I want an agent that's going to go, you know, purchase, you know, Google ads or Facebook ads or anything like this. Like, does this unlock the capability for agents to do more? Meaning like, right now people are using agents in whatever way they can imagine that's within the workflow that stops at the payment side. But like, are there significant unlocks other than just being able to pay if this is implemented? That's a great question. And let me let me roll that back to the agent owner. Right? Humans, as I mentioned, agent is a leverage on human output. So whoever you are, whatever your domain expertise is, whether you're a web developer, you're an insurance broker, you're buying and selling something. That transaction model, and if you trust it and if you trust, if you want to leverage an AI agent to do that for you, then you give that capability. So the opportunity here is pretty expansion of of what humans can do. So there's more buying opportunities. Instead of you being able to buy just eight hours of clicking, you can have twenty agents doing a thousand per second. Right? So when you enable that, it really opens up the whole world of commerce. Okay. So now we've got agents out there. Think with a lot of people who have maybe never built an agent, I think sometimes they think that it's very simple. Yeah. It's like they'll build one. Yeah. Anybody that's built one knows it's not that simple. You got to really get into the weeds of whether you're an N8N or another platform like that of how, you know, actually building if determining which LLM you're going use in each step of the process, and if then statements, and all this good stuff. How can you, if how can you get to a point where I want to not only build an agent, but enable it to go pay for me? Does that have to, does that payment, what you're talking about in terms of unlocking on payment side, does that have to be done at the retailer level, the person like the GoDaddy? Do they have to implement this, or is this something that happens at the stage when you're building the agent? This is synonymous to how we launch buyers and sellers today. So when a company launches their own websites to buy and sell items, they create the website, they test it, they load the catalogue, and then they integrate the payment. Right? So, they would do the same thing with agents. So today, nearly every website in the world has either a workflow for humans to pay on mobile, they have flows for humans to pay on the browser. They need to support a third one. Agents need to be that new buyer. So you have humans coming into mobile, humans coming through a browser, and agent coming in through their APIs. So is this something that the payment processors, like Stripe or others need to get on board and implement? Or is this something that the cop the person who's selling, you know, put in? Got it. So it's gonna be a very, very similar model where you have payment providers like ourselves, providing the payment technologies to the website developers or the sellers themselves. They would integrate these flows into just like they're entering our payment flows into today. How far do you think we are from AI agents who are buyers buying from an AI agent who's a seller? The answer is that the agent is already knocking on the door, says I want to buy. Think about it this way, when you sit on a LLM engine, you say, hey, I want to be able to find three great hotels, and please go book it for me. So, you get three great hotels, it goes, please go and book it yourself. Right? So, it's already knocking on the door, because these capabilities are not available. So, the next step is, you know, but we needed to have AI advance to this stage before we can build payments, right? We can't build before because we need to know where the friction is. The friction is right. Okay, you're sitting at LLM, you've got three hotels, what do you do next? Today, you go to those three sites, you pick one, you onboard, and then you pay. With an agent and wallet right next to your your LLM, you say, please go and book a hotel in in number two, and then send me the receipt. That's that's the eventual goal. And it's gonna happen pretty soon here. With Impla being at early stages of of growth here, what are some of your biggest concerns? Like, what are you worried about? What's keeping you up at night? We've talked a lot about the possibilities, the the plus side. But there's also the the other side where we are very conscious of. Right? How do we enable trusted transactions? That keeps us up all the time. Fintech is one of those use cases where you can't fail. Right? It's like you can't have a four zero four. You can't roll back the code. Right? It's the money's sent. Sometimes it can't be refunded because it's already sent. So we think about that a lot of times, you know, at the surface, we want to talk about friction, but we build trust. We want to make sure that the systems that send money and receive money are trusted, whether that be through our own algorithms or existing infrastructure or new infrastructure that's being built. That's really important is, making sure it's compliant. I say it back another way, you've got LLMs today that you're, you're, let's say you're just building a custom GPT, it gives you the wrong answer and you're like, well this doesn't work, and you move on, Not much loss other than your time plugging in the prompt. In the scenario where you are enabling an AI agent to go purchase something and they purchase the wrong thing or they go rogue and purchase something more than what you wanted, that money is gone. There is a pain point there. There is not, this is not something you just go, oh, nevermind. Absolutely. Exactly. Exactly. And then one of the things here that we want to think about is we try not to talk too much about agents future form, but also pull it back to, from our perspective, when we buy and we transact, we decide based on what we know, what we want and we want to write, Hey, I want to be sent fifty dollars on this service, on this site. We want to give the agent that rule. Only go to the site, only send this much, and only do this. Anything above and beyond that, come back to me. Because when that happens in our own minds, we say, you know what, is it worth it to pay ten bucks more? Yes or no. I think the comeback to me is in the sense of like, is it a text message? Is it a platform they're logging into? Like, you, are we, is the plan to enable people who want to give AI agents the ability to go purchase, is the idea that they're going to have like a SaaS platform to log into where they have their agents and they're determining which ones they're giving, you know Yeah. Budget to? So the just like how you have a fintech app or multiple on your mobile phone today, when it goes when you buy something, it notifies you of a, hey, UPS is gonna come and deliver your products. Hey, would you like to approve this? Which card would you like to use? That's the form that we we it handles. It's on a mobile device. So we can do it as seamless as possible. So the best example of what I have today is I do a transaction and boom, I get a text. I can just respond what the Exactly. You know, category is and Exactly. Exactly. Twenty thirty comes. Yeah. Right? Year twenty thirty, payments go get enabled for AI agents now. What does the world look like in terms of, you know, are people still manually on Amazon clicking a button? Two thousand and thirty is a long ways away for technology. But what I can tell you is that today, I'd love to have an application to book all my travel. I like to have a hotel to buy all the discretionary items that I need on a weekly basis. Maybe an AI agent to manage my bills up to a certain amount. I knew that today. Two thousand and thirty, it was truly AI is a leverage on human capital, human output. Then I think there'll be more and more outputs. But more importantly, with every platform shift, there are use cases and flows that are only able to exist on that platform. So we don't know what that is yet. But let's solve today's issues and today's needs with what's possible, right? The three things I mentioned earlier. And then last question I have for you, you're in a growth phase, early stages of the company. We talked about what's keeping you up. What are you most excited about? I'm excited about just just the the opportunities to really expand on commerce, the transaction, the business models. You know, we talk about how AI can increase GDP. Right? It's it's I'm seeing a path to that is when we get this technology implemented, when we get this across the board, we get the trust in place, we get the the sign off of of the buyers and sellers, I think it's gonna go really quickly. Alright. Jim Nguyen, cofounder and CEO of Inflow. Thanks for coming by. Appreciate it.