Hello everybody, I am Asheesh Birla and I am the CEO of Evernorth. And today I'm joined with a long time friend and former Ripple colleague Chandler Fang. We'll get to have him introduce himself in a second. But Evernorth is building the largest institutional XRP treasury out there. And in running it means that we're making hundreds of treasury decisions per day and maybe in the future that even accelerates. Why? Because crypto moves so frictionlessly and there's so much opportunity out there, which makes me question, is there a new way that we treasurers and treasury companies and the traditional corporate treasurer should be thinking about managing their treasury operations in the context of AI? And the reality is that a lot of treasury companies like Evernorth - we're actively managing our treasury versus passive management of treasury. That means that we're actually going to put that treasury to work into different kinds of yield operations across traditional means and also on chain using the XRP Ledger. The question is, can humans manage all that risk appropriately as the scale and speed of crypto comes into play, especially with decentralized finance or DeFi. So I brought in Chandler because Chandler is an awesome guy. One of the smartest people I've ever had the pleasure of working with. He's the founder of an exciting company called t54. Chandler, I think you're the perfect person to answer some of the questions that I just asked that I think that treasury companies like Evernorth and corporate treasurers are going to face in this convergence of crypto and AI. Chandler, why don't you take it away and give a quick intro.

Yeah, absolutely. Thank you for having me, Asheesh. Very, very excited to be here. My name Chandler. I'm the founder of t54 Labs. So we are empowering the trusted agent economy. When agents start to move your money around, doing transactions, payment, shopping, in this case, treasury management, how do we fundamentally make sure that agents are behaving within the guardrail? How to make sure that agents are fully leveraging this power and helping financial activities? So we started the company in January 2025 and super happy to be here.

Amazing before we get into like the infrastructure and the deep topics here Chandler your background as a quant trader at JP Morgan before coming to ripple and working with us on payments. You know most people are coming to AI in finance or from the crypto angle or solely from the AI angle, but you're coming from it slightly different lens bringing that traditional experience and then that crypto experience. And part of the reason you probably found this opportunity in t54. What are others missing that your experience in both crypto and with JP Morgan helped you see?

Yeah, I do have a relatively unique experience where I first of all observed how AI as a technology evolved over time. On the other side, I'm also working in a crypto, in a big financial institution, JP Morgan, in a crypto team before, versus, and then work in a big crypto company, I would say, Ripple, and focusing on some of the real life applications. So I definitely see a lot of great trend. Let me start with something about AI first. So back to 2019, when I was in JP Morgan AI team for the payment stuff and the AI as a technology was pre-transformer. It was nothing to do with large language model so the AI we were using back then are machine learning, reinforcement learning, time series forecasting, especially for language, we're using something called natural language processing. I'm not sure the audience, some of them are old enough know the legacy technology, but today nobody talks about those anymore because large language model itself pretty much can handle all the text basis task already. So number one is we see this consolidation of the technology, being very much focusing on large-language model as the approach to achieve the intelligence. With that said, the evolution or the speed of evolution is getting faster and faster. When I was in JP Morgan doing research on natural language processing, we checked the mainstream model, and three months later we see a small version. One year later we see a big version come out, and then the outcome is slightly better. But today, you see that the major large language model labs, OpenAI and Anthropic, they are launching new things in weeks. And we are seeing more and more impressive work coming from AI agents. So overall, I want to say this time is different, if you think about the projection, if you think about the speed of evolution.

Yeah, the speed is mind-boggling. There was a time early on in the internet where things would be released once a quarter, once a month. And to your point, we are seeing this transformation almost happening every other week. There was a gap in the product last week. It's closed the next week. And again, it seems like it's only accelerating from here. That's a very good point.

100 % and it's a very exciting period. And then if we go to crypto part, you know, I was fortunate enough to join JP Morgan’s crypto team. And when Jamie Dimon was talking about Bitcoin is worthless back then where I was in that team, I was reacting to like, look, what we are doing here if Bitcoin is worthless, right? But today you see that people are talking about that. So the point I try to make here is back to 2017-18 crypto is the wild west. It's a new technology. People are thinking, oh, that's interesting, but there is some fraud or there's some speculation. What's going on there? Up until today, you see it's getting more institutionalized and more financial institutions, professional financial institutions are talking about that, learning this technology and aggressively adopting that. So what I'm seeing here is a kind of like convergence where the real established financial institutions start adopting the technology of blockchain and try to utilize and support crypto assets. On the other side, the crypto company, especially a company like Ripple or Evernorth in this case as well, is getting more institutional, trying to serve the wider public, trying to make the real utility and try to make the entire framework, including risk management, more and more robust for crypto assets.

And during your time at Ripple, do you think you were in the payments part of the product? If you had to do it over again, would you see applicability of AI in terms of cross-border payment management?

Back then, unfortunately, the AI was not that advanced. But if I have the crystal ball, if I could go back to the past, I think number one challenge we're seeing when doing cross-border payment is where is my money? And in case for people, if people are not very familiar with cross-border payment solutions here, typically or traditionally, there are a lot of middlemen, intermediary banks, holding the fund in different geographics locations and release the fund whenever the customer needs it. The number one challenge for cross-border here is I'm walking to Bank of America, try to send money to Ukraine, for example. Three days down the road, I have no visibility where my money is and I have to contact banks asking different people. In this case, it can be fully solved by agents autonomously because agent has the API access to get every single piece of the data and run the analysis basis, they have the full visibility, and if you need something, they can tell you immediately. That's just one example here how agents can fundamentally reshape the financial ecosystem, especially associated with the process with a huge operational overhead.

That's definitely an interesting answer. They rethink the past using today's technology. Again, to your point, things are just moving so fast, we must have to rethink, are we doing things the right way? Are we building the right kinds of product in the right way, given how fast things are moving? So we talked a little bit about this as like, there's a lot of benefits to AI. They can act autonomously, they can do more and more. Agents can do more and more as the weeks go on and then crypto. It's magic money moves like this. You can trace it the velocity of payments and transactions are Almost limitless. If you if you look across all blockchains a lot of capacity there but for a corporate treasurer, they're not the most risk-taking group in in the company and you know there's drawbacks. Before a human can, if you're using stable coins or crypto, the speed of which it moves could also be a little bit, it could introduce some risk. By the time a human comes in and even looks at it, it might be too late. And then on the agent side of things, we don't quite know how and what the agents are going to do to the full extent without the right kinds of risk controls. And who is that agent moving money to? If they're moving money to another agent, do we know what the risk profile that agent is? You and I had a conversation a few weeks ago and you introduced some of these concepts and one of those concepts about agents having their own identity. And I thought that was fascinating. Thought you can talk a little bit about that and how should the treasurer really look at the different kinds of risks with crypto and then agents and AI.

100%. You asked a very great question. You are touching the core, in my view, the financial ecosystem. So if I take a half step back, think about the entire financial ecosystem or financial solutions we are leaving day over day is a human-centric financial ecosystem. What I mean by that is if you want to open a bank account, the first thing you need to do is show your driver license or passport to prove your identity. When you try to initiate a transaction, you either need to explicitly click the button or somehow tell the audience or the bankers in front of you say, yes, I want to send out the money to give explicit consent. Why is that? It's because fundamentally, human is the financial actor of our entire economy and the sole decision maker. It was the case, but not anymore. What I mean by that, you also talk about like now we have a smart thing, smart entity, let's call it AI agents. This is the first time in human history there's a separate entity who is as smart or maybe even smarter than a human being. They're good at doing financial decisions. They're able to access the data 24-7. They're able to process a lot of information and generate insights. Now, how can we make sure that if the agent decide to do something, the agent also bear the liability or we set up a guardrail or framework to make sure that if something goes wrong, as human beings, as the final decision maker, we still got covered. That's why I mentioned to you Asheesh about the concept of KYA stands for Know Your Agent. It's a similar concept to KYC, but the logic are fundamentally different. So there are four, if I may unpack the concept of KYA, there are four critical components. Verifying the agent identity, verifying the human's intent, verifying agent's mandate based on human's instruction. Last but not least, handling the dispute, especially associated with the liability. And let me use a real life example to explain the four steps you can immediately get to the point why KYA is so important. So Asheesh, if you are driving your Tesla car and you are asking your your agent on your car, say, I'm thirsty. I want to buy a coffee. Help me to locate a very close coffee shop and then I will just drive through and pick up my coffee. Right? Your agent does all the reasoning, does analysis, and then eventually say, oh, I got $6. Let me try to purchase a latte coffee regular milk with Starbucks agents. So now if you're a Starbucks agent, you suddenly see not only a human being, but an agent raising their money say, I want to buy a coffee. And the merchant going to say, wait a second, who are you? Show me your identity. Number one, are you really representing Asheesh? Number two, what is the human instruction give to you? Okay, buy a coffee. Is it very clear? Are you buying a latte coffee, cappuccino, regular milk or no milk? Okay, you confirm that. Let me make sure that you are making the right payment method. You're spending $6, you're going through XRP Ledger, you are making sure you are paying me in stable coin with RLUSD. But one last thing I want make sure. If I give you the coffee, Asheesh picks up, he sip it and say, this is not what I want. I want to get my money back. There is a mistake. Who is bearing the liability? If I stop here, you're going to see that when agents become a very active actor in our daily activities, touching our money, facilitating transactions, it's inevitable to understand their identity and especially how to safely guardrail their behaviors when something goes bad.

This might sound like sci-fi, but like Chandler did tell me that he has an agent that works for him that orders the team boba every day at 2 p.m. and it gets delivered via DoorDash which is like just fascinating. Chandler's already living into this feature, but that's fascinating. Think you're talking about identity, but you're also talking about risk as a core pillar of the financial system. You know, most companies in fintech are actually like risk management companies. So I think that intersection that you talk about in your four pillars of risk verification and real time risk assessment, I think is super, super interesting. So let's talk a little bit about XRP, like why XRP, how does it fit in with agentic commerce and AI and finance? And when we were looking at different blockchains, the thing that just kept coming back to me is that, hey, listen, to build a treasury and to build DeFi you want something that's fast you want something that's programmable. XRP Ledger built for finance, stood out; you were looking at other blockchains as well and XRP stood out for you and you are at the forefront of this intersection of AI and crypto. Tell me a little bit about why XRP and how that fit in with your plans at t54.

So I have a twofold answer. Let me start with why we need crypto first and then let me elaborate more why XRP is uniquely positioned, especially with XRP Ledger. So first of all, we need internet of money for agents. Basically, you also mentioned the same thing, agents are 24-7, but another part is agents are API-friendly or API-native. Different than humans previously relying on UI interface to click the button, agents can do everything frictionless if they got an API 24-7. With that said, if you look around your US dollar, the physical bill, or you go to Chase Bank, which closed during the weekends, how could you ever empower your agents to move money around with such legacy infrastructure. So in this case, blockchain, stablecoin, crypto asset is uniquely, natively positioned to support or fuel our agent economy. Within that, if you look across all the chains, what is the criteria to decide which one is more agent-friendly? I think I have two major criteria in my view. Number one is cost basis. Number two is their financial ecosystem connections. First of all, talking about cost, we have been underwriting and assessing more than 20 million agent-native transactions already. That 20 million is by count. If you look into the dollar value, the average value of each agent transaction is still relatively small. We classify them as microtransactions. If the rail itself, settling the agent transaction, is very expensive, charge a high gas fee, low throughput, there's no way the agent are willing to use that chain. So XRP is super cost effective. When we do the calculation, I don't even recall how many zero in front of it. It's kind of like four or five zero point one cost when agent try to make a one transaction that's settling through XRP. So the cost, the cost effectiveness is the number one advantage for XRP. Other than that, if you look at the entire ecosystem, Ripple has a very good reputation by connecting all the financial rails cross-border and all the financial institutions. Bank of America, all these great partnerships. So when an agent decides why I need to put my money flowing through this rail is because I can seamlessly connect to more than six zero different destination countries globally for cross-border payment. Or I can utilize this chain to connect to a treasury management platform. Plus, I got this RLUSD and XRP ecosystem, who all can funnel different use cases, no matter on demand side or supply side. So I think in this case, XRP is also uniquely positioned to foster this entire agent economy.

That's fascinating. You know, the other question I have for you is, if agentic commerce, like it's predicted, takes off, the number of agents out there are far going to exceed the number of humans very, very quickly. Does every agent get its own ledger account? How will that scale? How do you envision scaling the infrastructure, not just from a throughput standpoint, but from the actual accounts that are going to be needed to support all this activity on different kinds of ledgers.

Yeah, so there are two approaches what I'm seeing so far. One is through virtual account concept. One is through all the physical custodians or physical wallets set up on the blockchain ledger. There is a trade-off between both between these two. But I think overall it's inevitable that we still need a human to oversee this entire infrastructure, this account architecture, and make sure that agents are fulfilling their job within a certain guardrail. Meanwhile, you also need to yield a little bit of autonomous power to the agents because humans may sleep, right? They need to sleep eight hours a day versus agents who never sleep, right? So in this case, when you are sleeping, agents see a big market shift or big events, and somehow you have to carve out some of the rights or kind of this power allowing agents to do something, react to some of the market risk. That's a tricky balance here where you can either introduce a separate account to re-infuse, or make sure that only this money can be utilized by agents. And that's another way to think about how do we handle this multiple accounts system.

Fascinating. Let's talk a little bit about, you know, Evernorth and potentially working with t54. You know, how do you envision this collaboration and partnership, you know, working in the future? Where do you think there's, you know, potential great synergies between Evernorth and t54?

I need to share a little bit of my personal story as a treasurer or as an AI product manager supporting thousands of treasurers in JPMorgan and also spend some time within Ripple as well. When I was in JPMorgan's treasury management team, we had more than 2,000 corporate clients, or even more today probably. And as you said, Asheesh, a lot of treasury managers are not that tech savvy, to be very honest. They are risk conservative and they are not very friendly to all the latest technology. However, their job is tough. They have a lot of operational work. They're super busy. As an example, if I walk you through a story of a typical CFO or head of treasury managers, day over day job. Number one, in the morning, 8 a.m., when they walk into the office. They open their laptop, the first thing they need to do is, OK, check my start beginning balances and see if I got any alert, any issues happened last night. And then they say, everything's settled. Phew. And let me try to plan for my today's cash flow forecasting. How much money inflow? How much money outflow? How much vendors I need to pay? How much data I need to collect in order to generate all those forecasting things? Then I need to move my money around. shit. This is cross border. It took two days. Should I plan ahead? And this one require manually fill in the beneficiary information. The last piece is a shortage white list. Oh, this person I never met. need to confirm that this guy is legitimate. Anyhow, after you do all the preparation work, you also need to physically move the money. You log into your bank portal. You click a few buttons. You double check your number. You may have a make or check process. At the end of the day, you spend five hours, 10 hours a day trying to go through the entire process and you need to repeat it every single day. If I pause here, you're going to realize how much operational overhead each trader is managing, and to some degree, how inefficient and error-prone. Now going back to Evernorth where you guys try to manage the crypto asset, it's even more challenging. It's 24-7. It's frictionless, but at the same time, blockchain is irreversible. If you make a mistake, it's big. Sometimes it's really hard to get the money back. So all those things will lead to one common pain point - the operational overhead and efficiency within treasury management shall be fundamentally helped or supported by AI agents

Yeah, I think what was interesting about that answer as well was that a lot of those treasury tasks are, you know, it's a high pressure job. But what I'm also hearing is that it's repeated. A lot of the treasury operations and those tasks are repeated and AI can really, really help there. So I thought that was interesting. And then still waiting in 2026 for days for a wire to, it's just like still crazy to me, but I think that that was a great point you made as well. So I want to sort of close out and talk a little bit about the risks as well. And we talked a little bit about that at the beginning that the treasurer doesn't wake up every morning and think about like, what innovations should I bring to the fold, right? They're like, hey, like let's just make sure payroll gets out the door. But a couple of things, like one, with AI, there was a story a couple of weeks ago about a Meta safety professional at Facebook Meta, AI deleted, she had an AI agent and it deleted all of her email accounts. And it went sort of rogue and she woke up in the morning all angry and so forth. And then again, on the crypto side, it's great. to your point, moving money that's taking days is horrible, but then instantly, you you got that risk. How should a treasurer think about getting into agentic commerce or using AI in their treasury ops? Is this something that they really should like be diving into in 2026? Or should they wait until things get figured out? What are the pros and cons of jumping in versus waiting?

Yes, if I may have a say, we have to jump in with a little bit cautious or a little bit risk management mentality in our mind. Number one, the AI is evolving so fast. Anyone open your GPT or OpenAI solutions or Anthropic's agents, right? And you're going to be impressed by how much work, how smart they are. And this is not something happening, it happened. And this is something that you should aggressively adopt into. But on the other side, literally as you said, the elephant in the room for treasury manager is risk. At the end of the day, human hate losing money. Human hate unhappy paths. We hate uncertainty. So instead of thinking about agents are bringing more risk, I would flip the switch by talking about how agents actually can help you to remediate the risk. A lot of people doesn't trust Tesla autopilot because of the thought it's a machine, but the insurance company is offering a lower quote for cars with autopilot features in place. Because statistically, humans are actually more error prone. So with that said, I think I really appreciate or respect Evernorth is aggressively taking the approach by thinking about how to incorporate AI agent solution into your risk management daily task. Because fundamentally introducing that process not only help you guys to improve the efficiency and also help you to remediate the risk, especially within this 24-7 crypto volatile market.

That's a great answer. And I think that's your t54's approach of looking at risk. You don't hear that a lot from AI companies today, but to really, really dive into the intersection of crypto and AI, I think you have to start with risk and then build out from there. That should be the core pillar. That's why I like your approach quite a bit. This is a fascinating conversation. You know Chandler is, you know, one of my favorite people building in this space. I enjoyed my time working with Chandler at Ripple and now, you know, working together to figure out how we can use AI in treasury management here at Evernote and t54. So thanks for coming on today. We'll have more of these conversations and learn from each other as we go into the future. That's a wrap for today. All right, thank you. Chandler, anything you want to add at the end?

I have one more thing to say, a little bit small teaser or something that we are actually working on. We haven't fully published yet, but before we are actively working with some of the AI researchers in the industry, try to understand how can we better quantify the risk associated with agents financial activities. So today, if you look at the market headline, you see some agents make mistakes. You see the agents performance is 99 % as good as a human being. But very little people talk about if agents make mistakes, how should we fundamentally cover the loss? But even take a step back, how much loss an agent can generate, if they make a mistake. So fundamentally, I see there is a huge innovation area that needs to be filled in, which is how to fundamentally quantify and give a price associated with agents downside risk. Once we have that, we can introduce more product, guardrail, and solutions, and fundamentally build up an economy to make these entire things up and running. So look forward to sharing more in the future once we finalize our research, but this is something that we have been heavily working on.