The AI Impact on Banking and Finance with CEO and Co-founder at Treasury Prime, Chris Dean

All anybody's interested in today to an extreme. I, it's not the same, but one, there've only been 4 SaaS IPOs since 2021. There's only been 4. One was OneStream, which is a platform for CFOs, and they're about $600 million. They IPO'd at $500 million. And it was very interesting to me, the founder CEO said, and that they're being very cautious on AI. The CFOs don't want it. They don't trust it. They don't want any hallucinations. At the same time, there's been machine learning and AI there since, since, I don't know, since the '60s probably, right? Yeah. So what is AI in banking and banking as a service?

Banks are generally interested but scared. I don't know anyone who's using anything at any interesting scale. The fintechs, one of the things the banks are scared about is that they become AI agents, which treat the banks as a commodity. The banks love this long-term relationship with the end user, right? With the consumer. They really like that. And what if you just had an app that rebalanced your bank accounts every day? And there was just— your credit card just worked magically to pull money from whatever account it needed to. They are terrified.

So welcome everybody in the global Saastr community. Have a good friend, an OG part of Saastr here, Christine from Treasury Prime. There are a few folks. We're going to talk about a bunch of things related to BaaS, banking as a service, fintech, where we're going, what happened to my money at Silicon Valley Bank. Could things blow up on us? Where AI is in more conservative parts of the world. But Chris has been doing this, has been doing hardcore banking as a service since before anybody knew what this was. It's his second startup doing it. The first was bought by Silicon Valley Bank in the early days. Now he's got a YC-backed company that's raised over $50 million that has many top banks and fintechs and a marketplace for them. And we'll get into that. But Chris, thanks. Thank you so much for joining us.

Thanks. Yeah, I love Saastr. Here's my only Saastr swag I still have.

Oh, the fidget spinner.

It's a fidget spinner.

I love mine too. That is, that is a classic. That is a classic. Okay, I want to talk about some stuff and also I want to talk about a little bit about stuff that has blown up in the space. This whole Evolve Synapse thing, it's pretty nerdy for most people, but about $100 million of folks that deposited money in startups still remains at large, doesn't it? Do we, do we know where this money is? It's, it's, it's still unresolved, right?

It's, it's unresolved.

Yeah. This is crazy.

Different parties will say, oh, it's, it's over there, but it's always on someone else's room, right? It's never my room. It's someone else's room. And it's just pointing fingers, pointing at each other. No, no one knows where it is. I assume there's some level of malfeasance there, but I bet it's really small. Someone stole some money, but just tiny amount. Mostly it's just reconciliation is hard to make sure that books balance. It's hard. It's really hard. And they didn't do it for a long time and it just leaked out.

So let's talk about this for a minute. So give us some context. In a corner of the fintech world, this is a big deal, but most people haven't heard about it and I'll probably get it wrong. But Synapse was attempting to be a tool for fintechs to enable banking on their platform, especially for some low-end products, some consumer products. And Evolve was their bank, right? A real bank, a chartered FDIC-chartered bank, correct me if I'm wrong. And Synapse couldn't keep track of its money. It got over its skis by almost $100 million and no one knows where a lot of this money is, right? Is this a summary? And like, where is— and then tell me the summary and tell me where the hell is any of our money anyway?

Yeah. So Synapse was pretty much the OG US banking as a service company. They have an API where you can say, I want to open a bank account and then you could do that. There are rules about how you could open it, but you could open it and say, I want to send a wire or I want to issue a debit card, stuff like that. You can do all that stuff. And they managed it. And in our— my world, those are called program managers. Basically, they want to pretend to be a bank, but they're not a bank. They need a bank buddy, they need a partner. So they got Evolve, which was— it's like you said, it's, it's an FDIC bank. Pretty much every bank's an FDIC bank. It's tech. Its main regulator is the US Federal Reserve, the Fed. But basically over time, Evolve and Synapse didn't like each other. They fought constantly and eventually Synapse, the software company, the program manager, started to add other banks and they added 2, 3 banks and they moved money around between them pretty indiscriminately in ways that I would say this is not my real— legally it's not my field, but seemed fairly illegal to me. And so what's happening now is like One bank says, I don't know where it is. It must be over to that other bank. And the other bank says, no, I don't have it. It must be with you. And you saw, well, let's look at the ledgers. Let's look at the books. And they're a mess because Synapse was always a mess and they went bankrupt.

Yeah.

I looked at purchasing them and I backed slowly away. Once I opened up the kimono, I was like, yeah, no thank you. Where's the money really though? If it's in a bank, it's fine. But a lot of stuff aren't banks. Evolve. For all its troubles, actually an okay bank. But also I could have said the same thing about SVB, right? SVB, I worked there for a couple years. I ran their fintech group there for a while.

Yeah.

It seemed completely solid and ready. They were constantly fighting with the regulators, and at some point they just got over their skis. They were relying on certain deposit interest rates. They were relying on people like you and me to keep our money there and never move it. And the interest rates are down. They couldn't pay the yield that they needed to pay. And people like you and I— maybe not you and I, but people like us— moved their money and all of a sudden there was no money there. And whatever else is true, whatever competent or incompetent thing SVB did, you cannot run a— you cannot run a bank without deposits. That is the whole bank business. You buy and you sell money. I buy money and take deposits. I sell money in terms of loans and I make money on the difference. So that's how you do it. If you have no money, You have no deposits, can't make any money.

Yeah, that moment. There's some— there is some saying that maybe that I made it up, but if you do startups long enough, there's a black swan event every 4 to 5 years, right?

Yeah. Yeah.

So for our little tiny team at Saastr, March 2020 was one. We lost $10 million when we canceled Saastr Annual. And the other one was we had $10 million at SVB and we woke up and we log in and it's zeros. I was still— I couldn't— it's zeros. So Horrible. So I want to ask, I want to ask about all of that for a minute and then go into the future. But before we get there, does, does this, is my money anywhere? Like, how does this ledger system exist? Is, I had $10 million at SVB and then it's zero. Okay. Let's put aside FDIC insurance and other questions, but where the hell is my money? SVB loaned it out. So is it, are these, are there mainframes in the cloud that know where every, does this money even exist? This is where I get confused, right? I sold I tell you, when I sold my second company to Adobe, I was terrified to even wire the money anywhere. I'd never had that much money before. And I'm like, what if it goes to the wrong place? What if it disappears? And I put it in Vanguard. Yeah, because— and today I would put it in Wells Fargo because I'm like, at least I think Vanguard will survive. I didn't even know where to put that much money. How does this ledger really work?

So people think two wrong things about banks generally is they think banks are very secure and know what they're doing about— they don't. Bank secure, bank secure is like not very secure. And the other is that they, that your money's just sitting there in a big vault somewhere, whether it's like in dollars, physical dollars, or there's bits in a computer somewhere. But that's not how our banking system works. Basically no banking system works like that. It's fractional banking.

Yeah.

It's like that It's a Wonderful Life movie where like Your money is in the— where's my deposits? It's in the mortgage that I gave so-and-so to build their house. So it's in their house and they're paying it off. So if you had $10 million at SVB, likely $9 million of it was out somewhere. And so all of a sudden when everyone pulls their money out, normally there's just a little bit of money movement and they can handle it by having enough reserve. They have a little buffer. But if all the money, if the buffer disappears, they're screwed. Yeah. Where does their money go? Then they have to call the government and say, yeah, we're stuck, right?

Yes.

Call some bigger bank and say we're stuck.

Okay, listen, I don't want to spend all the time on this, but yeah, just to be my money therapist, when Adobe wired me a big check back in the day to Vanguard, did it really wire money? Here's where I get confused. Did money change hands? Did actual money— what the hell happened when that $5, $10— I don't care, right? But when it's your life savings, when it's your big win, What actually happened with those proceeds?

So it's turtles all the way down. There's like stacks of things recursively stacked on top of each other.

Yeah.

At the bottom of this, the stable base is the US Federal Reserve Bank. Every US bank, US commercially chartered bank, would say like, what, 4,200 right now, has an account at the Fed. Roughly, you can think of an account number as like the routing number. It's not exactly right, it's mostly right. Okay, so an entire bank's all their funds, all their billions of dollars will be in that one bank account at the US Federal Reserve. When you send a wire, what happens is there's a Fedwire protocol, which moves that many dollars from the account of JP Morgan to the account of Vanguard. You know, there's all sorts of extra little metadata stuff that says, by the way, that's Jason's money, don't lose it, right? And then it's up to Vanguard to increment your balance properly. But that's how it is. It's all electronic. It's all in the cloud, like you said.

Got it. So the meta ledger is guaranteed by the Fed for the banks, right? And then the individual— and this is maybe this part of where Evolve and Synapse got in trouble— the individual account holders, the individual accounts, it's up to the bank or the non— I don't think Vanguard's a bank— the non-bank to keep track of that. And that's where you can go awry, right? But the Fed is guarantee— keeps track of the total amount for each of these actors.

Yeah. That's right. Let's pick Wells, right? They, I don't know how much money Wells has, say $500 billion. I don't know if that's true, but like they have an account at the Fed. Actually, it's probably spread across 20 accounts, but pretend it's one account. That's their money, right?

Yes.

It's, it doesn't have $500 billion in there because that's just their total. They, it has maybe $100 billion in there because they've loaned out the rest, sent to somebody else.

For sure.

And they have a core banking system, which is basically just a ledger. It's a database. That says, okay, Jason has a dollar, Chris has $2, Jack has $7, etc., etc. And it's up to the bank, Wells Fargo, to make sure that the number that's at the Fed matches the sum of all the numbers on their bank accounts.

Yes.

And that act of making sure those are the same is called reconciliation. And if those are the same, everyone's happy. The books balance. It's just regular keeping the accounting. I can tell you every day at all the banks I work with every single day, there's a person who has to do something to reconcile those books. It is hard.

Every day. Every day. Every day. Every day they're reconciling.

Treasury Prime's unusual. The banks don't really love us for this either, is that we reconcile 24 hours a day, every hour, and we try really hard to do that. We were off once by— can't remember what it was anymore. It was like $10,000, $20,000. Yeah. And we, we were attached to $10 billion. So $20,000, it's a rounding error. I had a heart attack. Ah, what's going on? And of course, one of my folks said, oh, we found it and they fixed it. It was fine.

Okay.

It was all hands on deck for, I don't know, 5 or 6 hours.

So you're checking it hourly. Once in the almost decade-long history of the company, you found a reconciliator, a 10-K or a 5-figure delta.

Yeah.

And you tracked it down and addressed it.

Yeah, yeah, yeah. That's right. And often these things are Fine. Sometimes it's just timing. You send a wire, you want the wire to work instantly like it does in the movies. It doesn't work like that. Yeah, it goes to someone's desk. If it's a big enough number, like from Adobe, that they'll have to call and approve it. Adobe will, the bank will look at it. Person will look at a big enough number to say, yeah, we got to look at that. That all takes time. And sometimes you'll say, I think I have an extra dollar in the account, but it hasn't actually been wired yet. And so just, you just wait and that handles reconciliation. But you can't do that. You can't just hope that it'll be better in the future. You have to say if it's not good in an hour, that something's bad. If it's not good in a day, then something's wrong.

Okay. So for anyone that wants a platform, Treasury Prime's only had one issue since inception for 10-K and reconciliation was solved that day, right? That's the track record across 10 billion deposits, right?

Yeah. Sometimes we argue with the banks about— they say your software is wrong and I'm like, I doubt that, but then there's a big hullabaloo. But there's not— I don't think we've ever found anything serious.

Okay, so let me ask the other end of the spectrum. Let me ask the question that I think has been underdiscussed in all the articles. Okay, let's talk about some products which I use and enjoy using. Okay, I'm going to pick 3: Brex, Mercury, and China. A couple of questions. What if they went bankrupt tomorrow? It's not going to happen, but imagine they went bankrupt tomorrow. What would happen to the assets there?

The most likely thing that will happen— I guarantee the most likely thing that will happen is that none of those are banks. Those are all neo banks.

They're not banks.

They're not. None of them are.

That's the thing. But a lot of people think they're banks, don't they? Chime became super clear. We're not a bank. When people started doing this a year ago, I remember, right? But we use Brex and Mercury every day, right? They're very easy to use. They're very easy to use.

Yeah. The UX is amazing. And banks' UX is generally terrible.

Terrible.

My corporate account's at Wells, Wells, and I use it all the time. It's bad.

Horrible for what we want to do, right? Work quickly. But I think most people think Brex and Mercury are banks, no matter what their website says. I think they treat them as banks. I think they're marketed as banks. I think when Brex said move your SVB dollars over to us, it felt like a bank, right? And there's a convenience there. But SVB was also, for folks that have been in tech for a while, especially folks doing like CEOs and VCs, SVB was super convenient too. We paid the price, right? We almost lost everything. We almost lost everything. It's not going to happen. But I'm picking 3 leaders that won't go under rather than Synapse, right? But if Brex, Mercury, or Chime, which would not go under, if they did tomorrow with no notice, what would happen? We've got like $3 million at Brex. I don't know what at Mercury. What would happen to my money?

In theory, and this is most likely what would happen, is Brex goes under. The bank still has it. And there are 2 banks there, Column and Fifth Third. Fifth Third's a big giant bank in the Midwest. The bank would say, we need a copy of everyone's balance. They need to know Lemkin balance or Sastre balance, right? And then they would say, oh, you got $3 million. They would write you a check for $3 million and they— not a joke— would probably put it in the mail and send it to you and hope that eventually it'd get to you. That's in theory what would happen. That's what was supposed to happen at Synapse and Aval. But they went to the books at Synapse, the ledger that said how much everyone had. and no one, they said, you said it has $3 million. There's not $3 million here.

Yeah.

So what do you do? So as long as everyone's a good actor, it's fine. And that's the most likely case because actually Brex and Mercury and Chime are all very well run. Yeah. Like Brex used to be, have the best, by far best neobank. They've changed how they do it, but they used to by far have the best one, best design.

Okay. Last nerdy question. Then I want to, I still want to stay nerdy, but I want to abstract up one level.

Right.

We'll see if you agree with me after all this. Would you put 100% of your own money in a non-DCIP?

I wouldn't put it in one, but I wouldn't do that anyway. It's like all about the backup. I wouldn't put all my money—

now you'd have multiple. Maybe that's the right question. You would have multiple banks.

I always have multiple banks. It's funny, on my— on the Treasury Prime business.

Yeah.

Every single one of our big clients has multiple banks. And if you use your examples, Brex has at least two. Technically, I think might have 3, but at least 2. Mercury's at 3. They still have stuff at Evolve, if I remember right. And Chime, there are 2. There are 2 banks. Everyone has multiple, not only for the, we're scared that they're going to, the bank's going to collapse, reasonably, but also so they just economic, so they can say to Bank A, I'm getting a better deal at Bank B. I'm going to start moving more of my stuff over there. And then Bank A says, all right, pricing pressure, we'll make a deal for you. And that works.

And did— and that has accelerated. Do people pre-SVB always have two banks, or is that accelerated since the SVB, First Republic?

So the big people have always had multiple banks. It just moved that big number down. So a lot more people have it, right? It used to be like you needed, I don't know, $50 million in deposits, which is for a bank small, for fintech large, you had to have that. And now it's more, oh, 20, 10, sometimes 10.

Yeah.

Yeah.

Okay. All right. Next one I want to talk about, I want to talk about the rebound of fintechs, how that intersects with Treasury Prime. I want to talk about AI and then, and then a few more detailed things. But you've had a lot of different— you— Treasury Prime's got a couple of products. You're connecting fintechs and banks. You're providing a service for that, right? A platform, a marketplace and more, and a software platform. You've watched some of the quality of fintech deteriorate, right? Post-2021. But now it seems like it's back more like it's, it's back with a vengeance, right? On many levels. So what's— yeah, what's the pulse check? What are you seeing across fintech and how is it like impacting you at a product and customer level?

Yeah, it's night and day. So December 2023, I went over— I've done the end of the year stuff that you do. I'm looking at all the books and where we like, we killed it and where we face planted and I started just modeling stuff and I'm like, there's a whole line of business we have which actually either— which generously breaks even, generously. But in practice, we lose money on every deal, right?

Yeah.

And, and I looked at the— I looked at what's happened in the industry. I said, this whole industry is going to faceplant at least for the next 6 months, maybe for the whole year, maybe longer.

What do I know?

And so we made some hard decisions. I ripped— I got rid of like a whole bunch of wonderful people who, like, had done nothing wrong. It just— that was the market and that was right. I didn't think we closed a single bank in 2024 and we closed one at the end of the year and that was good. But since then, I can tell you there's multiple banks in play. Everyone wants this stuff and they want it partly because the fintech side is back. There's like people with interesting ideas. There's established companies, some SaaS companies who want to add banking like they used to in the good old days. And it's just taking off. And like part of its capital is around, but part of it also just— there's been a big switch with how they're regulated with this new U.S. administration. They're like changing how all the regulators are running. And mostly the bankers are feeling much more generous and able to do things, which they weren't feeling like that last year.

I see. It felt like the banks were at a huge walkaway, especially from anything vaguely started being fintech, but you're now you're saying they're back. They want the business.

They want the business. They're still— they're not— they haven't stopped being bankers.

And like, yeah, I don't know any vertical SaaS company in particular. Vertical SaaS is also quietly on fire, right? Shopify, Toast, Procore, all crushing it. I don't know a single vertical SaaS company that doesn't want to add some type of banking or fintech product, right? At a bare minimum, they're adding some sort of payroll, but they're also, they want to add credit, right? They want to add banking. They want to, it is every vertical SaaS wants to be the ERP for its industry, right? Whether it's hairdressers or construction. And everyone's decided that banking's part of that, right? You want banking and payments are part of that stack, right?

Yeah. There's a whole range here. It's like just doing simple payments. Like you can do that at scales you could never, you couldn't do before. Couldn't do 5 years ago.

Yeah.

Very sticky. For a lot of the, a lot of the SaaS customers, the end users, right? Because they like to get used to it and they don't want to switch around because it's sticky product. All the way up to a full banking experience. I want a bank account. I want to do some lending. We have 3 small, as well compared to Procore, small, right? Construction startups. They're all doing variants of the same thing. They don't actually really compete against each other, but they all have decided that they're going to— they added payments and then they added bank accounts. And now at least 2 of them have added lending since then. And I could just see them growing and they're just growing steadily. SaaS companies do, they go up, right? That's great. Yeah.

Shopify Pay is— Shopify is exploding at $11.2 billion in revenue, exploding. It's all Shopify Pay.

It's all—

it's always been a fintech, right, for many years. But managing that stickiness of making the whole— and Shopify does everything, right?

To that, this is a pitch I give all the time, but Imagine to the Shopify user, right? Not you or I, but like someone who's a merchant, you start up their store. If I could just say, if I come to you, Shopify, so look, I could just do everything for you, including handling the money and handling the payments, and it'll just be dead easy. You just click this button and we're good to go. And said, wow, why would you do that? Because I want to be helpful. Shopify is all about making the merchant successful, but also I'm going to take a few bits here, there, like a percent of a percent. Just here and there. Yes. And you won't even— it's so small you won't even notice. But at scale, that's huge amounts of money. You have to be at scale. But they are. It's huge. Huge.

Yeah. I really don't think— some folks notice it, right, where the margins are very low in the business. But for the most part, most of these vendors are pretty successful in— it's not just hiding it in the— it's just, it's not enough to matter, right? The value is so much higher than the little bit of take that they make, right? That it's at the edge of free. For the end customer.

Sure. The merchants would like gladly go somewhere else that was even cheaper than Shopify with a good product. But frankly, even if they go somewhere else, it's probably not as good a product and they're going to charge them basically the same thing anywhere, maybe slightly more. So yeah, why don't I just stick with the good one who's a good deal and can do everything? Yeah.

And then let me just— let me tie it back to Treasury Prime and then for the last couple of minutes, let's pull it back high. But the 3 construction, the 3 smaller procores that aren't direct competitors that came to Treasury Prime. How did they come? What are they looking to do in particular that they, that they needed you for? Where's the evolution of their product in terms of the services? Where's the insertion point and what are they trying? Where do they need banking as a service, a banking API to do what they want to do?

Sure. All different flavors of this, but remodel, like the new HomeRemodels, right? So what you do is the customer, it's a SaaS product. The customer is not the homeowner. The customer is the general contractor who's doing the remodel. I, what they do is they say, okay, guy doing the remodel, you need to manage like supplies, you need to manage payroll, you need to manage payments in and out of it. You need some sort of card so you can go to Home Depot and buy lumber or whatever. You need that to manage all that. So if I just give you a bank account, then the accounting becomes easy. I can just move money in and out. The homeowner can get an app that says, here's how much you owe, here's what you need before the— on the next stage. And it's all just laid out so everyone can see it. Very convenient, very sticky SaaS product, right? But it couldn't exist unless it was easy to open a bank account. So the Treasury Prime shtick is it's easy. You just do it. And they're like, people start out with— they call me and they say, I have this very hard problem. They have all these tricks they're going to solve. I say, or I tell them, you could just open a bank account. And then you would have access to a bank account and then you could just put the money there and store the money. You don't have to do all these, oh, I'm going to get this license here, become a PayFac or whatever. All these inside things like, don't do that. The bank has figured all of this out 100 years ago.

Yeah, got it. And so they need to provision— I should know this, but I don't— they need then to provision 100 accounts for 100 projects for the general contractors or in that use case, What are they repeatedly doing here? What are they flowing through the system in the API?

The different ones actually do a different one. Yeah. Creates a bank account per project and the other it's per user. So per general contractor.

But in each case there's a separate bank, you need a separate bank account to make this work. Yeah.

Right. Separate. And it's a real bank account. And the thing is it's a bank account that's owned by the general contractor. It's not owned by the SaaS company, that's software company. And so it's a comfort level for everyone and they're, the contractor's happy because it's his money. It has his name on it or their name on it. And the bank's happy because there's an end user over there who they can talk to if things go south.

It's their money.

Everyone's happy with it. And the banks always fantasize too that they can do upsells or cross-sells of other services. And generally that doesn't work because the end user, the contractor in this case, they're loyal to the SaaS company. They're not loyal to the bank. They don't even know who the bank is half the time. Yeah, I feel like, where's Mercury? I don't know.

I don't know. So here they are. Yeah. You're serving general contractors. These are expensive, important projects, right?

Yeah.

And so they can just go into the app. They, in, in a click or so, can deploy a new bank account for their customer that they own. It's their bank account, right? And the customer just wire or however they— or transfer over the funds and it's all done. It's very sassy, very elegantly. These accounts are provisioned, right?

Yeah, yeah, yeah. And this is a very common pattern throughout our thing. Like, we have a separate group of people who are their investment accounts. So AngelList, Sidecar, Sidewards, SYD. And people want to get together to invest in something. They'll create an LLC, they'll create a bank account attached to that LLC. They're all GPs, basically. They're GPs and LPs, and they'll put money into that bank account, and that money will sit there until they're ready to do the investment. And so the capital call is like put money into the account and then you, you deploy it later and they open one account basically per deal and that's fine with me. And the thing that is kind of an amazing thing for the banks is the banks are in this for a different reason than the fintechs. The banks want deposits and they want fees. But really, that's the question I was going to ask you.

How much do the banks make off these types of transactions?

A ton. And like, they, they want deposits and these investment things like the AngelList and Sidecars of the world, you might think, oh, there's no money there because how often does money sit in disaster investment accounts? You make a capital call, you have to call them 5, 10 times, and eventually they'll all show up and then you'll wire out your million dollars or whatever it is.

I always thought it didn't make much money.

Yeah, but the funny thing is there's so much of this and the money always sits there for a couple, 3 weeks that in aggregate it's a billion, it's billions of dollars in deposits. And that's real money. And you know who could— and so the clients make yield on that. There's interest and they make some money on that. Yeah, but the banks also make money on it. And that's, that's Shopify. It's a volume game, right?

Yes.

They can get $1 billion from a fintech or a group of fintechs. They can get half a percent on that. And that's real money. And then if they can double, that's double that money. And then it's on top of whatever fees they have and banks are really good at using deposits to make money because that's all they buy money and they sell money. They've been doing it forever. They're really good at it. But today it's hard to get deposits because they're competing against all these fintechs who are frankly better marketers than they are. Always better marketers than banks are.

Yeah. Okay. Last or penultimate, depending on how we go, question I want to get is—

hit me—

is AI. And obviously it's frankly All anybody's interested in today to an extreme. I, it's not the same, but one, there've only been 4 SaaS IPOs since 2021 so far. There'll be a lot more coming. There's only been 4. One was OneStream, which is a platform for CFOs, and they're about $600 million. They IPO'd at $500 million. And it was very interesting to me. The founder CEO said, and that they're being very cautious on AI. The CFOs don't want it. They don't trust it. They don't want any hallucinations. At the same time, machine learning and AI at a higher level has been part of banking for a long time, right? If fraud— if nothing else, fraud detection, like we've been doing a lot of the stuff we know comes out of early fraud detection of how do you analyze trillions of fraud and trillions of transactions. There's been machine learning and AI there since, since, I don't know, since the '60s probably, right? Yeah. So what is AI in banking and banking as a service today? What do you— what do people want? What can you do with it? If nothing, is it—

banks are generally interested but scared. I don't know anyone who's using anything at any interesting scale. The fintechs, one of the things the banks are scared about is that they become AI agents, which treat the banks as a commodity. The banks love this long-term relationship with the end user, right? With the consumer. They really like that. And what if you just had an app that rebalanced your bank accounts every day and there was just your credit card just worked magically to pull money from whatever it found it needed to. They are terrified.

It's easy to see that happening, isn't it? That the agent optimizes everything for you.

It seems obvious it's going to happen, right? Like people like me are spending their whole lives like making this. I can tell you the biggest problem that banks have always in these day-to-day stuff is reconciliation, and that's on the deposit side. And then on the loan side, it's determining what's a good loan or not. And yes, risk of all that, a lot of machine learning on the risk side, the loan side. That's very traditional, the AI stuff. Not a lot of traction. Like, most of them, like, let's have a meeting, but they don't want to do anything. And on the recon operational side, there's not a lot happening. People are worried, like the CFOs are worried that if there's a hallucination and it says, oh, everything's fine, but it's not fine. Yeah, that's their job. That's— CEOs get fired over stuff like that.

Yeah, for sure. That is interesting that agents should be able to absolutely optimize how your money works for you. Yeah, it's an obvious use case, but one that is not necessarily in the bank's interest. So there's a collision coming there if it hasn't already started. The one thing I'll tell you, one thing that AI could do, but that everyone does wrong in this stuff and this agent stuff is no one can get this right on an after-tax basis. No one can advise you, but everything matters after tax. People recommend things that are very tax inefficient. Like it's just one example of optimize my entire life. I want it to be the maximum tax efficient. And say, I want this amount of liquidity, I want this amount of access, right? No, I don't even think a human can do that.

Yeah.

At a practical level, in theory they can. But for you as the home banking yield network, you can't get it all right on an after-tax basis for 20 people in your family. There's too much going on.

Yeah, there's too much going on. And you need to know stuff that basically no one wants to tell you.

Yeah. All right. Related to that last question for Treasury Prime, what is something coming out on the roadmap in the next 12 months that you're excited about? What innovation's coming out that you're excited about?

I just spent 5 minutes dumping on it, but we have an AI product coming out like it's next quarter from now.

Okay, what's it going to do? Tell me more about it. That's all we care about is AI. What's the AI product going to do?

It solves the operational problems that we see every day. Like we still do a significant amount of professional services for the banks, right? And sometimes we charge real numbers for that and sometimes we do it for free because that's just the deal. And I can tell you on our side, we're like, we're smart and we're better at this than most, almost everybody. But it's not hard. It's just time-consuming. You can't make a mistake. So we have this, we have this model where there's like a human trains it and then there's multiple AIs that watch each other to make sure that you're satisfied with it. And then there's a manual check at the end because the CFO operation of doing like money movement and making sure that all the accounts line up at the end of the day, which is like reconciliation, but not— that's— it's time-consuming and hard. And a person can do it straightforward because even though it's hard because there's so many details, what's good at handling all these details is a computer, right? And it just does that stuff.

And so this new product is going to use AI to supplement, augment, replace humans doing some of this accounting at the end of the day?

Augment it. Yeah, for sure. You still need, I don't think, I don't think they would buy it and I wouldn't want to sell it right now where I would say you can just turn off the lights and trust it. 'Cause I wouldn't turn off the lights and trust it instead. I would like, but if someone says, this is a hard problem, but I'm an expert, so it's straightforward. I can do it in 6 hours. So what if you could do it in 1 hour? Would that be useful to you? Yeah. Generally it is. Right. Got it.

Okay. So using AI to take out a significant amount of the human manual headaches of accounts at the end of each day. Right. Okay. Two and two final questions on that. So how to banks, to conservative banks, how are you describing this product? How do you communicate what you're doing to a real-world customer, not someone that's on social media?

Sure, sure. Operate. It's like we don't even— we only use AI as a second thing. We introduce it as an operational tool. Okay. It's a, it's an operational tool to help. It's not recon. That's not reconciliation. It's an operational tool to help you manage your books and manage your account list. And that is something they go, okay, tell me more about it. We use AI to do XYZ. And they're like, oh, it's AI. Why didn't you tell me in the beginning? Because when I tell you in the beginning, all you want to talk about is AI. You don't want to talk about the problem. And, and actually, for better or worse, I actually want you to buy the thing and use the thing. I don't want you guys to talk about it.

All right. Okay. Last question here. When you— just because it's interesting, when you talked about how much of an expert the AI has to be, right? This is an AI for experts, for folks that know this cold, right?

Yeah, that's right.

And could a third— a horizontal third-party agent do that on your platform? Or is this one of the types of AIs that you've got to own the data, you've got to own the system to win? Like the great debate, the startups versus the incumbents. In this little area, you're an incumbent, right? Because you have $10 billion worth of transactions on your platform.

Yep.

So was this— will your AI only work because you are the, the, the first-party repository of this data?

Yeah, because it's not our data. It's the bank's data. And the bank— we generate the data for the bank, but it's still the bank's data. And so this only works if the banks agree that they can use the— all the huge amounts of data we provide them for this AI tool. Yes. Operations.

All right. Thanks, my friend. And thanks for all the time. And we will catch up, if nothing else, on AI and banking and banking as a service in 12 months. Yeah.

Let's see. See where it goes. All right. Thanks, Jason.