How Are AI, Stablecoins, and Agentic Systems Reshaping Banking?

HOST (Brett King):
Opens the discussion with the central question: as AI, crypto adoption, and stablecoin infrastructure converge, which structural components of banking will fundamentally change? How should institutions prepare?

What Structural Shifts Will AI Drive in Banking?

HOST (Jason Henrichs):
Less bullish on immediate retail disruption. More focused on AI’s impact on:

• Bank-to-ERP integration
  
• Agentic systems communicating across financial infrastructure
  
• Automated corporate workflows
  

HOST (JP Nicols):
Reinforces that institutional changes will matter more than consumer-facing hype cycles.

Are Stablecoins Overhyped or Underestimated? What’s Real?

HOST (Brett King):
Stablecoins aren’t washing anyone’s socks, yet. But the real value is not trading altcoins. It’s:

• Contract-level operations
  
• Autonomous trade finance
  
• Smart-contract systems at scale
  
• Cross-border automation for supply chains
  

Large institutions like J.P. Morgan and HSBC are already executing large smart-contract-based settlements.

QUOTE:
“This is structural automation of money at scale.”

What Will Replace Legacy Core Systems?

Smart contracts won’t run on traditional ledgers.

HOST (Brett King):
Explains why:

• Autonomous finance requires tokenized operations
  
• Rollbacks (not chargebacks) will replace dispute systems
  
• Agentic AI will manage transaction errors
  
• Decentralized marketplaces may emerge for goods and services
  

HOST (Jason Henrichs):
Banks will soon operate three synchronized systems:

• Core
  
• Data layer
  
• Token/contract layer
  

Reconciliation across all three becomes mission-critical.

Can Smaller Banks Keep Up?

HOST (JP Nicols):
Large banks have the advantage of scale, but small institutions must adopt new thinking:

• Platform partnerships
  
• Marketplace connectivity
  
• Outsourced foundational models
  
• Avoiding the “build your own cathedral” problem
  

HOST (Brett King):
Digital-native players like Revolut, Nubank, and WeBank can leapfrog incumbents due to agile infrastructure.

Where Will Agentic AI Enter Banking First?

HOST (Jason Henrichs):
Internal operations will be the earliest adopters:

• Liquidity management
  
• Inter-branch settlement
  
• Automated capital allocation
  
• Internal smart contract workflows
  

A top global bank already implemented smart-contract Basel liquidity balancing, unlocking billions.

Does AI Improve or Threaten Financial Inclusion?

HOST (Brett King):
AI may erode inclusion if citizens aren’t digitally onboarded:

• Automation reduces the utility of cash
  
• Healthcare, education, and services become AI-powered
  
• Those opting out become excluded over time
  

HOST (Jason Henrichs):
AI could create a new caste system, where the digital “haves” pull further ahead.

QUOTE:
“AI won’t enslave us, inequality will.”

Will Capitalism Break Under AI?

HOST (Brett King):
AI disrupts the original purpose of markets: value exchange for human labor. With fewer human workers:

• Wealth concentrates
  
• UBI might reinforce inequality
  
• New economic models will be necessary
  

Pop culture analogies (The Expanse) illustrate the bifurcation of society.

How Should Banks Prepare Today?

HOST (Jason Henrichs): Two Critical Pillars

• Data foundations
  Every institution must modernize data systems now.
  
• Cultural foundations
  AI must be adopted across the whole organization, not just innovation teams.
  

HOST (JP Nicols): Add Customer Intimacy

• Jobs-to-be-done
  
• Deep understanding of customer context
  
• Using AI to strengthen relationships, not obscure them
  

HOST (Brett King): Add AI Literacy

QUOTE:
“There are two types of people now, those using AI to advance, and those who will be replaced by them.”

Raw Transcript:

Welcome to Breaking Banks, the number one global fintech radio show and podcast. I’m Brett King. And I’m Jason Henricks.

Every week since 2013, we explore the personalities, startups, innovators, and industry players driving disruption in financial services. From incumbents to unicorns, and from cutting edge technology to the people using it to help create a more innovative, inclusive, and healthy financial future. I’m J.P. Nichols, and this is Breaking Banks.

Welcome back to Breaking Banks. Well, it’s time for the hosts to get together, and this is a rare occasion for us. But there’s been a lot of conversation recently about how AI is going to impact banking.

And we see the crypto crash right now, and there’s discussion about stable coins and agentic AI. And all of these things are coming together in a moment of time when banking is potentially going to see some fairly significant structural changes, particularly in terms of the way we operate cross-border transactions and business banking as a platform. And we wanted to talk about the implications of that.

So joining me, joining together today is our other host, Jason Henricks and J.P. Nichols. So let’s get at it, guys. I guess one of the big debates is, as we move towards this era of more autonomous-based banking, what are the fundamental structural things that are likely to change? And how should banking be preparing for that? Let’s maybe kick off with that.

For as much talk as there is on the retail side of it, I’m less excited about the impact of AI on the retail side. I think there are lots of interesting consumer applications where people are using. I love what Grasshopper Bank has done in terms of integrating.

But this idea of the self-driving money and the educational component, I’m less excited about. I think it’ll take longer than expected. Where I think it’s going to be really interesting is how your bank begins to talk to your ERP, and your internal agents begin to work with external agents when it comes to your finances.

Yeah. And somewhere in there fits in stablecoins, right? We have been told they will do everything, including wash your socks. I’m maybe a little bit more skeptical about that.

And I’d like to talk about where we think they are massively overhyped and what that means for banks and regulators who are being told right now, get on the stablecoin train before it leaves the station. And let’s start with what they are actually doing today, not the marketing deck version. If you look at the bulk of stablecoin trades today, most of it is still being used in the DeFi world, in the crypto world for just trading with altcoins, converting from USDC.

Having said that, the largest trades on stablecoins today in terms of dollar value are not trades. They are single contract implementations. And we see JPMC experimenting with this.

HSBC did recently a very large transaction. I think it was $300 million or something trade based on smart contracts. If you are looking at autonomous business at scale, when you’re doing either supply chain automation or you’re automating cross-border trade in line with trading agreements, that is where the very big potential for this is.

And I agree with Jason’s comment. This is not retail. We’re talking about structural automation of money at scale, where you need to, when you’re going to automate that, you’re going to pass it off to a smart contract.

The smart contract’s going to do it in tokens or in stablecoins. And then at either end, you may have a mechanism for moving it back into a bank account and resolving the currency difference between the stablecoin and fiat. But the more autonomous contracts you’re operating, the more stablecoin balances and CBDCs you may just hold.

But I think fundamentally, if you look at the platform piece of smart contracting for autonomous finance for an AI era, I don’t believe that’s going to be run on existing systems. It’s going to be on net new technology. What do you think that net new technology looks like, Brett? Where is the gap that existing is not solving it and it requires new? Not disagreeing, I’m curious to your expert opinion.

Well, there’s a few things that you need to take place to run a smart contract. Whether you’re running it as a decentralized autonomous organization, a DAO, and you’ve actually got a fully autonomous operation, or let’s just say you’re implementing supply chain automation, and you’ve got a dark factory autonomously operating, and it is automatically sending out requests for raw material, and it is initiating a purchase order for that. It’s shipping goods on automated shipping containers on autonomous ships on the ocean.

These are technologies we’re looking at deploying over the next 10 or 15 years. Those are all smart contracts that are essentially going to be operating on some marketplace or platform. It’s not going to be running on a core system.

It’s not a ledger based system, where essentially this is what we’re going to take from the DeFi world of smart contracts, essentially, and this is going to become TradFi. In fact, I think even the term smart contract is probably problematic, because we want to talk about autonomous digital contracts, or autonomous finance contracts, or something like that. It’s a little bit more TradFi acceptable.

But that essentially becomes a token system in terms of token operation. There is elements of agentic AI operation, both in terms of choosing agents to go out and do the particular piece. Let’s say an agentic traffic cop, plus specific agents that are designed in purpose for the type of smart contracts we’re talking about.

This may develop into some autonomous marketplaces with some bidding capabilities for goods and services. But this is a new glue between agentic agents that we need autonomous finance to operate in. It’s not going to be using fiat currency, and it’s not going to be linking directly to the core.

Yes, it will use real-time payment rails, but not like we think of it today. It’ll be more like DeFi and crypto wallets, because you won’t have chargebacks the way we think about it on card systems. You’ll have rollbacks.

When an AI makes a mistake, you won’t be thinking about, oh, this is fraudulent, and we need to charge back this to the counterparty or whatever. You’ll be saying, no, we need to reverse that transaction that occurred three steps ago in the wallet. Fundamentally, I think the operating parameters of this autonomous finance looks different from what we think of in normal trade finance and corporate stuff.

It’s not just the systems, it’s the thinking. To pull on that thread of systems was with a group of regulators and banks last week, and we were talking about this. The realization on how we think about these systems, we’ve already gone from the core system to most banks have a data layer, whether it be a data lake, a snowflake, you name it, call it your side core, your data layer.

Now, you’re going to add a third to this, which is actually how your coins are moving and where they’re branding. They all have to reconcile. I’d say that next big leap where we’re falling behind is there needs to be an operational layer and a systems level thinking layer that are not as well developed as they need to, because we need reconciliation across all three of those systems, which is the record of truth.

Then how do we think about, like you had said, it isn’t a chargeback, it’s a rollback. What are the implications of all of this in terms of what does a chargeback mean? How do we unwind these transactions in a way? How does it propagate across all of the systems? Yes, true. JP? You mentioned JP Morgan Chase and HSBC.

Those are huge multinational banks. Sounds like all the things that you’re talking about coming to scale is also pretty important. What does this mean for smaller institutions? This is where it gets interesting, because obviously larger banks that have already trade finance at a level where they have multiple parties that work through the banks and they do that from a relationship perspective because of the ease of moving money and trade, they will have an advantage because it’s just moving to a different system of trade.

But for smaller banks where currently, I don’t know whether you’re using letters of credit or whatever the case may be, or you’re just doing ACH or wires today, this does look different because you now need to connect to marketplaces. Maybe it’s using Stripe’s smart contract layer or things like that. You’re going to have to have some partnerships in place for that.

You’re also going to have to take a more platformification view of banking. How do you connect as a platform with the businesses that you serve? And what’s the layers? As Jason said, those agentic layers that you can plug into different components of your operations. Well, and where are the points of intervention in this system? Where do we still want human in the loop to do some of this? And then the idea of thinking through this a couple of weekends ago and being into whiteboard, this idea of you need agents queuing agents to even surface this to the human in the loop.

And there becomes a proliferation of agents that need to do first line, second line, third line reporting and execution and remediation. I do love that quote from Yuval Noah Harari. I don’t know whether you guys have seen that.

It comes out of his book Nexus. But I wrote about it in an article I put on LinkedIn this week, the difference in this technology, in agentic AI specifically, compared with any other technology that we’ve had in the past, whether it’s computers or the nuclear bomb or solar generation or whatever, is that this technology has agency. This is why we call it agentic AI.

But it has agency, can go and operate on its own. So again, I think one of the cultural and technical challenges of this is we’ve never had banking systems that have their own agency. And when we’re talking about smart contracts and stable coins and all of that, how are you going to have that rigorous design process as a bank? How are you going to ensure that if you’re going to start operating smart contracts for your customers, that it is going to work? And I think the mistake many people will make in having this conversation is that they’re going to say, well, we’re going to need a lot of work to do this because chat GPT and Gemini and all of this has shown a fairly significant hallucination rate.

No, this is very, very different technology. And this is where the foundational models are really critical and where particularly community banks are going to have to get in with other foundational models or other banks to sort of create those foundational models, because those foundational models will be trained specifically on this very, very type of agency that’s required, whether it is agreeing on and sending an invoice or initiating a payment. It’s so you’re not going to have the error rate or hallucination rate.

But having those access to those core foundational models is sort of a basic layer. You talked about data like Jason, but access to sort of a foundational layer is also going to be key. Well, and I think this is why the concept, even for large institutions that aren’t thinking about the foundational elements, a foundation is not laid in a single day, right? Like when you think of the great cathedrals, how much of the actual work that makes it enduring is below the surface and happens regardless of what’s going to be built on top of it.

The foundations need to be laid and those take time to get done. And there is no perfect in those things that for banks and institutions of all sizes to begin thinking through those implications, even if they are far off, is just the critical challenge for 2026. And I think in that classic, it is going to take longer than expected and happen faster than expected at the same time.

We need to see institutions get their head around. I need to, even if it doesn’t feel like it is changing my business today, I need to start thinking about how I’m going to adapt and what foundations I need to lay. That was going to be my next question to you guys, you know, in terms of the work that you’re doing with smaller banks, is how do you help them to have that cultural agility and the technical agility to be able to explore this optionality? Well, I think there’s two things.

Jason, as you were talking about the foundation, you’ve got me thinking of Familia Sagrada. How many hundreds of years are we still building? That’s actually what I was thinking. Yeah.

So, you know, I think there’s two things. One, when we, the lens we like to use the most is jobs to be done, right? Who’s the end user? What’s the job they have to be done? And until some of these things become a better tool to get those jobs done, we’re still going to be in the foundation building business. But the flip side of that is jobs to be done are getting more complex because there are more capabilities out there.

We are seeing this even just with, you mentioned Stripe or Plaid, just being able to connect to accounting systems, to AR and AP and all those sorts of things. So yeah, Jason, you answered it out of both sides of your mouth and I guess I am too, right? It’ll happen both faster than we think and slower than we think. But what I worry about the most for the smaller institutions is I think they ought not to get caught up in the trap of building their own cathedral, right? Because they don’t have the time to do that.

They need to figure out what are the rails they can connect to today while also keeping an eye on where does this go in the future. But it’s not just smaller banks, JP, you know? I mean, culturally, how many of the top 50 banks in the world are really ready for this transition as well, you know, even culturally? Because, you know, let’s say, okay, you can start with a data lake, but operationally, think about the complexity of all the different agents that you could deploy in a smart contract scenario and how many different departments and silos that’s going to touch within a bank and how you’re going to get agreement across the board on that, you know? I mean, that’s the reality of how complex this may be to do. Now, for a Revolut or a NewBank or a WeBank, you know, or a Monzo or a Chime, you know, or a PayPal or an Alipay, this is not going to be as difficult because they don’t have those same silos.

They may have some silos, some technology silos and some legacy tech, but at the same time, you know, they don’t have sort of the entrenched product lines and compliance structures that traditional players have. So this is why I think that, you know, yeah, I mean, most of the fintechs were talking about a retail focus, but at the same time, I think this is where we’re going to see, particularly for SME type automation, I think these new digital direct players are going to have significant advantages when it comes to deploying agentic AI personally. I do still think on the trade finance and the big cross-border stuff, the larger banks will have an advantage because it’s just the scale of the business they’re doing and they can build those trusted networks.

But it’s that SME style stuff where, you know, I think it’s less clear. Agreed. Well, and I think the early applications are actually going to be internal for most of this, like intrabank work, you know, not a stable coin, but the equivalent of a smart contract.

Can think of a global bank that was using this for settlement of all its various geographic regions that under Basel II, what it needed for liquidity in each region, they programmed it on a decentralized ledger and made it a smart contract to actually figure out, you know, freed up billions of dollars in capital, right? Like to your point around the scale makes it worthwhile to go do it. And I think the way you de-risk it is you look at internal applications first. Here’s a question for both of you.

Do you think all of this we’re talking about agentic AI and stable coins and the like and DeFi, does all that help financial inclusion or does it hurt? Well, you know, I think the reality is with AI, you know, there is a very big issue with the fact that if you’re not digitally included, you’re going to find that the choices that you have narrowing, whether it is access to health care, access to travel, you know, access to, you know, education for your kids, you know, I do think, you know, let’s look at highly autonomous societies, you know, and how they’re going to process that. For example, you know, someone who prefers to use cash, I think we can see the trend, the more autonomy you put into payments, the less utility cash, you know, itself will have. I mean, of course, Elon Musk’s been talking about this recently, that he thinks AI is sort of the death of not just cash, but the death of money, because the more automation you put in the system, the less utility money itself has.

But I think we need to start thinking about that, you know, from an inclusion perspective. You know, there is going to be real issues. If we, you know, we want the most advanced health care system in the world, for example, then we’re going to use artificial intelligence to diagnose, you know, causes and diseases.

We’re going to use it for longevity treatments and improving health span, because that’s also the largest cost that we have in the economy as far as health care goes. But if you’re not willing to use those tools in terms of access to health care yourself, you’re going to find yourself increasingly isolated in terms of options. So this is something from a societal perspective we need to work through.

In 30 or 40 years, it’s going to be less of an issue because, you know, as we see with things like internet access and the use of mobile phones, you know, society does adapt over time, but it’s those periods of transition which are messy and complicated. Sure. I think I mostly agree with Brett in my position.

It’s slightly differently working backwards from the social implication, which is I think it could and it should actually solve the inclusion issue, but it won’t. And it’s for that digital has versus have-nots and those who have resources will move faster. And I do question Brett’s last piece that there is a digestion period and it will catch up.

I think AI could be fundamentally different in this regard, that there is no catching up like mobile phones and internet, that we actually see a stratification that is very, very hard to have anyone catch up, at least societally, when we look at a macro level. Individuals, yes, be able to move between that, but does it actually embed a caste system that’s nearly impossible to break? Look, I think that’s a really fair point, Jason. I do think the greatest – people say, aren’t you worried about AI taking over the world and robots enslaving us? And that’s not my concern because I just don’t think that’s necessarily a logical leap to make from the tech that we have or that we’re developing.

However, the speed of change is going like this, and adaptation of society and policy is going like this. So it’s that arbitrage, that gap between the technological possibility and what’s actually happening in terms of policy and ethics and regulation and all of that, because we’re not used to moving as fast where you have that issue. And I think that’s certainly going to be very clear when you’ve got people that can’t adapt in terms of use of AI in their career and their workplace and so forth, is that is going to hit in terms of inequality.

It’s going to be significantly bad. And even if you talk about something like universal basic income being a safety net for people who have been pushed out of jobs because of technology displacement or unemployment, that is only going to reinforce that inequality. There’s a great sci-fi series – I don’t know if you guys have watched it – called The Expanse.

And it’s a great series if you want to watch it, but I’m a sci-fi geek. But in The Expanse, and this is sort of projected into the 2060s and 2070s, there’s two types of people in the world, those who live on basic and those who have the real wealth in the world. And this is a… We are already there.

I don’t know if you guys saw the really good Fareed Zakaria interview with… on New York Times, Ezra Klein did with him. Fascinating to see some of the hard numbers in terms of, from 2019 to today, how wealth capture has shifted. So something… and the figures might not be exactly right, but it’s something like that the top 0.1% of the US population went from $9 trillion of assets to $29 trillion between 2019 and today.

And the bottom 50% of the American population went from $800 billion to $4 trillion. So that inequality is already happening because of the high level of technology gearing of the market, the Magnificent Seven or whatever you want to say, all of these things. But AI is just going to supercharge that sort of wealth capture and really sort of be a problem.

Ultimately, this means, and this debate often gets confused in the Western world about an argument between socialism and capitalism. But I think with that level of inequality and uncertainty about the future, because let’s understand this sort of core basic element. I know it’s off the track of banking, but Adam Smith, Wealth of Nations, 1778, the core purpose of the market was to create value exchange for human labor.

And the way we have wealth distribution today is we give people jobs and that’s how they get distributed wealth. But if AI is taking those jobs from humans, even if we do supplement that with the UBI or something like that, capitalism itself is fundamentally breaking down in terms of its purpose. Doesn’t mean socialism is the solution.

It just means that we just need new economic thinking. And that’s not just about how people get paid and value for that, but how do we value humans in society and what’s going to be some of the structural elements for inclusion? Let’s take it. I’m no economist, but my long held view was that the free market economy solves most problems pretty well over the long run.

And all of those have their own asterisk, most problems, but not all. There are things that just don’t have a free market solution and we must intervene to do pretty well. It’s not perfect, doesn’t solve things.

So we need regulation on the margins to improve that. And over the long run and to your earlier point, the short run can be longer than you think and very, very messy. And I do wonder, does that view hold in a world where AI is much, much, much more prominent? I mean, I think one thing we all agree on is we are in just the first couple of pitches of the first inning here in AI.

Yeah. Well, as usually is the case with us, when we get together, we can talk about this for a long time. But how do we wrap up? What’s the plan of attack? If you’re a banker today, where do you start in terms of getting ready for this transition? I mean, for me, there’s two pillars.

One is no matter where this ends up, you need better access to your data and data capabilities. And even with CFPB running out of money and 1033 effectively acts, JP Morgan putting market pressure, doesn’t matter what size bank you are, get your data house in order. The second piece is a cultural one, which is you have to begin to use the tools.

And for the bigger banks out there, this isn’t just your innovation group, right? Because we know that within, in some regards, smaller banks are doing better because they’re using AI universally. Within the largest banks, there’s a small group that has been enabled to use AI and some executives, but a big chunk of the organization is not using AI. And you need to solve that.

And so I’d say data foundation, cultural foundation are the two biggies that you need to take the leap that this is going to be important. The third I’d add to that is I’ll go back to jobs to be done and customer intimacy, knowing your customer, and not just to be compliant and onboarding, but to be able to serve them very well beyond vague terms like service and relationship, to really, really understand who your customer is. The small institutions still do have a bit of an advantage there, and they can leverage this technology to understand even better and to dig deeper and understand the why behind the why.

And doing so has never been more important. Well, to Jason’s point, when I’m doing my talking head stuff on the stage around the world, I say, there are two types of people in the world today. Those who are using AI to enhance their career, to figure out how their business is going to change.

And the second type, those are going to be replaced by the first type. And I think that that’s one of the core problems. If you don’t know the language, if you don’t know the tools that are out there, if you don’t know what’s possible, how can you imagine this new future that’s emerging? Amen to that.

Yeah, very true. Well, on that note, I think that that wraps us up as we head into the beginning of the winter holidays here for us. So we’ll see you all next week with more Breaking Banks.

I’ll have my agent reach out to your agent to schedule that. That’s it for another week of the world’s number one fintech podcast and radio show, Breaking Banks. This episode was produced by our US-based production team, including producer Lisbeth Severance, audio engineer Kevin Hirsham.

If you liked this episode, don’t forget to tweet it out or post it on your favorite social media. Or leave us a five-star review on iTunes, Google Podcasts, Facebook, or wherever it is that you listen to our show. Those actions help other people find our podcast.

And in return, that helps us build an audience that can be supported by sponsorship so we can continue to provide you with our award-winning content every week. Thanks again for joining us. We’ll see you on Breaking Banks next week.