Welcome, everybody, to the second episode of our mini series about digital assets. And today we have a really interesting topic about the potential convergence of AI. Is this now just a coincidence because of two technologies kind of in parallel, we're gaining maturity, or is there true value in combining the two? So a true convergence.

And to deal with this very challenging question today I have three distinguished guests. We have Jess Houlgrave from WalletConnect, truly the expert on how the user interface for our customers will change. Then we have Michelle Neal. She is the CEO of Fnality, a relatively young company who plays a pivotal role in the new infrastructure for settling payments. And then from Archax, we have Graham Rodford.

As I said in the introduction, this is a really cool topic. We are talking today about what may be one of the most interesting and challenging developments in the world of finance, which is kind of the convergence between AI and tokenisation, whether this is a hype or is it a coincidence or is there true convergence?

And maybe, Michelle, I can come to you first. So do you believe that there is a convergence that tokenisation will influence the world of AI in finance or the other way around, whether AI is an enabler of acceleration of tokenisation and digital assets?

Thank you very much for having me. And I guess I tend to think about these two technologies, AI and tokenisation, as solving different problems in the financial stack where we stand today. So AI as primarily enhancing intelligence, improving analysis, forecasting, optimisation, and decision making. Tokenisation by contrast, modernising execution, making assets, money, and settlement processes programmable.

So where they start to converge is in the ability to connect better decision making with more efficient execution. So for example, an AI system may identify a liquidity need, optimise collateral usage or rebalance positions in real time. But tokenised infrastructure is what allows those instructions to settle instantly and with certainty.

So one without the other, you know, potentially creates friction. So I think we're still early in that journey. Because in regulated financial markets, it's not just about speed or automation. It's about delivering outcomes in a way that is trusted, understood and legally certain.

And that's important because AI doesn't remove the underlying constraints in the financial system. If anything, it can amplify them because you're making decisions faster than the system that can safely complete them.

So if settlement infrastructure can't keep pace, particularly in terms of finality, liquidity and risk management, then AI-driven finance eventually hits a scaling problem. So near real-time, high-quality settlement becomes a prerequisite for AI to scale effectively in financial markets, not just a nice to have.

And so that has to operate within frameworks that provide legal certainty and regulatory clarity. Otherwise, you sort of risk increasing speed and complexity without actually improving outcomes.

But I think the longer-term opportunity is much bigger than simply faster payments or more efficient settlement. Over time, the real transformation comes from liquidity, collateral, margin and asset movements, for example, can be dynamically optimised across markets in real time.

In that world, AI may increasingly become the decision engine for financial markets, but tokenised infrastructure is what allows those decisions to be executed safely, atomically and at scale. And I think that's where these technologies become genuinely transformative together, not just digitising existing processes, but enabling entirely new operating models for financial markets.

Yeah. Which means that there is a lot of change going to happen in the financial services industry, isn't it? So you mentioned actually two core elements there. One was trust, and we'll come back to that later. But you mentioned speed as well.

Graham, do you believe that customer journeys will be faster, that decisions can be made faster in the world of tokenisation, like Michelle said, executing transactions, making them programmable, so probably more purposeful? What's your view on speed of decision, speed of execution?

Yeah. Thanks for having me. So to build on what Michelle said, I guess just briefly, what do we think about when we talk about tokenisation? Really to us it means the register of ownership of something is on a blockchain.

So historically assets have been stored in databases or wherever else they may have been, but now they're on a blockchain. And at the same time, you've seen this not just the asset side, equities, debt or the like, but you've also seen cash move on-chain as well.

So whether it be tokenised deposits or stablecoins, you have this exciting situation where now you have assets and cash on the same technology. And to Michelle's point, that kind of opens up a lot of efficiencies for AI or otherwise, whereby transactions can be automated.

So delivery versus payment or certain compliance rules around transacting, which when combined with AI, calculates what it needs to do, and then it needs to be able to execute it. Using old assets and cash it may have struggled, but now everything's on digital rails, it's able to execute them more seamlessly.

So I think it's the decision making behind the transactions which can now be improved using AI, but the assets and cash themselves are now more free-flowing 24/7.

So I think it's important to remember those regulatory and trust points. They don't go away. But we're in a situation now where you can have agents executing 24/7 across a whole host of different financial transactions.

And do you believe transactions will become safer, more trustworthy, less fraught by the combination of tokenisation and AI?

In the long term, probably. But in the short term, I think we're going to see a lot of experimentation. There's a lot of people developing, much the same as algorithmic trading, actually.

If you think about the big quant trading hedge funds, for example, they developed algorithmic trading. Then everyone suddenly realised you probably need to train them in sandboxes and have separate test environments before you release them on markets.

I think the same will be true of agentic trading. There will be a lot of agents trading, and we're seeing a lot of exchanges adapting their infrastructure to allow it. But I think we just need to be wary of letting it out of the box too early and letting it run free.

So regulation is going to struggle to keep up. But there's no doubt there's going to be an increase in transactions. On the flip side, the fraud and identification side, I think that probably gets greatly enhanced as well, as does the ability to trace transactions on a real-time basis.

So this was more on the execution of transactions. However, from a customer perspective, obviously there will be quite some changes as well.

We have heard and even spoken in other parts of this mini series about the relevance of wallets. And I think, Jess, this is in particular your area.

So in that migration of interacting with accounts, either through your mobile app or through an IVR or a desktop website, what is the role of AI in that transition from accounts to wallets?

When we think about the role of the wallet today, it's really about enabling a different user experience than maybe our traditional financial systems have offered us. And I think that change is going to continue, partly driven by AI.

The way that we interact with these systems is going to change. Enabling my agent to take actions on my behalf safely and securely is going to be a drastic shift for users, whether that's around how they think about their investment products that they're accessing or how they think about everyday payments that they are making.

I think the transition to digital assets also opens up new possibilities for what the wallet becomes. So it no longer just becomes the place and the software that we use to manage our assets, but also perhaps things like managing our identity or managing our data.

And when you combine these two things together, you can have some really interesting new experiences that unlock the ability for me to give certain permissions on behalf of my agent or directly to a provider to access my KYC information, for example, is going to change the way that a user can onboard with a financial institution.

So the wallet really becomes a hub for more than just the financial assets that we control, but also other things like our data. And then when you layer on AI into that, it's going to unlock a lot of new user experiences.

I'm cautiously optimistic, let's say, in that I think that there's a lot of excitement around this. I don't think we quite yet have the security frameworks that enable us to offer this to end consumers in a very safe and secure way right now.

And AI, let's not forget, also brings the cost of the bad guys almost down to zero. So for those people trying to be bad actors, whether that's around fraud or anything else, life is getting a whole lot easier for them too.

And on the other side, we need to be building the systems and tooling that allow us to catch that.

Yeah. And that's probably again where part of the trust should be obviously on the permission side in the wallet, like you said, Jess. And then on the other side, there is programmability of transactions on the blockchain.

Which brings me a bit to the whole theme that I promised we would come back on of trust. Where do you believe, Michelle, in the future the trust will predominantly reside in the whole ecosystem of wallets, banks, the blockchain operators, the regulators, identity providers? I could go on, right?

Where do you believe the trust will go and how do consumers in particular understand that?

So I'm a little bit cautious about the notion, and this is one that sort of comes up a lot, that distributed ledgers eliminate the need for trust. What they really do is change how trust is structured.

People sometimes just describe blockchain as trustless, but financial systems are never truly trustless. The question is simply where does the trust sit? Is it in institutions and governance frameworks or increasingly in the design of the rules of the network itself?

And I think that distinction becomes even more important in a world where AI systems are participating more actively in financial decision making. Because if AI is making decisions and tokenised infrastructure is executing them, then trust, accountability and governance have to be embedded much more directly into the system itself.

So one interesting aspect of tokenisation is that controls, permissions and governance can potentially be embedded directly into assets and transactions, making oversight more proactive than purely reactive.

But transparency alone is not the same as accountability. So you still need clear governance, intervention mechanisms, legal certainty and ultimately confidence in the underlying settlement process.

And that brings us back to the importance of the settlement asset itself. So for wholesale financial activity, central bank money continues to provide the strongest foundation because it delivers finality and removes credit risk from the system.

In addition to that, central banks have a governing remit for financial stability. Of course, the role of monetary policy transmission and the fractional reserve banking system. And there's lots of schools of thought on whether fractional banking will be needed in the future.

And all of these things will obviously continue to evolve. So I do see the role of central banks and regulators as remaining central but evolving.

Distributed infrastructure can extend how trust is accessed and used across increasingly digital and interoperable markets, but it does not eliminate the need for this sort of trusted anchor underneath.

In fact, as systems become more automated, more interconnected and potentially more autonomous, the importance of that trusted foundation arguably increases.

Because if AI accelerates transaction velocity, financial decision making, markets need infrastructure underneath that can support that activity safely, consistently, predictably.

And I think that's ultimately the bigger challenge for the industry. So success will not really be determined by a single technology or a single model. It will depend on whether the system as a whole becomes more coherent, whether value, liquidity and risk, for example, can move efficiently, safely and consistently across markets, jurisdictions and different forms of digital money.

So we're moving from digitised finance towards increasingly programmable finance. And that has implications not just for efficiency, but for how trust, governance, accountability and resilience are built directly into the financial system itself.

And I think that's what makes this moment so important, because the architectural decisions that we make now will shape how financial markets operate for decades to come.

Yeah. And again, you mentioned speed here, both on the kind of decision making and on the execution. And that probably makes it even more scary, right?

Because when instant payments were introduced, there was this requirement of the two-way transaction to be executed in five seconds. And already in those days, we had a conversation about real-time fraud detection, real-time sanction checks, etc.

And these five seconds in this era almost sound like a century ago. I mean, this is literally milliseconds.

Do you believe maybe, Jess, back to you since you're more consumer-facing, isn't this really scary for the ordinary person in the street that everything goes at the speed of light and you can't trust a human anymore who will sit down, have a look at your transaction, maybe block it or recall it if need be? What about that speed? Isn't that the opposite of trust?

So I think what is important is to think about the speed that is theoretically enabled by the digital ledger system, and then what we actually want to be presenting to the end user.

I think having incredibly fast settlement layers, Michelle alluded to the importance of that moving forwards. But just because that is there doesn't necessarily mean that what we want to give to the end consumer is going to be that level of speed and finality.

So I think it's really important, as people are designing the products that sit on top of these blockchains, to think about all of the other pieces that are put in place.

Some level of friction, especially in a payment flow, is often really desirable and really important. It's why we have things like two-factor authentication against some payments so that we can do these additional checks.

And so I think it's very important for us to design the product experience and the user experience with these things in mind. Notwithstanding that underneath it all, we could do it faster doesn't necessarily mean that we should.

I think one of the key advantages, though, of this technology is that those things are possible. And with AI, when you layer this on, we have the ability to have much greater understanding actually of those transactions than maybe having that person in the loop.

Just because somebody's sitting there looking at your transactions and maybe deciding one is a bad one doesn't necessarily mean that the human is best placed to make that decision on your behalf.

And the data sets that can be consumed by some of these models now actually may mean that the machine is best placed to do that instead. So that's where the power of these two technologies really comes together.

I think the other thing that is really important when we think about this idea of speed and finality for end users is also just the fragmentation that we have out there.

So digital assets do not all exist in one cohesive ledger right now. They exist all over the place. There's huge amounts of fragmentation.

And so that user experience and the speed of the products that we're offering to people also need to really carefully balance some of this fragmentation to make sure that it's understandable and coherent to an end user in a language that they understand, whether they're making a payment or making an investment decision.

We need to translate a lot of this language around crypto, blockchains, digital ledgers, digital assets, tokenised funds into a nomenclature that they feel comfortable with. And that's how we'll bring users safely along this journey.

Yeah, it's all difficult language, isn't it? But I think you're making an interesting point here. There is the kind of theoretical speed of decision making and execution, but then there is the practicalities of a wide distribution of different ledgers and technologies. And there is still the ability to add in controls, in particular when transactions become programmable.

We're probably all geeks on this, financial geeks, right? We love AI, we love tokenisation. But at the same time, if I would ask you, Graham, freely speculate a bit about the future. Is this now one of the many hypes that we have seen?

Like AI was almost dead right after the innovation of machine learning. It was probably ChatGPT that saved the day for AI. And I'm a bit sceptical here.

Tokenisation or cryptocurrencies, distributed ledgers, that was arguably the last couple of years of the first decade of this century. So almost like 20 years ago. And now all of a sudden it's booming with stablecoins, with tokenisation, etc.

So if I ask you to freely speculate a bit, how are the two reinforcing each other? Or is this just a hype and then it will slowly die down again?

Yeah, I think what's interesting is I've known the panellists and their companies for pretty much the whole time we've been in the sector, which has been about eight years.

So for us, it certainly doesn't feel like it's something that's happened recently. But I guess for the first five years of our existence, we said this was coming. And really in the last couple of years it's really started to take off.

So pretty much every financial institution out there has started working on blockchain. They all started with a strategic innovation team that's gradually grown. Everyone's looking to replace all of their products now.

So I think there's no doubt the financial institutions see the direction when it comes to tokenisation. And then when we look around the side of that, you're seeing regulation being put in place globally now.

So I think they're all signs that tokenisation certainly isn't going away. And AI has been in there. I mean, I'm kind of new to AI really, but it's been around for decades and it's just been gradually growing.

But I think all of us have seen this real shift in the last year as we've all started to deploy it onto our desktops to help our day-to-day tasks. And whilst not perfect, you can see the impacts drafting an NDA, putting a legal agreement together, PowerPoints, models, whatever it might be, or even programming whole systems.

So there's no doubt that it's going to be impactful and not go away. I think we're all in a bit of the experimentation phase right now.

You're going to start seeing a lot of people creating apps in their bedrooms and realising there's a reason they're not product design and programmer all together. But everyone's kind of moving up this learning curve.

And I think in parallel, you've got this merge going on between banks, wallets, neobanks, payment firms. They're all coming together.

And I think part of the drive for that is due to people's patience really. Everyone wants everything right now in their pocket, one click. They no longer want to fill in a subscription document for a financial instrument and send it off somewhere. They want to open an app. They want to tap it.

So I think right now we're all in the kind of deep tech side of things, but we're going to start to see people implementing this in the background and start to obfuscate the detail and really give a good user journey.

So I think you're actually going to see this accelerate rapidly from here, with financial transactions being increasingly prevalent as part of everyone's day-to-day lives and then also automation of those coming in behind it.

And are you as optimistic, Michelle, about the future or the convergence, I should say, about AI and tokenisation?

I would say that the transition isn't really uniform across the financial system, the value chain or the transaction lifecycle, right?

Some areas, particularly front-office activity, analytics, trading, they tend to be a little bit more iterative, evolve quickly. I mean even the point that was made earlier about sandboxes for trading algorithms, the ability and the release cycle to change is just much faster.

But when you get to core financial infrastructure like payment settlements, market plumbing, it tends to evolve more incrementally because the requirements around resilience, governance and certainty really are significantly higher.

It's a more complex part of the infrastructure. So I think there you see progress may appear a little bit slower, but I think it's more durable as it goes through its evolution.

And I think that's an important distinction because there's a tendency to view this primarily as a technology shift. So in reality the harder questions are often around legal frameworks, regulatory alignment, governance, operational readiness.

And in many cases, the technology itself works. It's kind of the easiest part.

I know that when I think about my own business, a lot of people think of Fnality as a fintech, and we are. But actually, the technology is only like one of three legs of a stool really.

And the real challenge is whether things can operate reliably at scale within the constraints and expectations of the financial system.

So I think a proof of concept can demonstrate functionality, but production infrastructure has to demonstrate resilience, operational continuity, legal certainty and trust every single day, particularly when you're in wholesale markets like we are, when you think about the sort of size and scale of transactions and transfer of value.

But I think we are now seeing meaningful progress moving from pilots towards infrastructure designed to function in real-world conditions and solve real operational problems.

I guess maybe my last question for today. I'm sure if we listen to this podcast, there are so many words and terms we use, like stablecoins, distributed ledgers, wallets. And all of this is new, not just to the average person in the street, but to employees of companies as well, treasurers, bankers, regulators.

Don't we have an information gap to bridge here?

I think in some cases yes, but in others no. And what I mean by that is that there are very few consumers out there who can tell you how their money moves between banks or how ISO 8583 messaging works for their credit card.

There are also many treasurers out there who can't actually tell you how ISO 20022 messaging works underneath the hood. And they shouldn't need to, because what they need is a product that works for them.

There are some people who really deeply need to understand that technology. And it's exactly the same when it comes to distributed ledgers.

Some people who are building it should definitely understand it. The regulator needs to understand it. Their back-office teams who are building this stuff need to get their head around it.

But for the treasurer who is executing payments, or for the end user who just wants to buy a coffee with their wallet, they don't need to understand how the technology works.

What they need are products that are designed specifically for their use cases, that have the right controls in place, that make their lives easy.

This is why taking technology and making products from it are two completely different skills and both really necessary as we bring this into the real world.

So there is an information gap, but I think it's up to us as an industry and also the people trying to build on this technology to really understand what is it that I should know, and what is it that we should be actually educating the end user about.

And I think these are two different levels that we need to be cautious that we don't overcomplicate things for an end user.

There are many end users today whose money is already moving around on stablecoins. They don't know anything about blockchains. They just know that they get their money faster and that it is cheaper for them to use this particular settlement rail.

And that's exactly how this should be.

Yeah, I guess. But still, I mean, I need to explain to my mum that where she used to talk about her app, which was already a challenge, now all of a sudden she will get a wallet, right?

Well, I don't know. I think her wallet may well be part of her app going forward and she won't know the difference.

Yeah, yeah. So this has been hugely interesting with three very knowledgeable experts.

I have the task of summarising this and trying to bring this to a close. But I think we touched upon trust, as well as this shouldn't be a technology play itself.

We talked about the theoretical speed of decision making and execution, but then in reality things might be completely different.

And the initial question was, is this now purely a coincidence of two technologies more or less moving forward in parallel? Or do they actually emphasise and reinforce each other?

And I think the joint conclusion was, yeah, they are both there and they actually re-emphasise. I think Michelle initially framed it quite well. There is the speed of decision making, which may be more driven by AI, and then there is the speed of execution, more driven by tokenisation.

With that, thank you for the participation in this podcast. Michelle, Jess and Graham, thank you so much and hope to speak soon.

Thank you, thank you, thank you.

And that was it for today. Thank you very much, dear listeners, for tuning in to this episode. If you like our podcast, go and subscribe on Spotify and Apple.

And I will remind you, there will be a third episode in this mini series, which will deal with the broader application of digital assets beyond digital money.

So stay tuned for experts talking about how we're going to tokenize your home or your will, or any other asset that makes sense to be tokenised.

See you next time.