Andrew Briscoe is a product builder. He has spent his career designing, building and scaling new products inside corporate banking - embedded finance, multi-currency wallets, real-time and cross-border payments, treasury solutions - across many markets and regulatory regimes. He now advises fintechs and banks on product strategy, design and delivery.
Jackson Hatfield sat down with Andrew to talk about what actually stands between a good idea and a live product inside a regulated institution, and what AI does and doesn't change about it.
Thanks for speaking with me today, Andrew. The standard story about regulated industries is that regulation slows innovation down, but we wanted to start from the other end and ask: what does building in a highly regulated market give you that an open one doesn't?
Regulation is very often the enabler. It creates the wave you can ride. Open Banking in the UK is the one most people will have touched perhaps without thinking about it - being able to see all your other accounts, credit cards and mortgage in one bank app, initiating account-to-account payments, accessing alternative lending. One of my early projects was in India, around UPI, which was government-driven real-time payment rails. Brazil similarly has Pix. In each case the public sector laid the rails and the private sector built customer-centric solutions on top of them. Both schemes have gone from strength to strength and have become trailblazers for other markets.
You can almost categorise it. There is a significant body of innovation in financial services that was triggered or led by regulation, and another that came purely from the private sector, with regulation forming around it afterwards. Done properly - and in the UK it has often been done very well - regulation opens use cases that would not otherwise have existed.
And the relationship with compliance itself? That is where most people expect the friction to be, but my experience with compliance has genuinely been good, and I think the reason is engaging early and treating risk and compliance as part of the team, designing for it from day one rather than meeting the gatekeeper at the final step. I have worked with some outstanding compliance and risk people and they are no different from anyone else. Most of them enjoy the innovation angle. They like being asked what if we changed this, what if the model worked differently, what would it take to be comfortable.
That is not the same as it being quick. I’ve built and launched global digital wallet products, and the path through compliance and risk approvals was lengthy. Many steps, many teams. And when you are doing something globally you cannot have a person in the UK opining on whether it is fine to go live in Singapore, HK or the US, for example, because they are simply not close enough or formally accountable to those regulators. There is very little in the way of UN-style regulation. Countries rightly run their own regimes, and the compliance structure reflects that.
So if compliance is not where propositions die, where do they die? Inside institutions of that size, ideas are rarely scarce.
It is capacity, resources and budget. The large institutions have very large development budgets, but there is an enormous call on them, not least the regulatory work, which is non-negotiable. For example, if you are moving to ISO 20022, or you have CMA Open Banking obligations and deadlines, those things have to be done and they go to the top of the queue.
The challenge is everything below that, which by definition gets deemed discretionary. By discretionary I mean it is a great idea, the business genuinely wants to explore it, but the funding for it has to be carved out of what is left once you have done the mandatory work and the sunsetting of technology that is no longer supported. So it becomes an exercise in ruthless prioritisation. There is no shortage of ideas. The question is how you spot the right ones and make sure the right ones get prioritised, and I think there is real room for improvement there.
That is a different problem from the one most people describe. If choosing is the hard part, does AI help with the choosing, or only with the building?
Where AI is prevalent in financial services today, in things like credit checks, fraud controls, customer service, summarising calls and call notes, it is typically automating the current business process rather than re-engineering it. Some of those use cases are making a material difference.
What I have seen much less of, and what I think is the more interesting question, is how you use AI in the product design process itself. Standing things up quickly, building prototypes, creating test environments. What that gives you is the ability to take a simple return-on-investment business case, make it real and reduce uncertainty. Here is the business case, and here is the prototype, here is what it looks like in the customer's hands, here is how we would test it. You never get certainty, but you get earlier clarity about what the product is and how it will be received, and that de-risks the prioritisation and investment decision.
I would be careful about how far it goes, though. Thinking back to the UPI work in India, that was one of my first real exposures to experimenting in trying to build and launch something in a genuinely agile way, with the usual difficulties of running it internationally and virtually with no single scrum team in one room. Would AI have helped with the early prototyping and some customer-level A/B testing? Almost certainly. Would it have helped with governance approvals? Perhaps. Would it have removed the dependency on the core banking platform underneath? I am not sure it would. You can be nimble around the periphery. The core is still the core, and banking and money-movement regulation exists for a reason, and that is something the industry must not lose sight of.
Embedded finance means your product lives inside someone else's platform, and the customer may never know you are there. What does that do to the product?
Embedded finance is about offering financial services at the point of need. For example, insurance or buy-now-pay-later financing. There are different scenarios. On the consumer side it can be fully embedded, where the website or the software you happen to be using is providing the whole service. In the B2B case I worked on, the customer was aware that they were being introduced to the bank's services and indeed opening a bank account, but it was all done seamlessly and digitally.
The thing that does not change is that the bank is the licensed entity. If a customer has an account, that account is owned by a bank, so in a B2B example you are building a journey that takes them from the ERP or the software provider, through to you, the bank, either as a pass-through or as fully embedded in the partner's platform. What that means is you are one step removed.
Customer acquisition may well be the business case for doing it, but you are not the primary owner of the relationship, your brand is not front and centre, and you do not control distribution. Even the service journeys have to be designed in concert with the partner.
Then there may be a gap between who the end customer thinks is providing the service and who actually is. You can address that legally. The perception stays where it was. They assume the bank is the backstop, because they are the trusted institution, and they are the people who have the money.
You are now advising on payments, including digital assets and AI, in an industry where trust is a licensing condition rather than a brand value. What does the demand for explainability do to any of this?
Explainability is absolutely critical, and in a sense it always has been. Why can this person open an account, why was this payment stopped, why is this customer being given credit or not given credit and at what level. All of that has to be explainable, and the difficulty comes at the point where a programmed algorithm becomes a black box that nobody can account for.
The right place to start is your own domain, where you have full control and you know the data is yours or your customers', and that you are the rightful custodian of it. The flip side is that a model running on a closed and limited data set can be a little underwhelming. It does not have the wisdom of the world behind it.
And agentic commerce, where the agent is transacting on your behalf?
It is a natural progression, given the open banking protocols already exist. It has to be controlled, with real protocols around what information is released and how it is used, but done well it might give you more control rather than less. We already have the likes of OpenAI and Stripe’s Agentic Commerce Protocol (ACP) and Google’s Universal Commerce Protocol (UCP) developed alongside partners like Shopify and Walmart. Now it feels as though your data and history are everywhere because everybody is sharing it. Being able to control and orchestrate when and to whom it goes is potentially an improvement on that.
Whether the journey should be frictionless is still an open question, and I think the answer differs by the size of the decision and the individual. If I have driven into an airport like London’s Heathrow and I am meant to remember to pay the drop-off fee at the end of the day (through yet another portal), and instead my car recognises that I have gone through the barrier and makes the £10 payment and tells me it is done, that feels like a useful and safe use case. But if I am mulling over ski holiday options for the winter with Claude or Gemini, I would not expect it to go away, sort out the resort and the hotel, and come back telling me to be at the airport two hours before the flight. Much as when you are booking something with friends or family, I would want it to say “here is what I found, and ask me what I think”. To me this is good friction, and I would want it. Perhaps I will change.
The harder part is accountability and liability. There is already an issue with fraud accountability landing on the banks even when the fraud materialises somewhere else entirely, because the money movement is where it surfaces and the platform is not accountable at all. Agentic payments will need the same conversation, and it needs to be a thoughtful one, because a disgruntled customer who has bought something they did not intend to is a problem whether or not anything technically went wrong. New entrants have to take some of that accountability. Consumers may have to accept some risk, with guardrails around it. A maximum number of transactions a day, no more than a certain value, in the same way we have limits on contactless and on an ATM card.
If you were designing an innovation function inside a regulated institution from scratch - its mandate, where it sits, how it is funded - what would you do?
Money talks. An innovation team without money has limited relevance, because if all you are bringing is knowledge and ideas, and you are trying to sell those into the department that owns the purse strings, you have a problem from the start.
But funding on its own is not the answer either. I have seen innovation teams that hold the budget and are accountable for the spend without being accountable for sales or the revenue, and that disconnect is its own failure mode. It is fine to ring-fence some money and place some bets, but unless somebody is accountable for the commercial success of what comes out of it, you will not get the traction to scale anything. I recently heard a similar disconnect in an embedded finance context described as an unmet expectation around distribution. Who is going to sell it, how are they incentivised and who owns whether it works must all be defined upfront.
So the team needs its own budget and some skin in the game, and it needs joint performance scorecards with the part of the business that will actually scale the idea. What it cannot be is an ivory tower or isolated lab. People need the time and space away from BAU to explore and experiment, but if that happens in isolation from the wider organisation, it will not fly.
And who should lead it?
Somebody with both curiosity and pragmatism. Curiosity in the sense of someone genuinely interested in pushing boundaries, with connections into the wider industry, thinking about the future (“the day after tomorrow”). Pragmatism in the sense of being able to navigate the organisation, because otherwise you end up with great ideas that never see the light of day. The job is to push the boundaries and keep the ideas grounded enough in the structures you actually have so that you can build advocacy and take them through to something meaningful.
