By Shona Sabah (Senior Manager - Strategic Growth Lead)
MoneyLIVE returned to London this month and brought together leaders from banking, payments and financial technology for two days of fascinating keynotes, panels and fireside chats. From the sessions I attended, these are the themes that stood out to me:
AI came up in virtually every session at the summit, but the conversation has moved on from whether or not it matters (I think we know the answer to that?!) and onto why so many initiatives stall after the proof-of-concept stage, and what it actually takes to move from an interesting demo to something that works reliably at scale.
Several speakers made the point that the barrier to building a convincing demo is now very low. The harder work is everything that follows, specifically the governance frameworks, evaluation mechanisms, reliability standards, and organisational change that regulated businesses need before anything goes near a customer. It was highlighted that the pace of genuine, enterprise-scale deployment has lagged well behind the pace of AI experimentation, and that this gap is where a lot of organisations are currently stuck.
Harriet Rees (Group Chief Information Officer, Starling) pointed out that AI's returns in banking and payments so far have been focused on back-office efficiency and internal productivity. Those are great, but she argued it would be a missed opportunity if they didn't translate into better customer-facing experiences in banking, in payments, in the tools people and businesses actually use day to day.
Adam Bealey (Chief Executive UK & Ireland and Global Head of Partnerships, SWIFT) added some additional context from the payments side. Corporate clients are asking for more real-time liquidity, deeper insight, end-to-end trackability on cross-border transactions, and settlement as close to real time as possible, and these are specific gaps that better use of data and AI could close. The question is how to build toward them without the kind of incremental, siloed experimentation that has produced so many stalled pilots elsewhere.
Saira Khan (Head of Innovation and Partnerships, First Direct Bank) reiterated the importance of starting with the problem, not the technology. The organisations making real progress are those who identify a specific customer or operational issue first, then ask whether AI is the right tool. The pressure to be seen doing something with AI is understandable, but it tends to produce solutions in search of problems and those rarely scale.
Running alongside the AI conversation was an important but sensitive question: which parts of banking and payments genuinely require human involvement, and how should organisations design around that rather than away from it?
Tom Stoddart (Managing Director: Head of Markets Sales, Business & Commercial Banking at Lloyds Banking Group) argued that AI creates a new question about the division of labour - not just between tasks, but between types of thinking. There are things AI handles well (information retrieval, pattern recognition, processing speed) and there are things it doesn't (judgement under uncertainty, advocacy, the ability to back an idea and stand behind it). Getting that allocation right, he said, will be one of the defining strategic questions over the next few years.
To highlight this, he used the example of property. There is now a large amount of data available about any given house - comparable sales, environmental risk, local amenities. Yet the decision to buy still almost always involves a viewing, a conversation, often a trusted person's input. The data informs the decision but doesn't make it. The same logic applies in business banking in that a relationship manager's value isn't in processing information but in understanding a client's situation well enough to offer relevant counsel, and in being willing to advocate for them internally when it matters.
This connection between data and human judgement is worth pausing on, because it points to where the real opportunity lies. The organisations best placed to deliver high-quality human interactions are those who understand their customers deeply enough to know when and how to intervene, and with what. That's partly a technology question, but it's at least as much a research and insight question: what do we actually know about this customer, and are we using it?
That links nicely to another theme that stood out for me, that in retail banking, in business banking, and in payments, there is a gap between what organisations assume customers want and what those customers actually need. Several sessions returned to this from different angles, and the pattern was consistent between them - the organisations with the clearest sense of direction had done the most rigorous work to understand the real problems their customers face, not just their stated product preferences.
Hannah Dove (Customer Propositions Director, Lloyds Banking Group) described research that identified the biggest unmet need in retail banking as 'setting themselves up for the future’. She also raised a data point that's easy to read as a technology story, but is something else in my opinion. Half of adults now use ChatGPT for financial advice. Her read was that this reflects a support gap more than a preference for AI. People turn to it because a knowledgeable, accessible person isn't available, not because they'd prefer to talk to a chatbot. For banks, that's more usefully understood as an unmet need than a threat.
In the CX orchestration session, Tanvi Gokhale (Managing Director: Customer Propositions & Journeys, NatWest) and Piyush Chechani (Director of Engineering (Digital Channels), Nationwide) extended this into channel design. Gokhale's point was that friction is often poorly defined because even a single moment that goes wrong creates a lasting association with the brand, regardless of how smooth everything else has been. The goal isn't to make every interaction frictionless - it's to ensure the right response reaches the customer at the right moment, through whichever channel they're in – and to do that you need to understand them.
Another observation that I found interesting was that the industry doesn't have a data problem, or really a technology problem. The data exists. The technology exists. What's harder is using it to engage customers based on their actual needs at specific moments in their lives, rather than on demographic segments that don't reflect how people actually behave. Customers compare experiences, not channels, and relevance is determined by what's happening in someone's life, not by which age bracket they fall into.
The challenger bank panel offered a different version of this argument. Neil Chandler (Chief Executive Officer, Tandem) put it nicely: nobody wants the BNPL, they want the jeans; nobody wants the mortgage, they want the house. The product is a means to an end, and the organisations that understand the end clearly are better placed to build the right means.
Bianca Zwart (Chief Strategy Officer, bunq) described how this discipline compounds over time. bunq's founding insight was that traditional banking was structured around the bank's needs rather than the user's. Staying close to customers as the business scaled meant the understanding of who they were building for kept sharpening and that clarity has driven every subsequent decision, including the move into the US market, where they're targeting the underserved expat segment that finds it genuinely difficult to open an account without local credit history.
Understanding what customers need is one thing. The harder question, and the one that occupied the second day of the conference in particular, is how to act on that understanding at scale, in a way that feels relevant and timely to each customer rather than generic.
The cognitive banking panel covered this well. Jamie Renehan (Head of Behavioural Insights, Bank of Ireland) described the richness of the data banks hold - deep, longitudinal transaction records that, when properly used, reveal how customers' circumstances and behaviours change over time. Life-stage signals (a first salary, a change in spending patterns, the beginning of a regular saving habit) can indicate where someone is in their financial life and what kind of support would be most useful. AI (there it is again!) makes it feasible to do this at scale in a way that wasn't practical before. The aspiration is a bank that acts more like a financial coach than a product distributor.
Saira Khan added an important qualifier to this when she said proactive support only works if the timing and tone are right. Customers don't want to feel monitored. The value comes from intervening at a moment that's genuinely useful (a prompt about an upcoming large payment, a flag that a direct debit has changed, a nudge toward a better savings rate) rather than a stream of communications that feel intrusive. Getting this right requires understanding not just what customers do, but what they're likely to need next. That's where the combination of good data and genuine customer insight becomes commercially meaningful.
Moneybox offered a retail-facing version of this same ambition. Their Aurora platform takes customers through a structured onboarding process and produces a personalised financial plan - an attempt to deliver something close to the quality of individual financial guidance at a scale that makes it accessible to people who wouldn't otherwise have access to it. The underlying question their work raises is relevant across the industry: if AI can bring the quality of one-to-one financial guidance to the mass market, what does that change about how those services are delivered, priced and eventually regulated?
Raghu Narula (Chief Customer & Banking Officer, Starling) approached personalisation from the business banking side, and the framing was slightly different. For SMEs, the most useful thing a bank can do is reduce the operational burden of running a business - not just provide financing. His description of agentic AI as a potential 'digital CFO in the pocket of an SME' points to what this might look like - spend tracking by project, automated tax preparation, smarter invoicing and payment tools, and interactive conversations that give a small business owner the kind of insight that would otherwise require a finance director.
Conrad Ford (Chief Product & Strategy Officer, Allica Bank) described something similar in the savings context. Embedding savings pots directly into the business current account (including ring-fencing funds for tax obligations) and positioning the bank as a form of treasury director for businesses too large to operate without that function but too small to afford a dedicated person. The language of 'treasury director' is interesting as it's not describing a product, it's describing a role the bank plays in a customer's business. That shift, from product provider to genuine business partner, is what personalisation at its best looks like in practice.
Looking across the sessions, the clearest thread wasn't about any particular technology. It was about whether organisations had done the work to understand specifically who they serve and what problem they genuinely exist to solve, and whether that clarity was shaping their decisions, rather than following from them.
The AI question, the human touch question, the personalisation question are not separate issues. They're different expressions of the same challenge. In an environment where the technology available to every organisation is improving rapidly, the things that differentiate you have to be built on something more durable - a genuine understanding of your customers, and a clear sense of what you're trying to do for them.
That's easier to say than to act on. But the organisations that came across as most purposeful in these sessions were the ones who had the most honest and specific answer to it.
If you'd like to hear more about MoneyLIVE Summit 2026 or discuss how you can better understand your customers' needs, connect with Shona and drop her a message.