Icon Solutions, Citi, NatWest and UBS: why three Tier 1 banks bought into a payments framework

Interview with Tom Kelleher by Sean Murphy There was a time when it seemed like the payments world would get less complicated. Real-time would replace batch, digital would replace cash, a handful of dominant rails would absorb the rest. None of that happened. Cash use is declining but still material; cheques remain in widespread use across the US economy; China runs on QR codes; the eurozone is preparing for a digital euro; stablecoin volumes are growing; ISO 20022 is forcing a data overhaul. For the operations team inside a Tier 1 bank, their lives have not become easier, payments is more fragmented than ever.

Tom Kelleher, co-founder and director of Icon Solutions, has been watching this happen up close for the better part of two decades. The UK payments technology firm, founded in 2009 and recently awarded a 2026 King's Award for Enterprise in Innovation, is supporting the large banks that are affected by this complexity and fragmentation. Its proposition rests on a single observation: payment rails rarely disappear entirely, they tend to accumulate, and the banks underpinning the global payments system need a way to absorb that accumulation without rebuilding their core stack every time a new payment method appears.

Icon's answer is the Icon Payments Framework, or IPF, a framework that banks use to build and run their own payments engines rather than license a finished product from a traditional vendor. Kelleher's diagnosis of the market is that the modernisation roadmaps banks were being offered in 2015 looked materially similar to the ones they had been pitched a decade earlier, with the same architectural assumptions and the same dependency on the vendor for every subsequent change. “In the past it's always been rigid products,” he says. “The box in the corner, nobody touches it, and if you want any changes, you've got to go back to the vendor.” IPF is structured to change that model: their framework stays neutral, banks customise on top of it, everything built on top of the framework belongs to the bank.

That model has direct implications for the metrics operators are measured on. Icon's own figures put total cost of ownership reductions at up to 50 percent and time-to-market gains at up to four times on payments build, with real-time payments adoption achievable in six months. The underlying logic is hard to dispute; a bank that owns what it builds and runs the framework with its own engineers, supported where necessary by Icon consultants.

The next phase of fragmentation is about data. The information carried by an ISO 20022 message is materially richer than the legacy formats it replaces, and banks have worked out that the act of moving money is being commoditised; the value sits in what they can do with the information attached to it. That has reframed payments from a back-office utility into a strategic data layer. Icon's bet is that the orchestration of all those rails, in five years, will sit centrally rather than as a patchwork of point solutions. The bank that can cleverly route a payment from one rail to another, based on cost, speed, settlement risk or customer choice, will have a structural advantage over the one that cannot.

Three of Icon's client banks, Citi, NatWest and UBS, have backed that thesis with capital as well as deployment. UBS led a funding round in June 2025 that existing investors Citi and NatWest also joined, with all three already running IPF in production. The structure is unusual but Kelleher is clear that the founders have retained control and in many ways it’s the best of both worlds as the banks contribute to roadmap conversations.

On agentic AI, Kelleher is realistic. Icon is using AI to accelerate its own build process and has encoded IPF knowledge, such as functional and technical product documentation, within an AI-based server to give AI tools the context to be effective. In addition, as an extensible framework, IPF is suited to invoking AI agents from the payments flow e.g., for repair. Autonomous agents moving real money are not, in his view, near-term. “If money is going to the wrong place, there will be significant consequences,” he says. Fraud detection and engineering acceleration are where the genuine production applications live today. However, he does believe agent experience will become just as important as user experience for IPF. 

For operators, the takeaway is that the payments ecosystem is not consolidating any time soon. The choice is between continuing to bolt point solutions onto a legacy core or rebuilding around an orchestration layer designed for the fragmentation that is actually happening. Icon’s bet is on the latter. After seventeen years, with a King's Award on the wall and three of the world's largest banks on the cap table, their bet is looking more defensible each day.