Payments leaders don’t lack for buzzwords these days — instant payments, tokenized money, agentic AI. What’s harder to find is a candid conversation about what actually keeps CIOs and heads of payments up at night. That’s exactly what we got when we brought together banks, fintechs, and payment leaders from across the region for a series of lunch roundtables. The result was less a briefing and more an honest exchange about where the industry is heading, and where it’s still catching up.
The core system dilemma isn’t going away
One theme surfaced again and again: the question of whether payments should live inside the core banking platform or sit apart from it. Larger tier 1 and 2 banks are more likely to have it separated, while even some tier 3 and 4 institutions, which would normally favour keeping it embedded, are now looking to implement their payments capability as a separate, stand-alone instance alongside their core. Neither approach is universally right — it’s a trade-off between agility and costs that every institution has to weigh against its own risk appetite and legacy footprint.
Regional differences add another layer. The US juggles two instant payment rails that are not fully interoperable, the UK is still working through legacy constraints on its path to ISO 20022, Europe is gearing up for a digital euro pilot in 2027, and parts of Southeast Asia — the Philippines in particular — have leapt from negligible to majority digital payment adoption in just a few years. There’s no single playbook; culture, regulation, and legacy infrastructure all shape the pace of change differently market by market.
Tokenization is moving from theory to practice
For a while, tokenized money and cryptocurrency carried a whiff of speculation. That’s shifting. Legal guardrails for instruments like stablecoins are now taking shape in multiple jurisdictions, and traditional banks are starting to feel comfortable engaging — not because the technology changed, but because the regulatory backing did. Our Payments Domain Lead, Mick Fennell, put it plainly at the roundtable: a lot of the RFPs and RFQs coming in now ask about stablecoin capability almost as a checkbox, even when the institution asking doesn’t yet have a concrete use case in mind. It’s less considered strategy at this point and more fear of being left behind — but that FOMO is itself a signal the market is turning a corner.
Agentic payments: the challenge nobody has fully solved
If there was one topic that generated the most energy — and the most uncertainty — it was agentic payments. Mick was candid that the industry doesn’t yet have the answer here. As AI agents begin transacting on behalf of banks, businesses, and eventually consumers, payment systems will need to operate at a speed and scale most institutions simply weren’t built for. It’s not a future problem — vendors and banks alike are already being asked how their platforms will keep pace.
So what should banks actually be doing to prepare? A few threads worth taking seriously:
- Build for real-time, not batch. Agentic transactions won’t wait for end-of-day settlement cycles. Screening, sanctions, and fraud checks need to happen in milliseconds, not hours.
- Rethink fraud and risk models. Agentic payments introduce new fraud vectors — including the possibility of synthetic or derived currencies moving between agents — that existing fraud frameworks weren’t designed to catch. Fraud detection and scam detection are genuinely different disciplines, and both need to evolve in parallel.
- Invest in scalable, resilient infrastructure now. Whether that means cloud-native architecture, sidecar deployments alongside legacy cores, or modular payment hubs, the institutions best positioned for agentic payments will be the ones that didn’t wait for certainty before building flexibility into their systems.
- Engage on standards early. Industry working groups are already forming around agentic payment standards. Banks that participate now will have more influence over how the rules get written than those who wait to react.
- Treat this as a compliance and regulatory question, not just a technology one. Who is liable when an autonomous agent initiates a fraudulent or erroneous payment? That’s still an open question, and it’s one regulators and industry bodies will need to work through together.
The common thread
Across every topic —Platformmodernization, tokenization, fraud, agentic AI — the same undercurrent kept surfacing: nobody is claiming to have this fully figured out, and that’s precisely why forums like this matter. The institutions that will navigate this best aren’t necessarily the ones with the most resources, but the ones willing to build flexibility into their architecture and stay engaged with where the conversation — and the regulation — is heading next.
Thanks to everyone who joined the conversation and shared their perspective so openly.
