The end of the batch era

On the morning of November 22, 2025, the Society for Worldwide Interbank Financial Telecommunication retired the messaging format that had handled cross border payments for half a century. MT103 for customer payments, MT202 for interbank transfers, the language that moved trillions of dollars between institutions for 50 years, all went out of service. The handover drew almost no public attention.

ISO 20022, the structured data standard that had been coming for nearly 20 years, became the only currency of the network. The shift is not just a format change. Richer structured payment data changes how institutions handle fraud detection, reconciliation, liquidity visibility, and compliance automation. Most bank customers did not notice. Inside payments departments, the cutover closed an era.

It was one of four shifts to land on payments organizations in a 15-month window. The Federal Reserve’s instant payments service, FedNow, has now crossed 1,700 participating financial institutions and raised its transaction limit from $1 million to $10 million in late 2025. The average FedNow transaction value sits near $99,000, firmly commercial rather than consumer. The United Kingdom’s mandatory authorized push payment fraud reimbursement rules, in force since October 2024, moved a substantial share of scam losses onto bank balance sheets. The European Union’s Digital Operational Resilience Act, DORA, came into force for roughly 22,000 financial entities in January 2025 and is now reshaping how banks contract with technology suppliers and how they account for operational risk. Most institutions are trying to absorb all four shifts using operating models designed for a slower, batch-based world.

Dr Leda Glyptis has spent 20 years watching banks try to absorb shifts of this kind, first as a senior banker and now as the author of two influential books on financial services transformation, Bankers Like Us and Beyond Resilience. She argues the industry still talks about modernization as a technology problem first, when the breakdowns usually start somewhere else: in governance concessions, in institutional habits, in operating models built to protect political capital rather than deliver change.

Glyptis joins ACI Worldwide as keynote speaker at Payments Unleashed EMEA – London 2026 on June 29 at the Hilton London Bankside. Ahead of London, in a written exchange while she was traveling in Tennessee, she set out the early signals that a transformation effort will not deliver, where AI is changing decisions inside banks, what irreversibility now demands of fraud operations, and the assumption about payments she expects the next five years to overturn.

The earliest signal that a modernization effort will fail, in her telling, is also the most common, and it appears before the work has formally begun.

“On the journey to deciding to engage with a modernization effort, before the work even begins, a lot of political capital is used up,” she says. “A lot of energy is used, a lot of calories are spent, a lot of favors are called and political capital burnt.” That has two unintended and severely damaging side effects, she argues, both of which take months to surface and become almost impossible to reverse once they do.

The first is a perception problem. By the time the work formally begins, the people around it already feel exhausted by it. “When the work begins, most stakeholders already feel like it’s been ongoing for as long as the decision making has been going on,” she explains. “You will feel it when you are about six months into your efforts, and your stakeholders start getting restless because we’ve been at it for a couple of years already. Although we really haven’t. But saying that doesn’t change the organizational perception.”

The second is structural. To avoid burning further political capital, modernization leaders make concessions early, often without recognizing them as concessions. “Once the green light is given to begin, we give a few things away to placate the people who said yes to us in the first place,” she says. “We often agree to a communications schedule that doesn’t match the work cadence, or we allow a lot of passengers to join steercos, we commit to oversight and governance that isn’t fit for purpose, or a combination of the above.”

At the time, those decisions rarely look consequential. Six months later, they usually are. “These concessions don’t feel material when they are made but they become defining,” she says. “They set the pace and eventually priorities of the work and derail efforts by invisible but persistent nudges off the intended course.” Most large modernization programs do not fail on a dramatic technical breakdown. They fail because the operating conditions for success were quietly traded away in the first 90 days.

If the political dimension is what derails change, the deeper problem inside payments is something more durable. Asked where the human element shows up first, in incentives, in language, or in operating model, Glyptis chooses none of those.

“I am going to cheat and say habits, because it actually encapsulates them all,” she says.

The habits she has in mind are not personal preferences. They are the muscle memory of how senior decision makers spend their days. “What they prioritize without even thinking about it: which email gets read first, which meetings take precedence, how work is delegated and efforts rewarded.” Over time, those patterns harden into the operating model itself. “Those habits are formed by what is known to work inside organizations and in turn reinforce all those structures for each generation of workers.”

Incentive structures and established language, she argues, do not just shape individual behavior. They become the unwritten template for what success looks like inside a bank, and therefore the unwritten template for what failure looks like too. “The things that were recognized, rewarded, appreciated or expected, and the things that may land you in trouble and therefore should be avoided,” she explains. “They, in turn, become part of the operating model as well as a template for the behavior of those trying to be successful inside our organizations.”

The result is the line that anyone who has tried to change a major bank from inside will recognize on first reading. “Those habits are muscle memory and scar tissue all rolled into one and exceptionally hard to change but, for that exact reason, impossible to ignore.”

Inside the largest banks, the visible AI work to date has clustered in compliance and operational risk: alert triage in fraud and anti-money-laundering teams, sanctions screening assistance, false positive reduction. Most banks are still layering AI onto existing workflows rather than redesigning how decisions get made.

Glyptis sees that pattern as predictable. “Most implementations are either automation of existing processes or chaotic creativity, where individuals play around and build without much governance, oversight or method,” she says. “That is part of each organization’s learning journey and frankly to be expected.”

The shift she finds more interesting is inside compliance specifically. “As real time risk awareness becomes a hygiene factor, AI is being used to power operational simulations, find patterns against fincrime typologies and enable decision flows for the officers inside regulated institutions. That’s exciting and material.”

What interests her more is not deployment but how organizations think about what AI is. “We are seeing material impact in organizations that appreciate that AI changes the way we work and needs to be approached as a tool that will not just fit into a day,” she says. “Our day will shift around it. That is particularly true of engineers but not exclusively.”

The banks treating AI as a labor substitution exercise will, in her reading, struggle. “How we lean into leveraging these tools, what we ask of them, how we train them and how we train ourselves to operate with them is a journey,” she explains. “Organizations that understand that and don’t just look to agents to complete end to end tasks will win both in the short and long run.”

The fraud question, in 2026, is no longer primarily a customer experience question. It is a balance sheet question.

UK Finance reported more than £1.1 billion stolen through fraud in the United Kingdom in 2024, with authorized push payment fraud losses of £450.7 million. The mandatory reimbursement rules introduced by the Payment Systems Regulator in October 2024 moved a substantial share of those losses onto banks. In the first three months under the new rules, 86% of in-scope APP scam losses were returned to victims through bank reimbursement. The economics of fraud detection, in the United Kingdom and increasingly in markets that follow its regulatory lead, have moved out of the customer experience function and into direct financial risk.

Glyptis sees the operational consequence as larger than the industry has yet acknowledged. “Although we’ve known this is coming for a while now, the vast majority of banks is not ready,” she argues. “Real time detection across the value chain is mostly not possible and the material cost of what irreversibility means for banks is yet to be fully felt. The reality is that it’s here and the impact is going to be felt across the board.”

Two questions, in her view, follow from that. The first is internal. “What are the areas across your tech estate that you have neglected or intentionally not tackled because they are too complicated, not visible enough, too political? Where is your operation slow or opaque? Where are you carrying residual complexity? It’s time to tackle it. Because that is now the weakest link of your shop.”

The second is about the supplier base. The point landed in regulatory form three weeks before the SWIFT cutover. On November 18, 2025, the European Supervisory Authorities designated 19 ICT service providers, including the major cloud platforms, as critical under DORA, subjecting them for the first time to direct EU oversight. “Are your suppliers part of the solution or part of the problem?” Glyptis asks. “We often talk about partnership in our industry, but rarely mean it. It’s time to actually pack a punch behind the term.” In a real time, irreversible environment, the boundaries between in-house and supplier capability collapse. “Across your tech estate, third party suppliers are a key part of how you operate today. Going forward those suppliers should be partners and those partners should be part of how you bring a real time, transparent, resilient and robust estate to life.”

The standard for partner selection, she argues, has changed. “Partner selection has never been more significant. If they don’t bring acceleration to your simplification efforts and a commitment to future-proofing your operation in constant, true collaboration, then you may need to reconsider your partnerships.”

Asked which widely accepted belief about payments transformation she expects to look obsolete in five years, Glyptis works through two answers before arriving at the one she actually wants to make.

The first instinct is to say UX. “We keep saying UX is the battleground but, if it ever was, it hasn’t been for a while.” The second is the number of handovers in any single payment. “A single payment goes through so many processes and so many entities, each of which develop largely in isolation. If a consumer payment fails, the end user has no idea whether their card was declined because they have inadequate funds, because the issuer made a mistake, because the gateway timed out. Inside the payments industry, each of those handovers is a market of its own and the performance SLAs a battleground of its own.”

But the answer, in her reading, sits a level deeper. “The answer is a few clicks above that and encapsulates all of the above,” she says. “Payments is a huge space. It covers so many verticals, so many competences and so much breadth. Someone working in merchant acquiring is in payments. And so is someone working in clearing and settlement. But the two will probably never meet, talk about how they are innovating and what is changing in their corner of the organization.”

The separation of consumer and wholesale payments, she argues, has been an industry given for years and a structural assumption baked into how banks are organized. Consumer payments and wholesale payments operate as different P&Ls, regulated differently, processed differently, run by different leaders. “A payments person on one side of the bank will have no idea how things work across the hall,” she says. “That will not be the case much longer. Finally.”

In her view, what will matter most is not the smoothest customer interface but the ability to see across the entire money cycle, end to end, in real time. “Understanding how things fit together, where dependencies and vulnerabilities lie, and how handovers across the entire money cycle, from a single merchant initiated transaction to liquidity management, will stop being the superpower it is now and become a baseline. And it is about time.”

Dr Leda Glyptis opens the mainstage at Payments Unleashed EMEA – London 2026 on June 29, at the Hilton London Bankside. Two days of keynotes, panels, and structured working sessions for senior leaders across European payments. Topics include real time payments scale, fraud and scam liability, sovereignty without fragmentation, and the operating model shifts that will define the next decade.