Why AI-ready Infrastructure was the Real Payments Story at Sibos 2026

Sibos 2026 workshops and conversations revealed why AI-ready payments infrastructure, ISO 20022 data, multi-rail connectivity, and end-to-end fraud controls need to work as one.

AI dominated Sibos 2026, but the show’s more consequential question was whether payments environments are ready to support it. Across conversations about digital assets, tokenized deposits, real-time payments, fraud, ISO 20022 and agentic AI, one conclusion emerged: AI’s value depends on the infrastructure, data and controls surrounding it.

Across discussions about digital assets, tokenized deposits, real-time payments, relentless fraud, ISO 20022, and agentic AI, the same pressure point kept surfacing: financial institutions are asking more of bank cores and older infrastructure that wasn’t designed for today’s speed, complexity, or variety of payment activity.

Bottomline Global Head of Banking Sales & Growth, Vitus Rotzer, and Global Head of Banking Solutions & Growth, Jessica Cheney, both attended, and agreed that Sibos this year was less a showcase for any single technology than a look at how different tech needs to work better together.

“AI may have been the headline, but digital assets, tokenized money, and the infrastructure required to move value across traditional and emerging networks were arguably the biggest conversations on the show floor,” Cheney said after the show.

For banks, PSPs, and fintechs, the implication is practical. Modernization is more and more about creating operating environments that can accommodate new payment methods, use richer information intelligently, and strengthen control at the same time.

It’s a Job for Multiple Rails

Financial institutions already operate across established payment methods such as wire, ACH, and cross-border networks. This year, however, Sibos talk-tracks added urgency to an adjacent question: how should banks and PSPs prepare for tokenized deposits, blockchain-based settlement, digital assets, and other emerging ways to move value?

Rotzer said the stronger model is “one platform and one operational framework that can access multiple rails,” applying intelligence and automation across them.

It’s an important point because new rails don’t need to become another operational silo. Separate systems create separate views of transactions, liquidity, exceptions, and risk just as the market is demanding greater visibility across them.

The strategic value is therefore flexibility. Institutions need the ability to test new forms of payment and settlement without rebuilding their operating model for each one. They also need to manage traditional and emerging networks side by side as the lines between them continue to blur.

Cheney witnessed this new convergence throughout Sibos. Banks were no longer discussing AI, digital assets, and modernizing infrastructure as isolated innovations, she said. Increasingly, they were asking how those capabilities combine to move value safely and efficiently across a more diverse financial ecosystem.

Let AI Prove Itself

Agentic AI generated plenty of interest at Sibos, as expected. Agentic is showing up in some ambitious plans, but the more credible near-term opportunities may be less dramatic and more functional.

Rotzer pointed to payment operations, exception management, fraud and anomaly detection, liquidity forecasting, treasury decision support, customer inquiries and compliance investigations as areas where AI can help experts make better decisions faster.

That’s different from allowing an autonomous agent to make consequential financial decisions on its own. “Payments are fundamentally about trust, control, regulation, and accountability,” Rotzer said. Automate carefully.

Banks have good reason to move deliberately on agentic AI. Giving a system the ability to identify an exception or recommend an action is one thing. Giving it independent authority to move money, approve a high-value transaction, or make a risk decision is another.

The distinction provides a useful roadmap for adoption. AI can absorb manual work, recognize patterns, and bring information to the right people sooner. Institutions can expand automation as confidence, governance and regulatory frameworks mature.

The question is not simply what AI can automate. It is where automation improves the payment process without weakening accountability.

Make ISO 20022 Produce Returns

One of the clearest changes at Sibos 2026 was the role of ISO 20022 in the conversation and planning. After several years dominated by migration deadlines and implementation headaches, Rotzer welcomed a shift toward a more concrete question: what can banks and PSPs now do with the richer information they’ve worked so hard to create?

The answer extends across operations. Structured payment data can help reduce exceptions, improve remittance information and strengthen fraud and compliance processes, while also removing friction caused by incomplete or inconsistent transaction details. It can also give AI more complete context when analyzing payments or operational decisions.

That’s where the return on ISO 20022 spend begins to become tangible. Rotzer also sees an opportunity to change how institutions think about payments themselves. He said a payment should not simply transport money from one account to another. The information accompanying it can improve communication, support decisions, and create value throughout the journey from initiation to settlement.

Rotzer and Cheney agreed that ISO 20022 becomes more important after migration, not less, because the business case is ultimately about what institutions can extract from data rather than the format it comes in.

Treat Resilience and Governance as AI Capabilities

Cheney identified three prerequisites that repeatedly surfaced in discussions at Sibos 2026 about deploying AI effectively: high-quality structured data, strong governance and controls, and operational resilience.

The second and third can easily be overshadowed by excitement around the technology itself. An AI-driven action inside a regulated financial environment needs to be explainable and auditable. Institutions need to understand what influenced a decision, establish who’s accountable, and maintain appropriate controls over how models are used.

At the same time, intelligence delivered in real time is of limited value if the surrounding infrastructure can’t act on it reliably. Modern APIs, real-time processing, and end-to-end operational visibility become part of the AI equation because they determine whether an insight can actually improve an outcome.

That alters the definition of AI readiness. It’s not simply having models, data scientists or promising use cases. It’s having an operating environment in which intelligence can be introduced without compromising resilience, governance, or trust.

Follow Fraud Risk Across the Entire Payment Journey

Many Sibos sessions resonated with talk of connected fraud defenses as essential to outmaneuver criminals.

With payments becoming faster and moving across more channels and networks, fraud defenses can’t depend on a single screening point. Cheney described a model that starts before payment initiation with identity verification and account or payee validation, continues through real-time monitoring and anomaly detection, and carries into post-payment intelligence and investigation.

Fraud has become an end-to-end problem because criminals don’t respect the organizational boundaries separating payment operations, cybersecurity, compliance and investigations. The response should be well defined and data-driven.

Institutions need connected data and risk signals that allow suspicious behavior to be assessed in context. AI can strengthen detection by identifying anomalies and relationships that would be difficult to recognize manually, but its value increases when those findings are visible across the payment lifecycle.

“The strongest approach treats fraud prevention, payment visibility, and operational intelligence as one connected discipline rather than separate functions,” Cheney said.

That may be the most useful way to read Sibos 2026 as a whole.

The institutions best positioned for what comes next won’t necessarily be those deploying the most AI or betting earliest on a particular new rail. They will be capable of applying intelligence wherever a payment travels, across established and emerging networks, while preserving the visibility, control and trust on which the movement of money still depends.