What does it take to move from a single AI success story to a scalable, enterprise-wide AI strategy? At Money20/20 Europe’s AI Summit, a panel of senior leaders from Stripe, Swift, Starling Bank, and AWS gathered to share exactly what that journey looks like behind the scenes.

The message was clear: real AI adoption isn’t about having the most cutting-edge model. It’s about aligning infrastructure, risk governance, and organizational buy-in around a purpose-driven, scalable AI operating model.

Moderated by Jennifer Chang (AWS Worldwide Financial Services Lead), the panel included:

Together, they unpacked how their organizations are going from experimentation to infrastructure from scattered pilots to AI factories.

Rachel Levi of Swift emphasized the importance of strategic filters. At Swift, AI projects must meet two criteria:

Her example: Swift’s federated fraud detection model, co-developed with 10 global banks. Rather than trying to “own the solution,” Swift validated the need for cross-institution collaboration to meaningfully reduce financial crime across borders.

Stripe CTO Rahul Patil described the shift toward “agentic commerce” as one of the most transformative changes in tech.

Stripe is investing in foundational infrastructure, such as AI-driven fraud detection models trained on over 50,000 transactions per minute. These models not only catch fraud they optimize checkout experiences based on buyer context and geography.

And yet, not every investment can be measured in pure ROI terms. “Sometimes, you invest just to be first to shape the space,” said Patil.

From a CFO’s perspective, Declan Ferguson of Starling Bank made it clear: AI is now an essential part of how investors and boards view long-term value. “Having something in your equity story about AI? That’s going to be a prerequisite.”

Starling’s approach is rooted in pragmatism. With 600+ people in tech and over 100 dedicated to data and AI, the bank is doubling down on what has worked: use-case-focused AI aligned to their lean digital infrastructure. Investments in AI aren't made in isolation they’re baked into Starling’s DNA as a tech-first bank.

If there was one recurring principle across all speakers, it was this: AI must be responsible by design.

Swift built out its own responsible AI framework from day one, defining standards for:

“It’s non-negotiable,” said Levi. “We don’t wait for regulation—we get ahead of it.”

Ferguson added that banks must evolve their governance models, especially second- and third-line risk teams, to monitor AI at the same depth as they do model risk today.

Patil took the point further: not using AI is now a risk. Fraudsters use it. So not using it means falling behind in defense.

When asked about infrastructure modernization, Patil emphasized that speed is becoming the defining variable: “Every other day, there’s a new model. The question is how fast can you test, integrate, and iterate on it?”

Stripe is investing in AI-first systems that serve customers and internal teams alike. Marketing teams can build tools without code. Sales teams can launch workflows powered by AI agents. These aren’t side projects they're rewiring how Stripe operates.

What unified all four perspectives was a deep understanding that AI is not a tool, it’s a system shift.

You don’t scale it by launching pilots. You scale it by:

If the first wave of AI was about excitement, this panel signaled the next phase: enterprise-grade execution.

Not affiliated with Money20/20

Bobsguide is a ClickZ Media publication