How Google and Revolut Turned AI into Large-Scale Infrastructure

Alejandro Martínez · 2026-05-05

### AI as the New Backbone for Fintech For fintech companies chasing hundreds of millions of users, scaling infrastructure while weaving in advanced AI isn't just a tech challenge—it's a survival move. This fireside chat, recorded at FinTech R:Evolution, pulls back the curtain on how two heavy hitters, Google and Revolut, are making it happen. David Abitbol, AI & Infrastructure Lead at Revolut, and Ivana Martins, Head of Strategic FSI at Google, shared their tech choices, hard-won lessons, and practical advice for any fintech diving into AI. We'll walk through what they said and why it matters. ### Revolut's AI Journey: From Experiment to Engine David Abitbol kicked things off with a line that sums up Revolut's speed: "I've spent five years at Revolut, which feels like twenty years in startup time." That density explains how the company turned AI from a side project into a core part of its operational infrastructure. Revolut's AI adoption took two main paths: - **Customer-facing apps**: Slashing support costs without killing quality. The old problem was simple—as users grew, so did support costs. AI breaks that link by handling more requests without adding more people. - **Internal productivity**: Fighting fraud, staying compliant, and smoothing operations—all jobs that used to be manual. These gains aren't as flashy as customer tools, but they're just as crucial for long-term competitiveness. ### Rita: The AI Support Agent That Changed the Game Rita, Revolut's AI customer support agent, is their flagship use case. Launched with clear metrics, it started with a 35% resolution rate without human escalation (HWOE). Today, that number's around 80%. At Revolut's scale, every percentage point gained saves hundreds of thousands of dollars annually. But Rita's evolution tells a bigger story. Initially, a central team built and deployed all AI models. That became a bottleneck as demand exploded. Revolut's fix? They created shared frameworks, evaluation tools, and deployment playbooks that any team could use. They moved from a centralized model to a distributed one, where each team can build its own solutions. That shift unlocked speed and innovation across the whole company. ### Google's Role: The Foundation Underneath Ivana Martins brought Google's perspective, explaining how their cloud and AI models—especially Gemini—serve as the underlying infrastructure for financial services. The big challenge is proving these foundation models meet the industry's demands for reliability, traceability, and regulatory compliance. In financial services, trust is everything. A bank won't plug in an AI model without solid guarantees about how it works. Google's job is to provide those guarantees, making it easier for companies like Revolut to innovate without breaking rules. ### What This Means for You If you're building a fintech, here's the takeaway: AI isn't just a feature—it's infrastructure. Start with a clear problem, like Revolut did with support costs. Measure everything. And don't let your AI team become a bottleneck. Build tools that let your whole company experiment and deploy. That's how you scale. This fireside chat showed that the future of fintech isn't about one big AI breakthrough. It's about making AI a natural part of how you operate, every single day.