Revolut has spent the past year quietly assembling the pieces for what could be the most aggressive AI bet in consumer banking. The London-based digital bank, which reported more than 45 million customers globally as of late 2024, has built a proprietary financial AI model, locked in a major cloud infrastructure deal with Google, and signaled through its UK chief executive that a conversational financial assistant is coming before mid-2026. The headline figure of 13 million customers reflects an approximate count of Revolut’s UK user base, where the assistant is expected to roll out first; the company has not published an exact UK figure for 2025, and third-party estimates from 2024 placed the number between 9 and 10 million, suggesting the real total may be somewhat lower or may have grown since those reports.
No full app navigation overhaul has been publicly confirmed. But the trajectory is unmistakable: Revolut is positioning AI not as a support tool buried in a help menu, but as the primary way tens of millions of people interact with their money.
What Revolut’s App Looks Like Today
For readers unfamiliar with the product, Revolut’s current app is organized around a conventional tab-based interface. A bottom navigation bar offers quick access to a home screen showing account balances and recent transactions, a payments hub, a trading and crypto section, and a hub for extras like insurance, subscriptions, and rewards. Budgeting tools, analytics, and card controls sit behind additional menus. Customer support is handled through an in-app chat system that mixes automated responses with human agents. It is a polished but fundamentally traditional mobile banking layout, and it is this structure that Revolut’s AI ambitions appear designed to eventually reshape or replace.
A Proprietary Financial Model Called PRAGMA
In April 2025, Revolut researchers published a technical preprint on arXiv describing a foundation model called PRAGMA. The paper details a system built for core banking functions: credit scoring, fraud detection, and predicting customer lifetime value. These are not experimental side projects. They represent an effort to replace the patchwork of third-party vendor tools that most banks rely on with a single, adaptable in-house architecture.
Foundation models matter here because they can be fine-tuned for multiple tasks from one base system. If PRAGMA performs as the paper describes, Revolut could use it to power everything from real-time risk decisions to personalized spending insights. Traditional banks, which typically run these functions on separate legacy systems that do not communicate well with each other, would have a hard time replicating that flexibility quickly.
Google Cloud and the Road to 100 Million Users
Building a model is one challenge. Running it for tens of millions of customers in real time is a different one entirely. In early 2025, Revolut and Google Cloud announced a deepened strategic partnership explicitly aimed at supporting Revolut’s growth toward 100 million customers. The deal covers high-performance computing, data analytics, and the deployment pipelines needed to serve AI models at scale.
That 100 million target is Revolut’s own stated ambition, not an outside forecast. And the scope of the infrastructure commitment tells you something about priorities. Companies do not sign cloud deals of this magnitude to power a basic FAQ chatbot. This is the kind of investment you make when AI is the product strategy, not an add-on.
Revolut’s UK CEO Says a Financial Assistant Is Imminent
Francesca Carlesi, Revolut’s UK CEO, discussed plans for an AI-powered financial assistant in comments reported by Bloomberg, pointing to a launch window before mid-2026. She described a tool that would let customers ask natural-language questions about their finances and receive tailored, personalized responses.
This is not the kind of chatbot that redirects you to a help article. What Revolut is describing is closer to a financial advisor that already knows your spending patterns, upcoming subscription renewals, and savings goals. The distinction is important. Rule-based chatbots have been around for years, and most people find them frustrating. An assistant powered by a proprietary foundation model and backed by serious compute infrastructure is a fundamentally different product.
Whether this assistant will supplement the existing app interface or eventually replace traditional menu navigation is still an open question. Revolut has not published product documentation or app update logs confirming a full redesign. But the combination of a proprietary model, a scaled infrastructure deal, and executive statements about an imminent launch points strongly toward a major shift in how the app works.
The Competitive Landscape Is Moving Fast
Revolut is not operating in a vacuum. JPMorgan Chase has committed billions to AI and data infrastructure. Klarna’s CEO claimed in 2024 that the company’s AI systems were handling customer service work equivalent to roughly 700 full-time agents, though the company later acknowledged it still needed to rehire human staff for complex cases. Monzo and other digital banks have introduced AI-assisted budgeting features.
But there is a meaningful difference between using AI to optimize back-end processes or handle support tickets and making a conversational AI the primary way customers navigate their bank. That second move is far riskier and far more ambitious. Most banks still treat AI as an optimization layer sitting behind existing interfaces. Revolut’s research, infrastructure spending, and public statements suggest it wants AI to become the interface itself.
Sarah Kocianski, a fintech analyst and former head of research at 11:FS, has previously noted that neobanks often announce ambitious product visions well ahead of delivery, and that the gap between a compelling demo and a reliable, regulation-compliant product at scale is where most bold plans stall. That observation applies directly here: Revolut’s public signals are strong, but the finished product has not shipped yet.
Regulation, Trust, and the Hard Parts
Ambition alone does not guarantee execution, and several critical unknowns remain.
Regulatory scrutiny is intensifying. The EU AI Act, which began phased enforcement in 2024, classifies AI systems used in creditworthiness assessments and financial services as high-risk, requiring transparency about how automated decisions are made. UK regulators have signaled similar concerns. As Revolut’s AI handles more sensitive tasks, from credit decisions to fraud alerts, it will face growing pressure to explain how its models reach conclusions. Explainability requirements could limit how aggressively the company deploys complex models in customer-facing roles.
Human support still matters. A conversational AI that handles most queries could cut costs dramatically, but it risks alienating customers who need a real person for complex or stressful situations like disputed charges or suspected fraud. Revolut has not disclosed whether it plans an AI-first model with human backup or something more blended. Without published service metrics or user satisfaction data, the real-world experience is still unknown.
User trust is the biggest variable. Even technically impressive AI tools can fail if customers do not trust them with their money. Banking is personal, and people tend to be conservative about how they manage finances. Revolut will need to prove that its assistant is not just capable but genuinely reliable before millions of users willingly abandon the familiar menus and dashboards they already know how to use. Early customer reactions on forums like Reddit and Trustpilot will be worth watching closely once any beta or staged rollout begins.
Why Revolut’s AI Gamble Will Pressure Every Retail Bank
The real significance here extends well past one fintech company. What Revolut is testing is whether conversational AI is ready to become the default interface for financial services. It is arguably the most aggressive test case in the industry right now: a company with tens of millions of users, proprietary models, scaled infrastructure, and executives publicly committed to making AI central to the product.
If the assistant launches successfully and customers adopt it, the pressure on every retail bank, neobank, and financial app will be immediate. If it stumbles, whether through poor accuracy, regulatory friction, or simple user rejection, it will set a cautionary precedent that slows AI ambitions across the sector.
As of spring 2026, the experiment is already underway. Revolut has published the research, signed the infrastructure deals, and put its leadership on record. What happens next will play out in app stores, regulatory filings, and the daily banking experience of millions of customers across Europe and beyond.
