Revolut’s Unified AI: A Paradigm Shift in Banking
Rather than stitching together separate models for payments, lending and support, Revolut developed a single proprietary foundation model called PRAGMA to learn financial behavior across the entire customer journey. This unified AI architecture lets the bank convert fragmented signals into consistent, contextual intelligence that can be applied across risk, product and service functions. The result is a single source of predictive truth instead of multiple, competing models.
The Power of PRAGMA: Integrated Intelligence at Work
PRAGMA ingests transactions, app interactions, investments, support logs and product usage into one connected system. That consolidated data enables cross-functional use cases: real time fraud detection, credit risk scoring, personalized product recommendations and intelligent customer routing. Reported outcomes include a 64.7% improvement in fraud detection, 16% better credit risk prediction and 41% stronger product recommendation performance. Built on GPU-accelerated infrastructure such as NVIDIA hardware, PRAGMA is designed to scale for a user base north of 70 million, while letting teams reuse the same model for multiple business problems.
Driving Customer Experience and Agility
PRAGMA powers a visible shift in customer interactions. Revolut reports its AI assistant handles roughly 75% of routine support requests, freeing human agents for complex cases. The firm’s customer-facing agentic AI, AIR, acts as a lifestyle companion for budgeting, travel and daily finance tasks, using the account-wide context PRAGMA provides. That tight integration accelerates feature development, reduces time to market and creates a competitive gap against traditional banks with siloed analytics.
Key Takeaways for Financial AI Strategy
Moving from isolated models to a unified foundation compounds intelligence and produces business-wide gains. Institutions aiming to replicate Revolut’s approach should focus on three pillars:
- Centralized data platform that supports privacy and governance.
- A shared foundation model that powers multiple use cases and reduces redundancy.
- Cross-functional teams and metrics to measure ROI by use case, not by project.
Unified AI is not a single project. It is an architectural and organizational shift that turns data into a durable competitive advantage across fraud, credit, product and customer experience.
