Revolut Highlights Proprietary AI Foundation Model for Banking Data

According to a recent LinkedIn post from Revolut, nine of the company’s researchers have co‑authored a scientific paper with NVIDIA on an artificial intelligence model tailored to banking data. The post highlights that, unlike many financial AI models focused on text, this work targets long, complex sequences of transactions, app actions, and trades.

The post describes a new foundation model called PRAGMA, reportedly trained on 24 billion banking events to enable use cases such as credit scoring, fraud detection, and product recommendations within a single architecture. It also notes that the paper is openly available to the wider industry while the authors remain employed at Revolut, implying that internal expertise in specialized AI could support future product differentiation and operational efficiency.

For investors, the post suggests Revolut is investing in proprietary AI infrastructure that may improve risk management and customer personalization, potentially reducing credit losses and fraud while increasing cross‑sell and engagement. Collaboration with NVIDIA could further position Revolut within an ecosystem of advanced AI development in finance, which may strengthen its competitive stance against both traditional banks and digital peers over the medium term.