Revolut Builds AI Models From Transaction Data
Revolut is developing proprietary AI models trained on its own transaction data as it broadens its technology ambitions.
Revolut has one asset most AI startups cannot easily replicate: a huge stream of proprietary financial behaviour.
What happened
The company said it is developing proprietary AI models trained on its transaction data.
Revolut processes tens of millions of transactions each day and is increasingly positioning itself as a broader technology company rather than only a digital bank.
Why it matters
Financial data can support better fraud detection, underwriting, personalisation and customer automation.
Training models around proprietary transaction history could give Revolut a structural advantage over companies relying only on generic foundation models.
The key question is whether those models materially improve products rather than simply reducing reliance on third-party AI providers.
The bigger picture
Large fintech platforms may become important AI companies because they control specialised, high-quality behavioural data.
The strongest advantage may not come from inventing a new foundation model, but from combining existing AI techniques with datasets competitors cannot access.
Revolut's strategy reflects that shift.