Singapore – Revolut has unveiled Revolut Research, a dedicated division within its wider AI Department tasked with pioneering next-generation machine learning architecture for global financial services.
The unit, which sits alongside leading academic and technology institutions, will act as the central framework behind the fintech giant’s proprietary AI deployments and machine learning initiatives.
According to Revolut, the global banking sector is running up against a technological ceiling, with legacy institutions attempting to retrofit decades-old infrastructure using off-the-shelf software wrappers.
This results in mounting operational drag for banks, while customers are left with clunky interfaces and rudimentary search-bar chatbots.
By contrast, Revolut says it is embedding native intelligence directly into its core financial engine, an approach it believes converts local market data into a compounding global advantage.
At the centre of Revolut Research’s work is PRAGMA, the company’s proprietary foundation model developed in collaboration with NVIDIA.
PRAGMA has been engineered specifically to decode complex financial behaviours, powering real-time risk assessment, platform operations and tailored product recommendations.
Through this initiative, Revolut is restructuring its AI stack around a single foundation underpinned by one shared behavioural intelligence layer.
Anton Repushko, Head of Revolut Research, has already presented findings from the division’s work at several high-profile international conferences this year, including ICML in Seoul.
Early deployments of PRAGMA on historical data have pointed to notable performance gains over legacy baselines.
Among these, the model has demonstrated 2.3 times higher accuracy at identifying credit default risk compared with existing systems.
Fraud detection has also improved markedly, with 65% more fraud cases caught and 17% greater precision in alerts.
Meanwhile, product recommendations across retail and business accounts have become 41% more relevant, according to the company’s internal figures.
Underpinning this technical progress is what Revolut describes as an unmatched global dataset, drawn from more than 80 million customers across over 40 markets.
Revolut processes billions of cross-border transactions and a wide range of financial behaviours in real time, and this data is fed directly into Revolut Research’s models.
The result, the company says, is a self-reinforcing loop that establishes a new language of financial behaviours, with the models growing progressively sharper at detecting fraud, evaluating risk and anticipating user needs as the dataset expands.
Revolut contends that this creates a proprietary intelligence advantage that traditionally structured banks will struggle to replicate.
“To lead the future of intelligent banking, you cannot rely on third-party blueprints”, said Pavel Nesterov, Head of AI at Revolut.
“We have launched Revolut Research to institutionalise our ‘build, don’t bolt on’ philosophy. By training native foundation models on our global operational data, we are giving our engineering teams an unprecedented engine to deploy smarter features faster, eliminate systemic friction, and give our customers a safer, radically better financial experience”, Nesterov continued.
For his part, Repushko framed the division’s mission in similarly ambitious terms.
“Revolut Research has been established to responsibly build financial intelligence at its deepest layer, rather than patching together narrow, specialised models”, said Repushko.
“In PRAGMA, we are developing a single, unified foundation model capable of understanding the true nuance of financial behaviour in real time. Technology is in Revolut’s DNA, and by collaborating with global tech leaders, this division is engineering proprietary capabilities that set us apart from traditional banks”, he further explained.
