Global fintech company Revolut has announced the launch of a dedicated artificial intelligence division, Revolut Research, to pioneer proprietary machine learning technology. This new unit will focus on developing in-house models, centered around its foundation model PRAGMA, to enhance its financial services. The move intensifies the growing competition among European digital banks to integrate advanced, native AI capabilities.
A Strategic Shift to In-House AI Development
Revolut is championing a "build, don't bolt on" philosophy, moving away from reliance on third-party AI solutions. Head of AI Pavel Nesterov stated that leading intelligent banking requires moving beyond external blueprints. This strategy contrasts with legacy institutions attempting to retrofit outdated infrastructure with off-the-shelf software.
The new research division will serve as the central framework for all proprietary AI deployments and machine learning initiatives. Working alongside partners like Nvidia, the unit aims to embed native intelligence directly into its core financial engine. This approach is designed to turn local market data into a significant global advantage.
The Power of PRAGMA and Global Data
At the heart of this initiative is PRAGMA, a proprietary foundation model developed to decode complex financial behaviors. It powers critical functions such as real-time risk assessment, platform operations, and tailored product recommendations. Revolut is rebuilding its AI stack around this single, shared behavioral intelligence layer for greater efficiency.
Early deployments of PRAGMA have already demonstrated substantial performance gains over previous systems. The company reported a 2.3 times improvement in credit risk accuracy and a 65 percent increase in detecting fraudulent cases. Furthermore, the model delivered 41 percent more relevant product recommendations, enhancing customer personalization.
This technical advantage is fueled by an extensive global dataset from Revolut's 80 million customers across 40 markets. This data feeds the models in a self-reinforcing loop, making them progressively smarter at predicting user needs and identifying risks. This creates a proprietary intelligence edge that traditional banks may find difficult to replicate.
Engaging the Broader Scientific Community
Revolut Research has committed to contributing to the wider AI field by regularly publishing its scientific findings. The division also plans to open-source some of its technical frameworks to foster innovation and collaboration. This transparent approach aims to establish Revolut as a significant voice within the global research community.
The team will actively engage with peers through participation in major international conferences, including NVIDIA GTC and ICAIF. Additionally, Revolut will host quarterly meetups at its offices for the scientific community to share knowledge. These efforts underscore the company's dedication to advancing financial AI.
The Neobank AI Arms Race
This launch places Revolut at the forefront of an intensifying AI arms race among Europe's leading neobanks. The announcement follows recent AI-driven product launches from key competitors, signaling a sector-wide push towards smarter banking. This trend highlights AI as a critical differentiator in the competitive digital finance market.
For instance, UK rival Starling Bank recently introduced its own smart tools and an "agentic" AI assistant. Meanwhile, Dutch competitor Bunq upgraded its AI-powered customer support assistant late last year. Revolut's own in-app assistant, AIR, is another example of this competitive drive to innovate.
The establishment of Revolut Research marks a significant step towards the deep integration of proprietary AI in financial services. By focusing on in-house development and leveraging its vast dataset, the company aims to deliver a safer, more personalized customer experience. This strategic investment solidifies Revolut's position as a key innovator shaping the future of intelligent banking.