Oakbrook Uses Generative AI to Improve Credit Decisions for Non-Prime Borrowers

Oakbrook has launched Manticore, a new Generative AI-powered credit scorecard designed to improve lending decisions for non-prime borrowers. By identifying patterns that traditional credit scoring approaches can miss, the model helps build a more complete picture of each applicant, supporting fairer, more informed lending decisions.

The launch builds on Oakbrook's continued investment in Generative AI, with AI-powered scorecards now live across both of its core lending portfolios. Together, these models represent a significant step forward in using AI to enhance credit risk assessment and support smarter lending decisions.

Built for non-prime borrowers seeking personal loans of up to £5,000, Manticore is Oakbrook's most sophisticated credit scorecard to date. Trained on more than three times the volume of data used to develop the company's first generative AI-powered credit scorecard, it is the largest and most data-rich model Oakbrook has ever built. Unlike traditional scorecards, which rely on features manually engineered by data scientists, Manticore uses Generative AI to identify patterns in raw credit data that may otherwise remain hidden. Eighteen percent of the model's deployed features are AI-engineered, generated from Oakbrook's proprietary data assets rather than standard bureau attributes available across the market. The result is a more granular view of risk, helping Oakbrook better assess applicants whose financial circumstances may not be fully reflected through conventional credit scoring approaches.

Dr Jamie Ownsworth, Head of Data Science at Oakbrook Finance, said:

"Non-prime customers are often the most difficult group for lenders to assess fairly. Their financial lives can be more complex, their credit files carry more history, and the cost of getting the decision wrong is significant for both the customer and the lender. Satori showed us what was possible. Manticore builds on that foundation and applies those learnings to the segment where improved risk assessment can have the greatest impact. We trained Manticore on the largest dataset we have ever used, developed new AI-engineered features from our proprietary data, and successfully deployed the model into production. Most importantly, it allows us to identify patterns and signals that were previously difficult to capture using traditional modelling techniques. That's what responsible innovation in credit risk looks like."

The launch of Manticore marks Oakbrook's second Generative AI scorecard deployment in less than four months. With both Satori and Manticore now live, Generative AI has been embedded across the lender's near-prime and non-prime credit risk frameworks.

The deployment forms part of Oakbrook's wider strategy to use advanced analytics and AI to improve decision-making, strengthen risk assessment, and support fair, responsible lending for its customers.