Sberbank's FinTRACE AI transforms financial data into smarter decisions

A new AI approach called FinTRACE has been developed by researchers at Sberbank’s Practical Artificial Intelligence Center. The system, created in partnership with the Sber AI team, aims to improve financial services by making them more accurate and transparent. Its potential uses extend beyond banking, including medical forecasting and customer behaviour analysis. FinTRACE works by transforming raw financial transaction data into a structured knowledge base. This allows AI to detect stable patterns and habits before making decisions, leading to fairer risk assessments and personalised offers. The system combines information retrieval, explainable rules, and large language model reasoning with temporal data.

Banks and digital platforms can use FinTRACE to build more flexible and cost-effective AI models. The tool acts as a universal layer over event-based data, supporting tasks like predicting customer churn or credit scoring. It also helps financial institutions identify client challenges earlier, enabling timely support. A scientific paper on FinTRACE, led by Maksim Makarenko, Executive Director of Data Research at the centre, has been accepted for presentation at the 49th ACM SIGIR conference. Beyond finance, the approach could enhance medical forecasting, allowing for earlier disease detection and better treatment planning.

FinTRACE offers a way to improve decision-making in finance, healthcare, and other sectors. By turning transaction histories into actionable insights, it supports more reliable and transparent AI applications. The system’s adaptability means it can be applied across industries that rely on event-based data.