BORICA

BORICA, Bulgaria's national payments infrastructure operator, relied on batch-based fraud detection systems that could not keep pace with the speed and volume of real-time transactions. By the time suspicious patterns were identified, fraudulent transactions had already been processed.

BORICA built a centralized AI fraud prevention platform that risk-scores transactions in real-time. The platform uses machine learning models to analyze transaction patterns across the national payments network, assigning risk scores to each transaction before authorization.

The platform now risk-scores 70% of all national payments in Bulgaria in under 10 milliseconds, enabling real-time fraud interception at the network level. The shift from batch to real-time processing dramatically reduced the window for fraudulent transactions to succeed.

• National payment infrastructure operators can serve as a centralized AI fraud prevention layer, protecting all participating banks simultaneously.
• Sub-10ms risk scoring is achievable at national scale with purpose-built ML infrastructure.
• Moving from batch to real-time fraud detection fundamentally changes the economics of payment fraud prevention.