Banking sector integrates AI agents for automated workflows

The banking sector is undergoing a transition from using artificial intelligence as simple chatbots to integrating specialized AI agents into core workflows. According to analysis by Ajay Rawal and Dr. Kostis Chlouverakis of EY, these agents are designed to handle specific stages of banking processes, such as collecting and verifying documents, identity checks, loan file preparation, and fraud detection.

This model emphasizes human oversight, where AI agents operate within predefined limits and escalate complex cases to human employees. This integration aims to increase speed and reduce bureaucracy in tasks like opening accounts, processing loans, and managing regulatory compliance. To ensure transparency and accountability, experts suggest implementing an ‘AI decision flight recorder’ to act as a ‘black box’ that explains how and why specific decisions were made.

While banks are investing heavily in these technologies, moving from pilot programs to full-scale production requires a redesign of existing processes rather than just adding individual tools. Currently, fully autonomous decision-making remains highly limited, at approximately 2%, due to strict regulatory frameworks.