When a disruption occurs, banking systems require accurate and early recovery. However, when an error is detected, operators often receive an influx of messages and reports, making it difficult to pinpoint the cause of the event, and ultimately delaying the recovery time.

Trialled as a proof-of-concept, the new solution demonstrated a 98% accuracy in monitoring and responding to error messages during a three-month trial. The PoC added patterns likely to cause errors in incident response to generative AI and linked an application supporting a series of operations in event detection with IBM supercomputer watsonx, helping to shorten the steps to recovery and improve recovery speed. Utilizing watsonx also enabled Mizuho to facilitate in-house operations and allow for the flexible configuration of monitoring and operation menus by on-site personnel when higher levels of security and confidentiality are required. Mizuho and IBM say they plan to expand the event detection and response PoC and apply it to production environments in the coming year. Mizuho has been an ardent exponent of the use of generative AI. The Japanese bank has been working with Fujitsu to apply generative AI technologies to work processes and tasks including the generation of source code, system development, and maintenance. In June last year the bank went a step further to embed the technology across the enterprise by making Microsoft’s OpenAI platform available to its 45,000 workers.