Breaking the Cycle of Recurring Failures: Generative AI and Root Cause Analysis in Legacy Banking Systems

The banking legacy systems and fragmented ownership often lead to superficial incident resolutions without addressing root causes, resulting in recurring failures. To solve this, we use Generative AI (GenAI) agents combined with the “Five Whys” technique. Our approach draws knowledge from problem descriptions, change request data, and historical incident records. This method uncovered that 70% of the issues previously blamed on management or vendors in a financial company were due to internal code problems. In a case study, we analyzed 5,000 projects and identified over 400 files with the same root cause. Automating root cause analysis moves the process from reactive to proactive, effectively resolving code-level issues.