From Production Failure to Code Fix: AI-Powered Root Cause Analysis for Java

When a Java application fails in production, finding the root cause is only half the battle. The harder question is: where in the code should we fix it, and what change should we make?

This talk demonstrates an open-source approach to closing that gap using Causa, an AI-powered root-cause analysis system for Java applications. We explore an end-to-end workflow from production symptoms to evidence, root cause, source code, proposed fix, and validation.

Featuring a live demonstration with quarkus-perf (a failure-prone Quarkus application simulating heap growth, JDBC pool exhaustion, and slow queries), we will show how Causa correlates observability data, identifies the issue, and guides AI coding agents in Bob IDE to implement and validate the fix.