This article outlines how to create a Production Engineer agent that quickly identifies and contextualizes service failures in complex systems. It emphasizes the importance of structured memory and effective communication in avoiding confusion during incidents. The design relies on GraphRAG for managing dependencies and historical context.