When a Db2 issue occurs, resolution relies on time-critical manual analysis. This session explores how agentic AI evolves the operational model - beyond scripted automation toward adaptive approaches for complex runtime scenarios. Through a live demo grounded in client feedback from a large financial services organization, we show how these approaches cut resolution time, and how AI-assisted decisioning supports routine maintenance within governance boundaries. This session will cover how agentic AI complements mainframe automation, architectural patterns for mission-critical environments, and human-in-the-loop models for trusted, auditable operations. Attendees gain practical guidance on trade-offs in adopting agentic AI for mainframe operations.