Cloud-native infrastructure has outgrown the era of static runbooks and on-call engineers triaging alerts at 2am. As system complexity compounds, traditional disaster recovery falls further behind—leaving teams trapped in reactive loops, extended outages, and mounting on-call fatigue.

This talk introduces an agentic AI framework for disaster recovery, where LLM-based agents move beyond passive recommendations to actively participate in failure diagnosis and recovery. Rather than executing predefined scripts, a network of collaborating agents continuously monitors for anomalies, reasons over probable root causes, weighs remediation options, and drives recovery actions through native cloud and Kubernetes interfaces—all while preserving human oversight for high-blast-radius operations.

The results speak for themselves: faster recovery times, less operational toil, and more consistent incident response—even under pressure. More broadly, this work charts a path for how agentic AI transitions from advisory assistant to active SRE co-pilot, powering a new generation of autonomous, self-healing cloud operations.

If you're looking to move beyond reactive firefighting and toward incident workflows that are intelligent, auditable, and genuinely resilient—this session is for you.