As enterprises move from AI assistants to autonomous agents that write code, trigger workflows, call tools, and make operational decisions, reliability can no longer be managed only through traditional DevOps, SRE, or LLMOps practices. A new discipline is emerging: Agentic Reliability Engineering. This session explores how organisations can design, deploy, monitor, and govern agentic systems that operate safely in production. It introduces the core building blocks of Agentic Reliability Engineering: agent observability, task-chain tracing, evals-as-control, policy-at-runtime, human-on-the-system oversight, failure containment, rollback paths, identity for agents, provenance, and decision accountability. The talk will help technology leaders understand what changes when software is no longer only shipped by humans, but increasingly planned, modified, tested, and operated by autonomous systems. The goal is simple: make agentic AI useful, scalable, and trustworthy without losing operational integrity.