The full arc of an agent from design and build through deployment, monitoring, and eventual retirement.
Traditional software front-loads the work: build, test, ship, then maintain. Agents flip that. They keep changing after launch, so most of the real management happens in production: reviewing behavior, catching drift, updating rules, and deciding when an agent should be changed or retired.
Related: AI Agent Governance, AI Agent Monitoring, Agent Sprawl
See also: What Is AI Agent Governance?