Use APScheduler when a custom Python agent needs repeatable background work rather than a chat-triggered task: scheduled report generation, periodic source refreshes, delayed follow-ups, retry queues, or maintenance jobs that must survive process restarts. The operator defines the task function, chooses a cron, interval, calendar, or date trigger, configures a persistent job store such as PostgreSQL, MySQL, SQLite, or MongoDB when durability matters, then runs scheduler and worker processes inside the Python service. It is especially useful when the agent already owns the business logic and needs dependable timing, concurrency limits, misfire handling, and restart-safe job state across simple single-process deployments or larger multi-node worker setups. Invoke this instead of using a product scheduler when the schedule belongs inside an agent application and must share its code, context, data store, and deployment lifecycle. The boundary is durable Python job scheduling for agent workflows, not a generic listing for every APScheduler feature or for external cron replacement.