Operations teams have run playbooks for decades — step-by-step instructions when something breaks. "If replication lag > 30s, restart the replica. If it doesn't come up, failover to standby. Alert on-call. Log incident." A human followed these steps, checking conditions and branching based on outcomes. Errors and delays came from human interpretation and manual actions.
AWS Systems Manager Automation turns this runbook into code — a YAML/JSON document that AWS runs, not a human. Each step is an action (call Lambda, run shell, wait for condition, ask for approval, call API), and the workflow branches on results