Patch night. Nothing should go wrong. It does.
The scenario:
Patching is routine until the night it takes a client offline.
The difference between a clean maintenance window and a disaster is a runbook that says exactly what to do, in what order, and how to back out if it goes wrong. Routine work still needs a process when the stakes are uptime.
The prompt:
You are building a maintenance window standard operating procedure for a managed service provider applying patches or updates to client systems.
Works in any AI tool.
Context:
- What you patch: [systems]
- Client notice norms: [how much warning you give]
Build:
- Pre-window checks
- Client notification timing
- A backup and rollback plan
- The order of operations and verification after each change
- A go or no-go decision point
- A short communication template for before and after the window