Maintenance Window
Plan work that requires downtime, degraded service, user notice, or rollback preparation.
Purpose
AI-assisted work needs boundaries for data entry, output verification, disclosure, and accountability. The person using the tool remains responsible for the decision and final artifact.
Operating record
| Cadence | Set a normal review interval and event-driven trigger. |
|---|---|
| Owner | Identify the role accountable for keeping maintenance window current. |
| Source data | List the systems, logs, reports, tickets, or records used during review. |
| Findings | Record exceptions, decisions, blockers, risk changes, and follow-up actions. |
| Evidence | Keep the smallest durable proof that the review occurred and produced a result. |
When to use
- When the work repeats often enough that memory is no longer reliable.
- When more than one person may request, perform, review, or inherit the work.
- When risk, approval, evidence, or handoff needs to be visible later.
Common failure modes
- Sensitive details are entered into unsuitable tools.
- Generated output is accepted without verification.
- Automation hides who approved the final result.
Review guidance
Review this page after a material incident, after a role or system change, and on a normal cadence appropriate to its risk. During review, check whether the owner is still correct, whether inputs are still complete, whether the output is still useful, and whether related pages need updates.
External guidance
These resources are references for terminology, control thinking, or review design. DailyWF adapts the ideas into lightweight operating pages rather than reproducing full standards.
- NIST AI Risk Management FrameworkAI risk governance structure for mapping, measuring, managing, and governing AI risk.
- CISA Secure by DesignSecure-by-design principles for software and technology evaluation.