# AI guardrails checklist
- Define use-case and expected outcome
- Complete risk screen (NIST AI RMF)
- Document data sources and consent posture
- Human-in-the-loop defined
- Monitoring plan and rollback path
- Decision log entry created