<aside> 🏛️ Make autonomy decisions with evidence and challenge, not optimism. 45-60 minutes.
</aside>
Objective: Make autonomy tier transitions with evidence and structured challenge, not optimism, so the team increases AI independence only when data supports it.
When to use: When the Architect proposes promoting an AI decision type to a higher autonomy tier. Typically 45-60 minutes.
Inputs: Hierarchy of Agency database entry, performance data (minimum 90 days at current tier), rollback procedure test results, monitoring plan for higher tier.
Position: Architect proposes. Sentinel evaluates risk. Scout challenges (devil's advocate role). Sponsor approves or rejects.
Architect presents evidence that thresholds are met.
Sentinel evaluates residual risk at higher autonomy.