Trust But Verify

Silver Tier Lesson

This lesson is part of the Silver Tier. Complete Bronze Tier (L01-L07) first.

Coming Soon

This lesson will teach you to implement Human-in-the-Loop (HITL) approval workflows that prevent your AI employee from taking sensitive actions without your explicit permission.

What You'll Build:

• Approval request file format with metadata
• Permission boundaries table (what auto-approves vs requires human)
• Folder-based workflow: /Pending_Approval/ → /Approved/ or /Rejected/

Key Concept: Trust your employee with routine tasks, but require sign-off for anything with real consequences (payments, new contacts, public posts).

---

Placeholder Content

See L00: Complete Specification for the full HITL architecture.

Reference sections:

• "Human-in-the-Loop Pattern"
• "Permission Boundaries"
• "Approval Request Format"