Guides
Practical guides for permission planning.
Use these guides to decide what a customer-facing AI agent may access, change, approve and commit your business to doing. Each article focuses on one planning decision you can document in the worksheet.
For official documentation, implementation tutorials and software, explore the agent permission-design resource library.
What Is Permission Design for AI Agents?
A plain-language introduction to deciding what an AI agent may access, change, approve and commit a business to doing.
How to Decide What Your Customer-Facing AI Agent Is Allowed to Do
A practical guide to building a complete permission plan with allowed actions, limits, approval triggers and evidence requirements.
How to Decide What Your AI Agent Can Do on Its Own
Separate independent actions from customer confirmations, employee approvals and actions the agent should not perform.
Designing the Handoff When an AI Agent Reaches Its Limits
Plan what happens when a request exceeds the agent’s authority, evidence is uncertain or employee approval is needed.
When an AI Agent Makes a Mistake: Design for Correction and Recovery
Decide how wrong actions would be detected, paused, corrected and explained before granting the agent authority.
Can Your AI Agent Undo Its Mistake? A Worked Refund Example
A fictional refund example showing which mistakes can be reversed, which need a separate correction and which leave consequences that cannot be fully undone.
How to Check Whether Your AI Agent Followed Its Permission Plan
Compare recorded actions, approvals and system outcomes with the permission plan the agent was supposed to follow.
From Permission Plan to Practice: How AI Agent Rules Fit Into Your Platform
See how a permission playbook can become business requirements for platform settings, workflows, tool checks and approval paths.
When an AI Agent’s Tool Fails: Pause, Verify and Avoid Duplicate Actions
Plan what happens when a refund, account change or other action times out before retrying or telling the customer it failed.
Who Owns Your AI Agent’s Permissions, and When Should They Change?
Assign permission owners, track connected systems and review permissions when policies, tools, roles or agent behavior change.
