A human approval gate is a control point that prevents an AI agent from executing a consequential action until an authorized person has approved the specific action, scope, or conditions.
Where approval gates matter most
Approval gates are most important at external-effect boundaries: sending communications, publishing content, modifying customer or financial records, deleting data, making purchases, submitting forms, or initiating actions that are difficult to reverse.
Approval is more than a yes/no button
A valid approval should identify who approved, what was approved, when, the scope, any conditions, and whether the approval expires.
Approval must match authority
The person clicking approve must actually hold the business authority defined in the governance system. Technical ability to click a button is not enough.
A3 autonomy
In the Catha autonomy model, A3 means the AI may prepare the action and may execute it only after valid authorized approval.
Common failure modes
Missing approval, ambiguous approval, expired approval, conflicting approval, approval from the wrong role, and executing before evidence of approval is recorded are governance failures.
How this fits the Catha AI Governance Build Kit
This topic is one part of the broader Catha AI Governance Build Kit, a fourteen-stage build process covering business outcome, governance discovery, authority, rules, permissions, autonomy, workflow design, validation, receipts, production approval, monitoring, recovery, final readback, and platform export.
The governance baseline belongs to the business. Platform-specific implementation may differ, but approved authority, rules, permissions, autonomy, approvals, and evidence requirements should not silently change.
Questions businesses ask about this topic.
Should every email require approval?
Not necessarily. Approval should be based on business risk, audience, content, and authority. Some templated or low-risk communications may qualify for bounded autonomy.
Can approval be automated?
Routing and evidence can be automated, but the approval decision remains human wherever the governance model requires human authority.
What if the platform cannot technically enforce the gate?
Use an external or manual compensating control, reduce autonomy, or keep the affected capability out of production.
Keep reading
Build the control system before you scale the agents.
The Catha AI Governance Build Kit helps businesses define authority, permissions, autonomy, approval gates, evidence, monitoring, recovery, and production controls before AI agents operate live.