The approval gate: a pattern for AI agents that write to production systems

Description and use

The approval gate pattern separates decision from action, letting humans approve AI‑agent proposals before execution, reducing errors in critical production systems.

Problem

AI agents that write directly to production systems risk corrupting critical data, eroding trust, and incurring high cleanup costs when writes are incorrect.

Outcome

The approval gate pattern reduced the impact of erroneous writes, improved auditability, and restored trust among revenue team users, enabling continued automation.

Workflow

• 1. Agent generates a proposed action with diff, evidence, and idempotency key.
• 2. System validates the proposal against target schema and permissions (preflight).
• 3. Proposal is queued for human review.
• 4. Reviewer approves or rejects the diff within seconds.
• 5. Executor applies the approved change and logs the full envelope, including inverse action for reversibility.
• 6. System records audit trail for future reference.

Evidence

At 95 percent action accuracy, one in twenty writes is wrong, and those writes compound into a database humans use to make decisions.

Source

Show the reviewer from and to. A human can approve or reject a diff in about two seconds.

Source