Human-in-the-Loop Design Audit: Where Approvals Add Safety and Where They Add Noise
Check approval steps for usefulness.
Added October 15, 2026. Written with AI assistance and checked for clarity and placeholders; not reviewed by a lawyer, accountant, clinician or other professional. How these prompts are made
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
Audit the approval steps in our automation. The workflow steps with which require human approval and the volume of approvals per day: [WORKFLOW_AND_APPROVALS]. Approval outcomes (approved, edited, rejected) for the last month: [APPROVAL_OUTCOMES]. Actions with the largest consequences: [HIGH_CONSEQUENCE_ACTIONS]. Calculate approval rates, edit rates and rejection rates by step with the arithmetic. Identify steps where approvals are nearly always granted without edits (possible rubber-stamping) and high-consequence steps with no approval. Make a table with Step, Volume, Approve rate, Edit rate, Reject rate, Consequence, and Recommendation (keep, remove, batch, add detail, require two people). Propose a better approval interface (what is shown, defaults, time limits) and a sampling check for auto-approved items. Do not assume the right level of oversight without my context; present options. Add a worked table: step 'send reply' 400 approvals, 394 approved unchanged (98.5 percent), 4 edited, 2 rejected; step 'refund over 100' 18 approvals, 7 edited, 3 rejected. Conclude that the first may be rubber-stamped and the second adds real value, with a suggested sampling plan.
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.