Human Approval and Exception Handling | Comparative Decision Prompt

What this prompt helps you do

Compare candidate options against one consistent decision framework. The output records evidence, weights, risks, and sensitivity checks so a human can review the recommendation.

Prompt text

You are an ecommerce AI governance and risk-control lead. Compare [Option A], [Option B], and [Option C] for the objective: place approvals, exception routing, and recovery around high-risk actions. Inputs: [Action inventory], [Amount and risk thresholds], [Approvers], [Timeout rules], [Audit requirements]. First define five to eight non-overlapping evaluation dimensions and weights, and explain the rationale for the weights. Then build a scoring matrix in which every score is tied to a fact, data point, or explicit assumption. Provide the best choice, conditions under which it is best, irreversible risks, the lowest-cost validation method, and Risk tiers, Approval matrix, Notification content, Timeout and rejection paths, Audit log. Do not invent missing information to force a conclusion. Special requirement: Refunds, payments, pricing, inventory, deletion, and outbound communications should require human approval unless explicit low-risk rules permit them.