
Ask the model to judge its draft against your brief before you judge it. It finds structural problems quickly, and the revision is usually noticeably better.
Why it works
Generating and evaluating are different tasks. When you ask a model to critique a draft against explicit criteria, it focuses on gaps it skimmed past while writing — missing steps, weak evidence, unclear structure.
The general version
Critique your answer against my original request. List the three most important weaknesses — missing information, unclear reasoning, unsupported claims or poor structure. Then rewrite the answer to fix them. Keep everything that was already good.
For plans and decisions
Imagine this plan failed six months from now. List the five most likely reasons, rank them by likelihood and impact, and revise the plan to reduce the top three risks.

AI Mastery Bundle
The next step after the basics: frameworks, precision prompting and a 1000+ prompt library.
$50 (was $100) · Look insideFor facts
List every factual claim in your answer. Mark each as: well-established, likely but should be verified, or uncertain. For anything time-sensitive, say what source I should check.
Do not overdo it
One critique round captures most of the gain. Repeated rounds can make writing blander. If quality stops improving, the issue is usually missing context from you, not more critique.
Our AI Output Checker turns this into a guided checklist with a trust score.
Put it into practice
Open the free AI Output Checker: review an AI answer before you trust it. This article is part of The Prompt Engineering Handbook.
Sources & further reading
Product facts were checked against official sources on 16 October 2026. AI tools change quickly, so check the provider’s current documentation for anything you rely on. Our editorial policy.


