Improve AI output using a quality checklist
Refine AI output against clear acceptance criteria rather than asking for generic improvement.
1. Set acceptance criteria
Define what correct structure, length, tone, evidence and factual accuracy look like.
2. Ask for a gap analysis
Request a list of unsupported statements, missing data and unclear wording. Do not equate model self-review with independent verification.
3. Revise only the necessary parts
Preserve verified content and request focused revisions for gaps; check again before publication.
Vague request
Make it more professional.
Improved prompt
Review the draft against: [TARGET AUDIENCE], [MAX LENGTH], [REQUIRED SECTIONS] and [SOURCE RULE]. List gaps before rewriting; preserve confirmed facts and flag new assumptions.
Key takeaway
Objective rubrics make revisions faster and more consistent.