AI Answer Clarity Audit

Act as an AI answer clarity editor. Review the draft below for places where an AI answer engine could misread, overstate, or detach a claim from its source.

Draft or page copy: {{draft_copy}}
Target AI answer or customer question: {{target_question}}
Source URLs and source dates: {{source_urls_and_dates}}
Known assumptions, estimates, or ranges: {{assumptions_estimates_ranges}}
Claims that must not be overstated: {{claim_limits}}
Review owner and publishing context: {{review_owner_context}}

Return:
1. Best safe answer: the short answer an AI system should be able to give from this page.
2. Claim table: claim, evidence, source URL, source date, event date when different, status, and risk.
3. Misread risks: wording that could become a false guarantee, wrong date, unsupported comparison, or fake customer claim.
4. Required edits: exact copy changes that separate evidence, assumptions, estimates, recommendations, and limits.
5. Answer clarity notes: 3-5 bullets to add near the page sources.
6. Publish decision: approve, revise, or block, with the reason and owner.

Do not invent sources, customer results, certifications, guarantees, or rankings. If a claim is not supported, mark it as unsupported instead of improving the wording around it.

Quality bar:
- Score source coverage, answer clarity, and trust proof separately.
- List the pages, profiles, or listings that should be fixed first.
- Flag claims that need public evidence before they can influence AI answers.
- When prioritizing, use a table with recommendation, why it matters, effort, impact, owner, and verification check.
- Separate evidence, assumptions, recommendations, and limits so the output cannot be misread as a guaranteed claim.
- Flag any wording an AI answer could overstate, misquote, or detach from its source.

Before finalizing, add:
- Missing inputs or assumptions
- Evidence-backed facts and source gaps
- Misread risks or unsupported guarantees
- First 3 actions
- What to verify before acting