Checklists for judging whether an AI's output is actually correct, for common task types.
How to use itLoading…
Advertisement
Checklist for reviewing an AI-drafted customer support reply before it goes out.
Updated Aug 8, 2026
Checklist for verifying an AI-generated summary is accurate and actually useful, not just fluent.
Updated Aug 8, 2026
Checklist for judging whether an AI translation preserves meaning, tone, and register — not just grammar.
Updated Aug 8, 2026
Checklist for validating AI-extracted structured data before it feeds into another system.
Updated Aug 8, 2026
Checklist for verifying an AI-generated SQL query is correct and safe to run, especially against production.
Updated Aug 8, 2026
Checklist for auditing AI classification/labeling output before trusting it at scale.
Updated Aug 8, 2026
A checklist for judging whether AI-generated code is actually correct and safe to merge.
Updated Aug 8, 2026
Checklist for whether a retrieval-augmented answer is actually grounded in the retrieved sources.
Updated Aug 8, 2026
Checklist for reviewing multi-turn chatbot conversations for quality and failure modes, not just single replies.
Updated Aug 8, 2026
Checklist for evaluating AI-generated creative writing against what it actually needed to do.
Updated Aug 8, 2026
Checklist for verifying an AI agent is calling tools correctly and safely, not just producing plausible-looking calls.
Updated Aug 8, 2026
Checklist for catching confidently-stated but wrong or unverifiable claims in an AI's factual answer.
Updated Aug 8, 2026