1. Does it answer the actual brief?
Compare the output with your objective, audience and constraints. Check that the work solves the problem you asked about, rather than a similar-looking one.
2. Which claims need checking?
Identify factual statements, numerical claims and cited sources. Verify them against material you can inspect. If you cannot verify a claim, qualify or remove it.
3. Is the context complete?
Look for missing assumptions, exceptions and dependencies. A recommendation may change when the business context changes.
4. Does it expose information?
Review the output for personal, confidential or commercially sensitive information before sharing it with another person or system.
5. What could go wrong if we use it?
Consider the consequence of an incorrect result. Decide whether the work needs a specialist reviewer or a human approval step.
6. Who owns the next decision?
Assign a person to approve the result. Record the assumptions, changes and follow-up checks so the decision can be revisited.