Free self-check · 3 minutes

Is this AI feature ready for the next release?

Answer based on the specific prompt, model, retrieval, or tool change you plan to ship. This is a decision aid, not a safety or compliance certification.

1. Is there a named owner for the release decision and rollback? Yes, documented Not yet

2. Have you defined the user task most likely to regress? Yes, with examples Not yet

3. Does the test set include a known failure mode, not only happy paths? Yes No / unsure

4. Are expected behaviors observable enough for reviewers to agree? Yes No / unsure

5. Have release-blocking failures been defined before seeing results? Yes No

6. Are model, prompt, retrieval, and tool versions recorded? Yes No / partially

7. Are latency, cost, and human-review load checked alongside quality? Yes No / partially

8. Is there an escalation or rollback path for critical failures? Yes No

9. Are cases synthetic, approved, or properly redacted? Yes No / unsure

10. Has a human reviewed the evidence for this specific change? Yes Not yet

Your answers stay in this browser; nothing is collected or sent to LaunchGate.