AIP-C01 Web TestEngine demo

Select all that apply.

• A

  Run a regression set of representative prompts and compare quality, safety, and formatting against expected outcomes.

• B

  Use canary or staged deployment metrics to compare the new version against the previous version before full rollout.

• C

  Skip testing if the system prompt is shorter than the previous version.

• D

  Evaluate only one successful happy-path prompt because GenAI behavior is nondeterministic.

• E

  Delete prior evaluation results so reviewers focus only on the latest output.