Make the expectation explicit
Assert observable product contracts: a response appears, required controls work, citations are present when promised, and an error offers a way to recover.
Put it into practice
Keep semantic quality evaluation separate from UI E2E. In AuraCheck, make the interface criteria explicit and use human review when a response requires judgment.
Try this on your next release
Choose one journey, define the expected result, and inspect the evidence from a completed run. Expand the suite after the first test is useful.