Inspection Readiness Simulator
An inspection is a retrieval test with a conversation attached. This runs the retrieval part: a question is asked, you commit to where the evidence lives and who answers, and only then does the model show what a prepared team would have said.
The commitment comes first on purpose. Reading the modelled answer and then deciding you knew it is not practice, so the system of record and the answering role stay hidden until both have been chosen — and the option lists carry plausible wrong answers, because a question whose only offered answer is the right one teaches nothing about where evidence actually lives. A pick that differs from the model is reported as different, never as wrong: a defensible answer can differ, and this surface has no standing to grade one. Nothing you enter is stored, transmitted or scored, the self-assessment is your own judgement rather than a result, and the questions are synthetic and generic — a good round here is not evidence that a real site is ready. To see what an inspector would be asking about, the requirement-to-evidence trace follows one obligation from the authority that set it to the record that satisfies it.
One question at a time. Commit to a system and a role, then compare.
Show me the audit trail for this batch record, and tell me who reviewed it.
The inspector is testing whether audit trail review is a real activity or a validation-era claim.
Practice questions are synthetic and generic. Nothing entered here is stored, scored externally or shared, and a good result here is not evidence of inspection readiness.