The outcome you are walking toward
Confidence that the evidence model is real: you have reconstructed one actual action end to end — who did it, under which brand context, which gate released it, and what resulted — from the record itself, not from a narrative assembled for you.
1. Know what is captured
Before the live session, read the audit trail & evidence model: capture happens at the moment of action, not afterward. That property is what you are verifying — everything else is detail.
2. Pick one action worth reconstructing
Choose something consequential and specific: an approval gate release, a split configuration change, a role grant. Vague actions produce vague evidence; a named action forces the record to be specific too.
3. Reconstruct it in the console
During the review session, ask for the record of that action: actor, timestamp, brand context, gate decision, outcome. Cross-check the gate release against approval gates — who was allowed to release it, and is that who did?
4. Push on the failure posture
Evidence systems are proven at their edges. Ask what happens when capture itself fails — the same fails-closed posture as database isolation should apply. What 'counsel-grade' actually means is the short essay on why this standard, and the diligence resources page is where follow-up materials and briefing requests live.