Not the assistant's summary of itself. A record made independently, sealed on your own machine, that still holds up later when someone who was not there needs to know what happened.
You are personally answerable for work an agent touched — in law, banking, accounting, healthcare, engineering or trading. The fear is not that the model is wrong. It is that you cannot show what it said.
What it records and what it catches →You are doing your own serious work with an assistant and nobody is checking it but you. It catches what is checkably wrong, and keeps a record while it does.
The checks, and the receipt →An exit valve halved high-severity findings in a pre-registered trial across three vendors. An acknowledgment trial came back null. Both are on the page, with the commitments that hold either way.
The floor, and the open questions →One product underneath all three. The record, the checks and the seal are the same; what differs is who is asking and what they are afraid of.
We measured nine models across five vendors, 1,275 runs. Asked to vouch for its own record, no model caught a falsification more than 45% of the time — the best is worse than a coin flip. Asked about another agent's record, every model improved, by 22 to 80 points.
Then we made the forgery internally consistent, and the best judge caught almost none of it. That result is why this product exists and why it says what it cannot do on every page. It is also why the systems doing the work have a page of their own.