The board will ask whether the AI can be turned off, and whether anything it did can be undone. Most platforms answer in a slide. This one answers on the screen. Pull the switch below and watch the AI stop, the trace disappear, and the deterministic layer keep working.
This is an illustration of the behaviour, not a live console. In the product, the counts are read from the gateway itself.
Coverage gaps. Orphaned citations. Broken lineage. Evidence whose source went dark six days ago and nobody noticed. Amzaa finds all of it with no model, no prompt and no inference.
Because it is deterministic, the same inputs give the same findings every run. There is nothing to hallucinate and nothing to explain to a regulator. It is the same for every tenant, including the ones who buy no AI at all.
And it has no off switch. That is deliberate.
A regulatory deadline in Amzaa carries the URL it came from and the line of the page where the date was printed. A human verifies it once. Then, with the AI switched off, the platform keeps re-reading that page and tells you the day the proof disappears.
And when the agent finds no date it can point at, it proposes nothing. A wrong date on a filing deadline is a liability, so it does not guess. The refusal is not a failure. It is the behaviour we designed for.
The graph, the floor, the full system of record, and a zero AI trace you can hand to an examiner. The whole platform, without the part your board said no to.
Governed agents on the same floor. They propose, a human approves, the engine writes. One action takes it all away and leaves nothing behind.
Regulated enterprises and PE firms, with early access and real influence on the roadmap.
Apply to the programme