Every technical claim in your system — and in the materials going to a regulator or an investor — put through four examinations.

First, the claims themselves. What would you observe if this were false? A claim nothing could contradict cannot be verified, and cannot be relied on either. You get claims that can fail, claims that cannot, and claims whose wording reaches further than the thing beneath it.

Second, the architecture as described against the architecture as built. For each claim that survives, the code path, the configuration, the flag, the test. What is the denominator, and how was that list produced? Who produces the evidence — the acting system, or something that does not share its interest in the answer?

Third, what is enforced against what is merely recorded. Whether the record of a decision establishes what it appears to establish, and whether anything downstream is obliged to honour a control's result. A check nothing acts on is documentation.

Fourth, what the materials do not say. Undeclared assumptions, dependencies with no named owner, and any place a limitation exists and is not written down.

You get one report: what was examined, against what criteria, what I found with the evidence quoted, what I did not examine and why, what remains unresolved, and what the report does not establish. Interim findings the day I find them — I do not sit on something material until the report.

Five days of work across two to three weeks. Fixed fee, unaffected by what I find. Right of reply on every finding, and no veto on any of them.

I run this on my own published claims first. That record — including the two figures on it that do not reconcile, and why — is at shango.in/transparency.