These are not case studies and they are not marketing. Each one is a measurement Certisyn published on its own account, built from files anyone can download, with the digest of every file published beside the finding. A record names what it could not determine as plainly as what it found, and no page is served unless the data behind it reconciled. A stale page is better than a page asserting a reading that did not verify.
Certisyn does not sell a determination, a licence or a subscription to any party named in a published record, and discloses any approach from one. The rule is published on the records themselves and in the machine-readable catalogue, where it can be checked rather than taken on trust.
Every input is a public file that needs no account and no key. The digest of each one is taken at the moment it is served and published beside the finding, so anyone can obtain the same files and reach the same result.
Each record states what it does not look at and what it declines to conclude. A negative from an instrument whose coverage has not been characterised is not a negative, and saying so is part of the record.
Every publication is gated end to end against the live surface. If a single check fails, the previous page keeps serving.
The 33 operative paragraphs of General Assembly resolution 80/251 reconciled, one at a time, against the documents the United Nations has published since it adopted them.
Method CR-1.0 · rebuilt 2026-10-05
The same 33 commitments read as a worklist: what would close each open line, whose desk that sits on, and how long it has been waiting. 25 of the 33 close by publication, citation, attestation or counting, and not one of those...
Method CR-1.0, closure reading · rebuilt 2026-10-05
What the five permanent members of the Security Council were recorded as saying at the General Debate, set against the mandate reform their Assembly had adopted six months earlier.
Method CR-1.0, proof of concept · rebuilt 2026-09-21
Three sealed volumes over the same 193 member states: conduct against the five truce stipulations, the structural conditions underneath it, and what the states actually undertook in the primary documents.
Method TO-2.0, SC-1.1 and CR-1.0 · rebuilt 2026-10-05
A provisional reading of the Milano-Cortina 2026 Olympic Truce window over all 193 member states, taken from the preliminary public series because no vetted release codes 2026 yet.
Method TO-2.0, provisional reading · rebuilt 2026-10-05
One page for each of the 193 United Nations member states: what was recorded on its territory inside the Paris 2024 and Milano-Cortina 2026 truce windows, whether it put its name to the resolution, and how much of the...
Method TO-2.0, per-state reading · rebuilt 2026-10-05
22 statements about the OpenAI agent that gained non-public access to a Services Australia portal, reconciled against the documents published since by OpenAI, the Australian Government and the press.
Method CR-1.0 · rebuilt 2026-09-30
18 statements of fact from a public post about the Firmus Technologies IPO, read against the prospectus reporting, the business register and the press, with the open lines named and the 24 hours of coverage after the float was...
Method CR-1.0 · rebuilt 2026-10-10
30 statements about AI oversight in US national security systems, the OMB AI memo it sits beside, and the reporting around both, reconciled against the primary documents, with the open lines named.
Method CR-1.0 · rebuilt 2026-10-01
One text can carry hundreds of separate mastheads. Over a contiguous 24 hours of the public GDELT record, this counts how many of the articles published were copies of each other, groups the domains that keep carrying the same...
Method MA-1.0 · rebuilt 2026-10-09