Cases: scenarios of work and what they show

Cases show a customer in advance what the work produces and what it requires from them. These are methodological scenarios, not reports about delivered projects: no client is described or implied.

The case lives in a tube, the conclusion in the text The case lives in a tube, the conclusion in the text
The case lives in a tube, the conclusion in the text

Three scenarios describe the work step by step: red teaming of a multi-agent system, a behavioural audit of an assistant that runs for months, and readiness for compliance. Each lists what the customer receives and what is needed from them.

One case is our own and is published in full: a letter of ours that two detectors read in opposite directions — one found almost all of it machine-written, the other found none. The documents are attached, so that the method can be judged rather than trusted.

Anonymised cases from real work will appear as the work happens, and only with the written consent of the customer. Until then this section shows the form of a case, not a portfolio.

Materials on this topic

Other topics

AI red teaming: what a test shows, and what it does notAI standards: where the rules actually applyAI supply chain: what enters before the model runsMulti-agent systems: behaviour between the agentsIncident chronicle: documented consequences, not forecastsBehavioural audit: how agent behaviour drifts over timePrompt injection: where text becomes an instructionRussian practice: who is held responsible for AIMethodology: how the Institute tests AI systemsData and benchmarks: what our conclusions rest onSociology of AI: relations between the agentsGovernment customers: how to commission a testGovernance and funding of the Institute

Page updated 02 November 2026 · site build 2026.11.13 · ANO «ISAIS»