EU AI Act Compliance demo
Each high-risk AI system's Annex IV file drafted from cited sources and kept true after retrains. 1:33
What the demo shows
Builds the technical documentation, risk file and conformity evidence for high-risk AI
The technical file for one high-risk AI system is shown as a dossier with a tab for each of the nine Annex IV sections, each with its fill level, the documents feeding it and the missing items in red; the right column carries the Article 9 risk file, the 17-point conformity checklist and the 2 December 2027 countdown. A classification agent tests each system against Article 5, Article 6 and Annex III and explains its reasoning; drafting agents write each section only from cited passages and never invent text; an assessor agent reads the file as an authority would and raises questions; a change monitor checks every retrain against the change plan. Owners answer with documents and accept residual risk; the AI compliance lead signs the declaration.
Transcript
- EU AI Act Compliance builds the technical file, risk file, and conformity evidence for high-risk AI.
- Four AI compliance teams.
- One breach can cost 3% of turnover.
- Kestrel & Vane provides six high-risk systems, due December 2027.
- An agent shows why the short-listing model is high-risk.
- It ranks applicants. Agents found 23 documents and drafted seven of them.
- nine sections. Every sentence cites its source.
- Click one and the passage lights up.
- Gaps show in red with the article and the person asked.
- An assessor agent reads like an authority and asks for accuracy by age.
- Lucia uploads her sign test.
- The gap is 0.04 in side tolerance.
- Section three turns green.
- Sign-off stays locked until every question closes.
- Every step is cited, signed, and locked.
- 76% done with 414 days to spare.
- Built on Botminds: agents, data, screens, and governance on one platform.
Transcribed from the film’s narration. Captions are on the player.