Changelog
What we shipped, and when. For changes to the regulation itself, see AI Act updates.
Live demo, and a typeface that actually loads
A read-only demo workspace is now one click from the homepage — real classified systems, generated obligations, running controls and produced documents. No form, no sales call. Writes are refused server-side, so the data stays intact for the next visitor.
Inter was being downloaded on every page load and then never applied — a missing Tailwind font configuration meant the browser silently fell back to the system typeface. Fixed, and headings now use a distinct display face.
Cards now sit on genuinely separated surfaces with a light top edge and a hover lift, instead of flat panels carrying shadows that were invisible against a dark background. The dashboard leads with one headline number rather than a wall of equal-weight cards.
Organization, WebSite and SoftwareApplication structured data, an /llms.txt manifest, an About page, and social cards on every shared link.
Nightly backups
The backup job is scheduled by the deployment itself rather than depending on a separate manual setup step. Dumps run nightly, are kept for fourteen days, and fail loudly instead of leaving an empty file that looks like a backup.
EU AI Act documents, framework crosswalks and a help centre
Fundamental Rights Impact Assessment (Art. 27), EU Declaration of Conformity (Art. 47), post-market monitoring plan (Art. 72) and GPAI technical documentation (Art. 53) — guided, per system, exportable as PDF.
For teams who answer more than one questionnaire: honest mappings of the AI-relevant controls, without pretending to be a full security-compliance platform.
Written answers to the questions people actually ask, searchable, inside the app — so you are never blocked waiting on a reply from us.
Nine in-depth EU AI Act guides, eight sector-specific pages and honest competitor comparisons — free to read, no sign-up.
Continuous monitoring, payments and the governance registers
Sixteen AI Act controls checked by a deterministic, rule-based engine — never a language model. Each run captures its own evidence and raises a notification the first time a control starts failing.
Risk register (Art. 9) with a 5×5 heatmap, incident register with the Art. 73 fifteen-day clock, AI model registry, and AI literacy training records (Art. 4).
Public pricing you can pay from the website — no quote, no sales call. Tax and invoicing handled by our merchant of record.
A shareable public page proving your AI compliance posture to customers, and a way to send AI assessments to your own vendors.