AI system register
Every AI system in your product — model, vendor, data subjects, EU exposure, owner, deployment scope.
Art. 26 · deployer record
For B2B SaaS selling AI into Europe
A living record of your AI systems, article-mapped obligations, and the evidence behind them. Exportable as a System Readiness Packet the customer's legal team can read without a call back to you.
The trigger
The first AI Act enforcement moment most SaaS teams actually feel isn't a letter from a regulator. It's a 30-to-60 question AI governance addendum that lands during security review for a deal you were about to close.
“Which articles of the EU AI Act apply to your system?” “What's your risk classification?” “Who is the named control owner for Article 14 human oversight?” “Can you provide evidence the system was reviewed in the last 90 days?” Legal forwards it to compliance. Compliance forwards it to engineering. Engineering sends back a Notion link. The deal stalls.
Attevera is the record those questions assume you already have.
What the packet contains
Each part is a real deliverable — not a slide or a checklist. Exportable as a single PDF per system.
Every AI system in your product — model, vendor, data subjects, EU exposure, owner, deployment scope.
Art. 26 · deployer record
Each system's obligations mapped to the exact articles that apply, with plain-language rationale.
Art. 5 · 6(3) · 9–15 · 50
Named humans — legal, product, engineering — own each obligation. No diffused accountability.
Art. 14 · oversight
Linked controls, source URLs, acknowledged reviewers, 90-day staleness flags.
Art. 12 · 17 · record-keeping
Every classification, assignment, evidence upload, and sign-off preserved with actor and timestamp.
Art. 12 · logging
One exportable PDF per system — the artifact you hand to procurement, counsel, or a regulator.
Deployer-facing output
Against the DIY version
AI system list in a Notion page that nobody updates after quarter-end
Register that's the source, not a derivative
Obligation mapping in a Google Doc a consultant wrote six months ago
Article mapping updated when systems change — not when the doc gets opened
Evidence in three Drives, two Slack threads, and one person's inbox
Evidence with links, review dates, and staleness flags in one place
A PDF generated once for a deal, already stale by the next deal
A packet that marks stale automatically when a system changes
Answering the same 40 governance questions from scratch every customer
Export, send, move on
Honest scope
In scope
Out of scope
Pricing
Starter covers up to 5 AI systems and 3 team members — enough for a single product with one or two AI features. Growth at €332/mo billed yearly opens it up to 25 systems and 10 members, which is where most AI-heavy SaaS teams settle. No procurement dance, no demo gate — sign up, import your systems, export a packet the same day.
Questions teams ask
Next step