AI systems: the register and the system record
The inventory every duty hangs off — four counts, the register filters, list or board, CSV import, two figures per system — and the seven-tab record each system opens onto.
AI systems (under Position) is the inventory. Before the EU AI Act applies to anything you have to know what you have, so every obligation, assessment, document and piece of evidence in Veritome is anchored to a row on this register.
The register
Four counts run along the top — Systems on register, High-risk systems, Overdue obligations and Unclassified — each a link into the rows behind it. Below them the register filters — Stage · Tier · Role · Source — narrow the table, and a List / Board switch changes the reading: the list is a sortable table (System · Role · tier · Stage · EU AI Act · GDPR · Owner); the board is one card per system.
Every system carries two readiness figures, never added together: its EU AI Act obligations done over total, and its GDPR duties done over total — or no personal data when it processes none. GDPR duties never count toward the Act's stages or gates.
Two ways in:
- Register AI system opens the guided registration and classification flow — see Classify a system: the guided flow, screen by screen.
- Import CSV loads a spreadsheet of systems in one go, so an inventory you already keep elsewhere does not have to be retyped. An imported system is classified from its record like any other.
The system record
Opening a system lands on its record. The header carries the reference, name, risk tier, your role and the owner, two actions — System actions (edit details, Review classification, the read-only regulator view, decommission) and Export dossier — and the EU AI Act journey card: obligations done of total, the next obligation, and the active stage, with the six-stage strip beneath it — Classify · Scope & literacy · Build & document · Conformity assessment · Register · Operate & monitor.
Under it, seven area tabs:
| Tab | What it holds |
|---|---|
| Overview | About this system, and the next steps the engine suggests. |
| Obligations | The journey: an EU AI Act journey tile (stages done, obligations done, the next gate) and a GDPR duties on this system tile, counted apart; the stage strip; then one lane per stage, with that stage's GDPR duties in a lane of their own beneath it. A locked stage says what it waits on; a GDPR lane never locks. Where a standard is switched on, what it asks about this system is listed too. Click a row to open the obligation. |
| Risks | The Art. 9 risk management view for this system — its risks, scores and treatments. |
| Incidents | Art. 73 case files logged against this system. |
| Evidence | Everything filed as proof for this system — generated, uploaded or connected. |
| Documents | The documents this system owes, with the editor that owns each. |
| Activity | The per-system audit feed (Art. 12) — every recorded action, newest first. |
A System card tab appears where a system publishes a card of its own.
The topic tabs of earlier releases — Transparency, Human oversight, Data governance and so on — are gone. Those are domains now: filters on the Obligations register and on this record's Obligations tab, never screens.
Reclassifying
A classification is never frozen. When purpose, data or deployment context changes, open System actions → Review classification: the same guided flow opens with the current answers, and on confirmation the engine replaces the obligation set to match, the dossier records the change, and a fresh classification evidence record is minted. Nothing changes until you confirm on the last screen.
What the register is for
Regulators expect an operator to know, and be able to show, every AI system in scope. A system you assessed and found out of scope is a better answer than one you never looked at — so register generously and let classification decide.
Still stuck? Open a support request in your workspace, and a person will answer there and by email.