What exactly is recorded?

Every action taken through Datapace: proposals with their evidence, approvals with the approver, executions with their window, verifications with their result, and every read an agent or an AI system makes of governed context. Each entry carries the actor, the database, the objects, and the time.

Can an entry be edited or removed?

The ledger is designed to be append-only: corrections are new entries that reference the old one. How long the trail is retained, and where, is decided with each partner.

How does this help with the EU AI Act?

Article 12 asks for automatic logging of events over a high-risk system’s lifetime. The ledger records what an agent proposed, who approved it, and what ran, which is the kind of record that article assumes; whether a given deployment falls under it is your assessment.

An audit trail for every action, by a person or an agent

When an agent or a person changes a database through Datapace, the record is the point. Every proposal, approval, execution, and verification is written to the audit ledger with who did it, what it touched, on which database, the evidence it was based on, and who approved it. The trail answers what touched production, when, and why, for your own reviews and for what regulators now ask of systems where AI acts.

Today. Changes to production reconstructed from tickets, chat threads, and memory when someone asks what happened.

With Datapace. Proposal, approval, execution, and verification recorded as they happen, with the evidence and the approver, queryable by database, actor, and period.

What Datapace does here

Who, what, where, why: Each entry names the actor (an agent or a person), the action, the database and objects it touched, the evidence it was based on, and the approver, so the record stands on its own.
Agents and people on the same ledger: An agent’s proposal and a person’s approval are two entries in one trail, so a change is readable end to end without reconciling logs.
Verification recorded too: After a change runs, what was verified and against which evidence is written next to the execution, so the trail says whether the change did what it was approved to do.
Queryable by database, actor, period: The trail is filtered the way reviews are run: this database this quarter, this agent this week, every approval by this role.
What the regulators ask for: The EU AI Act asks that systems where AI acts keep logs of what happened; the ledger is designed so that answer is a query, not a project.

Questions teams ask

What exactly is recorded?: Every action taken through Datapace: proposals with their evidence, approvals with the approver, executions with their window, verifications with their result, and every read an agent or an AI system makes of governed context. Each entry carries the actor, the database, the objects, and the time.
Can an entry be edited or removed?: The ledger is designed to be append-only: corrections are new entries that reference the old one. How long the trail is retained, and where, is decided with each partner.
How does this help with the EU AI Act?: Article 12 asks for automatic logging of events over a high-risk system’s lifetime. The ledger records what an agent proposed, who approved it, and what ran, which is the kind of record that article assumes; whether a given deployment falls under it is your assessment.

Go further

See this on your own data

Bring a use case. We will show you what Datapace reads on your live database, what your experts would confirm, and what the agents would propose first.

Book a call