What the term really means
European Union regulation establishing rules for developing, providing and using artificial intelligence systems.
How it works in practice
The AI Act follows a risk-based model, so duties depend on organisational role, intended purpose and the impact of use.
The decision to make before implementation
An inventory should identify provider, users, data, decisions, documentation and human oversight; a note that the company “uses AI” is insufficient.
How to verify that it works
Review use-case classification, required information, staff competence, incidents and feature changes that may alter risk. Compare results with an agreed baseline and review routine cases, difficult exceptions and human hand-offs separately. A practical AI Act test should have an owner, a review date and a recorded example of an outcome the team will not accept.
The PAR HOUSE Agency approach
We approach AI Act from the workflow rather than a tool demonstration. An inventory should identify provider, users, data, decisions, documentation and human oversight; a note that the company “uses AI” is insufficient. We then build a small measurable scope, record assumptions and expand only after quality review.
