AI maturity in the organisation: how to assess where you are and what to do next
An organisation's AI maturity is measured on five levels, from level 1 (no structured AI use) to level 5 (organisation optimised by AI, in a continuous learning loop). Most French companies in 2026 sit between level 2 (uncoordinated experimentation) and level 3 (controlled deployment on a few processes). Assessment rests on four dimensions: actual usage, skills present, governance and culture.
Why assess AI maturity before any other action?
Most AI projects in business fail not for lack of available technology, but because of a mismatch between project ambition and the organisation's actual maturity. Deploying an advanced AI tool in an organisation that has not yet stabilised its core processes is like putting a Formula 1 engine in a car whose brakes have not been checked.
Maturity assessment has three practical uses. It avoids purchases of unused tools. It prioritises training investment. It provides a documented baseline to track progress.
The 5 levels of AI maturity
Level 1 — Non-existent. No AI system is used intentionally. Employees may use personal tools individually, without oversight. There is no usage policy, no training, no governance. This is the situation exposed to the AI Act Article 4 obligation without even being aware of it.
Level 2 — Experimentation. AI tools are used ad hoc by some employees, usually on their own initiative. Uses are disparate, undocumented and not evaluated. There is no usage charter. This level is "shadow AI": the organisation uses AI, but does not really know it.
Level 3 — Controlled deployment. The organisation has identified 2 to 5 priority use cases and deployed tools on these scopes in a controlled way. A usage charter exists, at least one awareness session has been held, and a point of contact is appointed. Results are measured informally.
Level 4 — Systemic integration. AI is integrated into several core business processes. AI skills are developed in a structured way, with differentiated learning paths by profile. Formal governance exists (AI committee or equivalent), indicators are tracked, and feedback loops are in place.
Level 5 — Optimised organisation. AI is a lever for continuous improvement. The organisation learns from its AI deployments, shares experience systematically and adapts its strategy based on measured outcomes. This level is rare in 2026.
How to move up a level
From 1 to 2: map actual uses, raise awareness of AI Act obligations, appoint a coordinator.
From 2 to 3: prioritise use cases, draft a charter, train regular users.
From 3 to 4: structure skills frameworks, integrate with HRIS (Human Resources Information System), formalise governance.
From 4 to 5: systematise after-action reviews, measure ROI, align strategy.