AI literacy measurement and evidence

Anyone can show their people watched the training.

Almost nobody can show what those people learned. Testamur measures what your workforce actually understands about AI, assigns training to the gaps we find, and re-measures to prove it changed. You end up with evidence, not attendance records.

Check where you stand See how it works

The loopMeasure · Close · Prove

Training on its own proves nothing

Most organizations bought a course library and assumed the evidence came with it. It did not. A completion rate tells you people opened something. It does not tell you whether your finance team knows when to distrust a generated number, or whether your recruiters understand what they are legally required to oversee.

We run the whole loop, because each part is worthless without the others.

01 Measure

Baseline

A scenario-based assessment across five constructs, banded by role. Sentiment measured alongside it and reported separately.

02 Close

Assign

Training assigned from the results, not from a catalog. People get the modules their answers say they need, and skip the ones they do not.

03 Prove

Re-measure

The same instrument, six months later. Movement is the finding. This is the part almost nobody else does.

Output

Evidence file

Who was assessed, what they knew, what they were assigned, what changed. Dated, versioned, ready to produce.

Why nowArt. 4 · Live

The obligation is already in force

Article 4 of the EU AI Act requires organizations to support a sufficient level of AI literacy among staff and anyone operating AI on their behalf. It has applied since February 2025, and national authorities began enforcing it in August 2026.

It reaches organizations with no European entity at all, because scope follows where AI output is used rather than where you are established. A US company with European customers is generally in scope and frequently does not know it.

The regulation deliberately does not define what sufficient means. That leaves you to set the standard and to be able to show your reasoning, which is precisely the problem Testamur exists to solve.

Read what Article 4 requires

FitWho this is for

Built for the middle

Enterprise listening platforms are designed for organizations above three thousand people and sold on annual per-seat contracts with long deployments. Free maturity checklists are self-scored and skip the workforce entirely. Between those sits almost everybody.

Size200 to 2,000 peopleLarge enough to carry the obligation, too small to be served well by an enterprise platform rollout.
TriggerEU exposure through customers, staff, or outputOften surfaced by a customer security review, a board question, or counsel.
OwnerPeople or L&D, with Legal or Compliance co-signingThe obligation usually lands between two functions, which is why it stalls.
SignalYou already bought AI tools and trainingThe gap is not effort. It is that nobody measured whether any of it worked.

TrainingIncluded

The training is part of the product, not an upsell

Our role-based AI literacy program is included in every engagement above the baseline assessment. It is not a catalog you point staff at and hope they browse. Modules are assigned to individuals based on what their own answers showed they were missing, which is what makes the assignment defensible as proportionate and role-appropriate.

It also means the training carries evidence with it. Completion is recorded against the gap it was assigned to close, and the re-measurement six months later shows whether it did. That chain, from measured gap to assigned learning to demonstrated change, is the thing you would want to be holding if anyone ever asks.

On the free courses. Microsoft, Google, and the European Commission all publish AI literacy material, and some of it is good. Use it. What none of it gives you is a record of who needed what, what they were given, and whether it worked for your organization. That record is the product.