AI literacy measurement and evidence Instrument rev. 1.1Assessed, not attended San Francisco

Anyone can show their people watched the AI literacy training. Almost nobody can show what those people learned.

We measure what your workforce actually understands about AI, assign learning to the gaps we find, and re-measure to evidence the change. You end up with a record, not attendance data.

Take the five-minute assessment or see how it works

Northwind Logistics · baseline 412 of 430 assessed

Cohort score

61 /100

Weakest construct

Confidence gap

+1.4

C5 Calibration · confidence gap by function

UnderstatedOverstated

Sales +2.4

Operations +1.7

Finance +1.1

Engineering −0.8

Reading: sales overstates most and scores lowest on boundaries. That pairing is the highest-exposure finding in this cohort.
Illustrative data. Not a client.

Instrument

v1.1, in build. Ships December 2026

Pricing

Published, from $9,500

Individual results

Never shown to your employer

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.

EvidenceThe second measurement

A baseline proves you looked. Only the re-measure proves anything changed.

Six months on, the same constructs, using parallel forms so nobody is answering remembered items. The hollow dot is where the cohort started.

Baseline, Jan Re-measured, Jul

C1Comprehension

+5 pts

C2Applied judgment

+18 pts

C3Boundaries

+27 pts

C4Oversight

+5 pts

What did not move: the sales confidence gap held at +2.1. The module reached them and their boundaries score improved, but the overconfidence did not follow. A report that shows only gains is a report nobody believes.

Illustrative cohort. Not client data.

What someone can actually ask you, and what you can answer today.

A customer security questionnaire, a vendor review, a board question, a regulator. They do not ask whether you ran training. They ask what it produced. The blanks are where most companies are today.

Take the five-minute assessment or read the methodology

AI literacy question a customer or regulator can ask Your recordWith Testamur

§1Course assigned

YesYes

§2Completion rate

100%100%

§3What they understood

C3 weakest, 61%

§4Where they are overconfident

+1.4 overstated

§5Why that training and not other training

assigned from result

§6What changed six months later

+22 points

§7Defensible on request

dated, versioned

Right column illustrative. Figures shown are not client data.

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 people and upLarge enough that you cannot answer the question by asking around. There is no rollout, no integration and no per-seat contract, so there is no size at which we stop being useful.
TriggerA customer security review or vendor questionnaireMost organizations meet this commercially before any regulator is involved. EU exposure through customers, staff, or output is the underlying condition.
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 exactly the kind of record an obligation of effort is judged on.

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.