Approve AI budgets on evidence, not on conviction.

Four questions, and each has a measurable answer. Who in this organisation can actually do AI work today. Which of our planned initiatives are ready to run and which are being funded on hope. What did the last cycle return. And what evidence supports each of those answers.

Most AI board updates answer none of them, because they are assembled from what teams report about themselves. A measured baseline answers all four in the same terms every quarter, which is what makes the trend meaningful rather than the snapshot.

By instrumenting the workforce rather than surveying opinion. People work through realistic scenarios; the results roll up by team into one index. Culture is measured separately, because capability without permission produces nothing, and a company can have plenty of the first and none of the second.

What that avoids is the maturity grid. A five-point scale filled in by the people being assessed tells you how they would like to be seen. It is why two companies at "level 3" can be nothing alike.

No, and it is built so that it cannot be. Individual results are never attributed to a named person for their employer. Cohort figures are withheld entirely below five people, enforced in the queries that produce them rather than in the interface that shows them. No AI model reads an individual’s result on a sponsor’s behalf.

This is not a courtesy. It is the reason people answer honestly, and honest answers are the only kind worth measuring. It is also, in practice, what gets a deployment approved by HR and by a works council.

Days. Setup is a guided runway rather than a consulting mobilisation, and the instruments are short enough that a cohort completes them inside a week.

The first artefact is a readiness baseline by team with the gaps named. The board pack follows from it, which is the point: the document is how the answer arrives rather than the thing being produced.

• AI readiness
• SkilliScore
• AI Opportunity Matrix
• Privacy floor (n ≥ 5)