Data & AI skills for the public sector

Public organisations answer to citizens, not customers, and that difference runs all the way down. A decision supported by a model has to be explainable to the person it was applied to, defensible to a council or a supervisory body, and traceable long after the person who built it has moved on. The Dutch algorithm register exists precisely because this is now expected rather than optional.

The EU AI Act sharpens it further. Several typical public uses fall in the high-risk category, which means documentation, human oversight and demonstrable competence are legal requirements rather than good practice. Competence is the part organisations most often forget to arrange.

And the workforce is unusually varied: policy advisors, service desk staff, planners, inspectors, drivers. A single generic course reaches none of them properly, while procurement rules make it hard to keep changing supplier until you find one that fits.

The behaviour a public organisation actually needs is this: someone who can explain a model-supported decision to the person it was applied to, in language that person follows, and who knows which decisions may never be automated at all. Competence of that kind is now a legal requirement under the EU AI Act, not a nice-to-have.

Building it takes knowledge of what has to be documented and who oversees a model, skill in the tools your people already have, practised in the browser on your own processes and cases, and the judgement to spot where automation genuinely helps and where it would harm the people it is meant to serve. A policy advisor, a planner and a service desk lead each need a different version of that, which is what the persona framework is for.

Adoption has to outlast a project budget, so champions and community are built in from the start and ownership passes to your own organisation at the end. Growth is measured from a baseline before the start and benchmarked against other organisations — a defensible figure for an accountability document rather than an attendance list.

We work with public organisations including Provincie Limburg.

Across all our clients, cohorts finish at 79% completion over the last twelve months and more than 10,000 learners, against an industry standard of roughly 12%.

Our Data & AI Skills Framework is what lets a programme be aimed at a policy advisor, a planner and a board member separately, and it is what turns "we trained people" into a measured skill level you can put in an accountability document.