AI training for public administration
AI training for administrations: on your own casework.
Two days on site, small group. Your staff work on a case from their own organization and build a working AI routine for it.
Arrange a first meetingThe AI training is a two-day workshop format for local administrations. Participants bring a case from their own unit, such as citizen enquiries, reporting or recurring review tasks, and build a working AI routine for it. The training also delivers the AI literacy that Article 4 of the EU AI Act requires from deployers of AI systems.
A real case, not an exercise
The training works on cases from your organization, not on invented examples. Short input sessions alternate with hands-on work on the case, with the trainers working alongside. Practical work makes up at least half of the training time.
A result for daily administration
The result is a working, verified routine, not a slide deck. On day two the draft becomes a recurring routine; at the end, every team presents what runs in its own area from then on.
Model competence
Participants learn to judge which model suits which task and when not using AI is the right decision. The relevant models are compared, from Claude to other established systems and open-source alternatives.
Verification before the decision
Participants see how incorrect AI output ("hallucinations") arises and how to spot it. This includes plausibility rules, cross-checks and clear sign-off points, so that no unverified result reaches an administrative decision.
The legal framework in daily work
What the EU AI Act and the GDPR mean for the participants' own cases and data is a fixed part of day two. Their use cases are assigned to risk classes; data flows and personal data are reviewed.
The Article 4 training duty
Article 4 of the EU AI Act requires deployers to ensure a sufficient level of AI literacy among their staff; the duty has applied to municipalities since February 2025. The training counts directly towards this duty and is documented accordingly.
A curriculum with defined outcomes
The two days follow a defined learning path with outcomes per section: confident use of the established models in daily administration, deepened on request for one tool of your choice, up to Claude Code for your IT unit. Participants practice in the tools they will keep working with.
Trainers from delivery
Our trainers build and run their own AI systems, from assisted-answer functions to production automations, and bring that experience into the training. AKARA Solutions GmbH from our consortium is an official Anthropic partner.
How we start.
- 1
Preparation
Participants, cases and focus areas are agreed in advance with the units involved. Tools are selected with a transparent overview of license costs.
- 2
Day 1: understand and build
Models compared, demonstrations on real examples, then work on the participants' own case. Every team produces a first working draft.
- 3
Day 2: verify and hand over
Limits and risks, the duties under the EU AI Act and the GDPR, verification of results. The draft becomes a routine for daily administration; the results are presented at the end.
- Now
A first meeting to take stock
Send us your request via the form on this page. We reply within one business day with a proposed date.
Frequent questions.
Who should attend and what prior knowledge is required?
No prior knowledge is required; the training runs from the basics to application. It suits staff who bring a case of their own, for example from citizen services, casework, finance, IT or HR. The group stays small so every team has a trainer within reach throughout.
How is personal data handled?
No real citizen or employee data is processed in the training itself; work is done on anonymized training data without personal references. No commissioned processing within the meaning of Art. 28 GDPR takes place. How the routines then run with real data in line with data protection law is part of day two.
Does the training meet the Article 4 duty under the EU AI Act?
It counts directly towards it: Article 4 requires deployers of AI systems to ensure a sufficient level of AI literacy among their staff. The two days cover model competence, handling incorrect output and assigning your own use cases to risk classes; attendance is documented.
How is the training treated in budgeting and procurement?
Training services are operating expenditure; nothing is capitalized. For a first orientation we provide a non-binding cost estimate after the initial meeting; a binding offer follows once the scope is settled. Services of this size are procured by many municipalities as a direct award; your municipality's procurement rules and thresholds are decisive.
What happens after the training?
The routines you built stay in your organization and keep running; the first training enables multipliers, and further sessions for additional teams follow on request. Also on request, ongoing user support follows as a separate module: everyday questions, new use cases, model changes. Scope and response times are agreed in advance.
Further services
The assessment records your administration's AI applications, maps them to the duties under the EU AI Act and the GDPR and provides the basis for leadership and council deliberations.
AI delivery: from the as-is process into live operation.From process capture through the target process to applications your administration keeps running itself, integrated into your systems and the e-file.
Let us take the next step together.
Book an initial assessment meeting.