The following five templates should give you a step-by-step guide to how municipal managers can get the organization on board.
Formål: To create common expectations and take the temperature of the team before you start the AI project.
Introduction (5 min.): project manager presents the goal for the workshop.
Poll "How do you feel about AI in your work?" (5 min.)
Use Mentimeter or Forms → response options: excited, curious, skeptical, concerned.
Case demo (10 min.): show a concrete AI example from another municipality.
Sticky-note exercise (15 min.): participants write potential gains and risks.
Vote priority (10 min.): vote with stickers on the two main wins and one top risk.
Next step (5 min.): owner agreement, schedule and communication.
Output: a photo of the board + download of answers from the poll for reference in the project plan.
Formål: Quickly assess whether an idea is worth the investment.
| Field | What you write | How to use it |
|---|---|---|
| The process today | "Manual journaling of emails" | Provides baseline for measurement. |
| AI-idé | “Mail classification via Promte model” | Describes the solution. |
| Expected gain | "-30% time consumption" | Quantify money or time. |
| Legal complexity | Low / Medium / High | If “High”, schedule DPIA first. |
| Data requirements | Structured, unstructured, sensitive? | Determine security and anonymization needs. |
Score each field 1-5 and add the numbers together – ideas < 15 go into the backlog, > 15 go to proof-of-concept.
Formål: Create a common risk picture on one sheet.
Draw the matrix (probability vs. consequence).
Place risks from the use-case canvas.
Define "yellow-to-green" actions (e.g. pseudonymization).
Revise the map after each sprint.
Use the color codes actively in the management reports – they make complex legal issues easy to understand.
Formål: Determine whether to do a full DPIA.
Examples of questions:
Is data about health, ethnicity or union affiliation processed?
Does the solution profile citizens to make (partially) automated decisions?
Is personal data transferred to third countries?
Is the target group vulnerable (e.g. children, elderly)?
If ≥ 3 “Yes”, the Norwegian Data Protection Authority recommends full DPIA before production.
See the full list here: Mini-DPIA Checklist (10 yes/no questions)
Formål: Document and share the human gains.
Indsamling: Ask employees to fill out a short form when they experience time or quality gains.
Lagring: Save quotes in a shared sheet with fields for name, department, use-case, quote.
Genbrug:
Internal newsletter ("AI win of the week").
LinkedIn posts to strengthen recruitment and citizen communication.
Executive presentations to policy committees.
This is how the toolbox creates value
By using these one-page templates, you don't need long consultant reports. You get a common language, clear prioritization and quick decisions, all in a format that can be copied and pasted directly into Teams, Miro or a steering group report.
