From 'are we even allowed to?' to five concrete AI use cases and two pilots, in a single training day

The first AI training at the Municipality of Helmond: a one-day programme for i-advisors on generative AI, prompting and AI agents, preceded by an intake survey. Outcome: a 9/10 rating, 100% pilot interest and 5 concrete use cases.

The challenge

I-advisors at the Municipality of Helmond advise the organisation on information management, but lacked a shared view of what AI can and cannot do inside the municipality. The intake revealed a wide range of starting levels (self-assessment 1–5): most used ChatGPT or Copilot only lightly, mainly privately and 'instead of googling'. The AI policy was known but still brief and not yet lived, and without paid licences experimenting was hard. The main concerns: output quality and accuracy, citizen privacy, transparency and governance.

The approach

We started with an intake survey among all i-advisors to map starting level, tools used, desired use cases and concerns. Based on that we built a tailored one-day training at the NextPrompt Academy on Brainport Human Campus. Morning: how generative AI works, effective prompting, reliability and risks, and assessing AI output. Afternoon: AI agents and workflows, and building their own use case. Throughout, we practised on recognisable Helmond cases, from parking information to drafting a briefing note. Delivered in September 2025.

The result

The training scored an average rating of 9/10, with learning goals, trainer expertise and materials all at 4.67/5 and guidance at a perfect 5/5. All participants want to continue with a pilot (100% pilot interest). Together we identified 5 concrete use cases, including a briefing-note generator, automatic minute-taking with tasks pushed to the planner, and an HR agent. Two of these were selected as first pilots. The biggest eye-opener: having AI generate prompts itself (self-prompting).