AI for the public sector: modernize without exposing citizen data
Your team may already be using ChatGPT to draft correspondence. With citizen data involved, that is a direct compliance risk, and there is a better way.
Local government offices face a double constraint: growing pressure to modernize, alongside tight budgets and strict data protection rules (nLPD in Switzerland).
Unmanaged use of public AI by staff is especially sensitive here: every piece of correspondence contains personal data.
- Legal exposure on citizen data
- Pressure to modernize with no extra budget
- No written AI policy at most agencies
See for yourself
A prompt structured in 5 blocks: Instruction · Role & Audience · Data · Constraints · Format, with compliant anonymization built in from the start.
On-site, we build this kind of prompt with you for the public sector, each with its estimated ROI, its safeguards, and Swiss-hosted tools. No citizen data in the examples.
Réponse à un courrier citoyen (nLPD)
Anonymization built in: no citizen data ever touches the AI.
See it in Chaptal →PV de séance → communication publique
Meeting notes → official minutes and a public announcement, closed-session content excluded.
See it in Chaptal →Politique IA communale (cadre d’usage)
The written framework your leadership can approve in session.
See it in Chaptal →Training tailored to the public sector · quoted on request
Government use cases, anonymization rules, sovereign tools.
A written AI policy
A clear framework for your staff, approved by leadership.
Ongoing monitoring
If needed, continuous Shadow AI monitoring (quoted on request).
Swiss-hosted tools preferred · compliance at the center of the approach · real experience with public-sector constraints.
Foire aux questions
How do you guarantee compliance on citizen data?
We favor Swiss tools and Swiss hosting, and we systematically work with anonymized data in our examples and workshops.
Do we need to run a public procurement process to engage you?
It depends on the amount and your agency’s own rules. We adapt our proposal (quote, scope of work) to whatever process applies.
What budget should a small agency plan for?
Training is quoted on request, sized to your agency. We always scope the engagement before pricing it.
Do you work with Swiss-hosted tools?
Yes, that is a priority for the public sector. We systematically favor sovereign, compliant solutions.
How long before we see concrete results?
The first use cases (correspondence, meeting minutes) are operational right after training. A written AI policy typically follows within a few weeks.