AI for municipalities and government: an assistant, not a decision-maker
Municipalities are under pressure: fewer staff, more legislation and regulation, residents who expect an immediate answer, and a regulator that tests every automated decision against the General Administrative Law Act (Awb) and the AI Act. We build AI applications that support civil servants — chatbots for resident enquiries, NLP for Woo requests, automation of Wmo intakes — without the algorithm making the final decision. Human-in-the-loop is not an extra; it is the starting point.
Discuss your case View applicationsWhy AI for government is a different conversation
A municipality is not a web shop. Decisions affect people's lives — benefits, care assessments, permits, enforcement. That changes how AI may be used, not whether it can be used.
The childcare benefits scandal has permanently damaged Dutch trust in automated government decisions. For digitalisation managers and CIOs at municipalities, that means every AI application is tested for explainability, for bias, for compliance with the General Administrative Law Act (Awb, art. 4:8 duty to give reasons), for the AI Act high-risk classification, for GDPR, for the information security plan (IBP) and for the Baseline Informatiebeveiliging Overheid (BIO, the government information security baseline). That may sound daunting, but it is good news: organisations that deploy AI responsibly achieve measurable gains without the risks of a Syri or childcare benefits scenario.
In practice, many municipal processes are not about decisions, but about providing information, filling in forms, summarising case files and looking up documents. A chatbot that explains at 11 p.m. how to renew a passport is not making a decision. An NLP pipeline that searches a Woo request and proposes draft redactions is not making a decision — a civil servant still presses publish. A council document summariser that reduces 200 pages to ten bullet points for councillors is not making a decision. That is exactly where the first, safe layer of AI impact lies: support, not replacement.
We build that layer. No experimental algorithms deciding in an unreadable black box who does or does not receive a Wmo assessment — instead, concrete tooling that halves the turnaround time of a Woo request, relieves the call centre and helps councillors go into meetings better prepared. Our municipal software practice and government apps portfolio form the foundation; AI is a component, not a standalone product.
Where AI works in the municipal organisation
Six areas of application where impact, feasibility and legal scope are all in order. Each scenario is built around support, not replacement.
Resident chatbot for 24/7 enquiries
An AI chatbot on the municipal website or via WhatsApp answers questions about passports, driving licences, moving house, waste, parking and events. The model draws on your own knowledge base and refers people to the right desk or form. Multilingual support (Dutch, English, Turkish, Arabic, Polish) makes information accessible to the whole population. When in doubt, it escalates directly to a civil servant.
Speeding up Woo/Wob requests
The Open Government Act (Woo) has multiplied the number of requests. NLP models search thousands of emails, memos and attachments for relevance to the request, propose redactions based on the Woo exemption grounds (article 5.1) and keep track of which documents have already been reviewed. A civil servant decides whether to adopt each proposed redaction — the tool speeds things up, it does not decide.
Wmo/Youth Act intake automation
The first contact conversation for a WMO (Social Support Act) application generates a file recording the services involved, living situation and support needs. AI speech-to-text plus structuring fills in a standardised intake form in line with your local bylaw. The case officer corrects and validates it. The decision on the assessment remains with a human — Section 4:8 of the General Administrative Law Act (Awb) requires a reasoned decision that the applicant can challenge.
Permit AI for local bylaws and the Environment and Planning Act
The Environment and Planning Act (Omgevingswet) requires integrated assessment. AI models check applications against the local environmental plan, flag conflicts and suggest conditions based on earlier comparable decisions. For simple applications (dormer window, driveway) this speeds up processing; for complex applications the permit officer gets a well-founded starting document instead of an empty template.
Council document summarisation and searchability
Councillors receive hundreds of pages of documents every week. An LLM pipeline generates a fixed-length summary for each document, recognises motions and amendments, and makes the entire archive searchable through natural-language queries. For council clerks this saves manual reading; for councillors it improves the quality of deliberation.
Anomaly detection without bias risk
Where Syri failed, anomaly detection can work: not to score citizens on suspicious behaviour, but to monitor internal processes. Unusual invoice patterns, irregularities in grant applications from the same organisation, duplicate claims from suppliers. The model looks at transactions, not at citizens — a fundamentally different risk profile under the AI Act.
How Appfront builds AI for public organisations
We don't start with a model, but with a process analysis. Which task takes civil servants the most time? Where do citizen complaints about unavailability arise? Which workflow is buckling under the Open Government Act (Woo) or the Environment and Planning Act? Only once that is clear do we assess whether AI makes sense — and if so, whether it becomes a generic LLM application, a specifically trained model, or no AI at all but smart automation.
Our projects land within your existing infrastructure. For authentication we integrate with DigiD and eHerkenning, for identity data with the Personal Records Database (BRP, formerly GBA) and the Register of Non-Residents (RNI). Suwinet integrations are only used where the Suwi chain permits and with explicit authorisation. We follow the VNG's Common Ground principles: data at the source, reuse through standard APIs, no shadow administrations.
Pseudonymisation of BSN (citizen service) numbers and other personal data takes place before data reaches the AI model. Training data is anonymised or generated synthetically. Models run by default within Dutch or EU cloud regions, or on-premise within the municipal data centre. We follow the BIO and the IBP — not as a checklist, but as an architectural starting point.
Four phases: from problem definition to production AI
An AI project for a municipality requires more preparation than a commercial project. Not because of the technology, but because of governance. Our phasing reflects that.
Problem exploration and DPIA
We discuss the use case with the process owner, the DPO (data protection officer), the CISO and legal affairs. A DPIA (Data Protection Impact Assessment) determines whether the application is high-risk under the AI Act and which additional safeguards are needed.
Proof of concept with dummy data
The first validation takes place on pseudonymised or synthetic data. We show that the model works without putting production data at risk. The PoC result is tangible: a working demo that process owners can test.
Pilot with human in the loop
A focused production pilot in which the model makes suggestions and civil servants carry out the final action. We measure turnaround time, quality, user acceptance and false positives. The DPIA is updated based on the pilot.
Production and governance
Rollout with monitoring, audit logging and periodic model retraining. We document to AI Act requirements: datasets, evaluations, limitations and escalation procedures. Municipalities can demonstrate what the model does and does not do.
Test your idea first: a working prototype in 1 day
With OneDayBuild, we make your idea tangible in a single day for €1,150, so you know whether further development is worth the investment. Decide to go ahead with the full build? Then we deduct the cost in full.
View OneDayBuild →Technology and standards we use
The choices for public organisations differ from a typical enterprise stack. Open source is preferred wherever possible; vendor lock-in on a US cloud LLM is unacceptable for many municipalities because of the Schrems II implications and the requirement for data residency within the EU. We therefore work by default with models that can be run within EU regions, or deployed fully on-premises.
For chatbots and document AI, we deploy Llama, Mistral and Phi models on hosted GPUs within the Netherlands or the EU, using retrieval-augmented generation (RAG) so the model answers from your own knowledge base. For classic ML applications (anomaly detection, classification of applications) we use scikit-learn and XGBoost: explainable, auditable, not a black box. We build frontend components with React and the NL Design System component library, so the application fits the NLDS standard and the municipality's house style.
AI Act, GDPR and Awb: what this means for you
Compliance is not an afterthought in a municipal AI project. It is the starting point. Four frameworks we factor into every architecture decision.
AI Act high-risk classification
Applications that affect access to public services, social security or essential services fall under Annex III and count as high-risk. This brings requirements around risk management, data quality, technical documentation, transparency, human oversight and accuracy. We design the system so that documentation and oversight are built in, not reconstructed after the fact.
Awb Art. 4:8 and the duty to give reasons
A government decision must be reasoned in a way that citizens can challenge. "The algorithm said no" is not enough. Our AI applications produce a written rationale that can serve as the statement of reasons, or deliver a substantiated proposal that the civil servant confirms in their own words. The legal decision remains with a human.
GDPR, BIO and IBP
Personal data is pseudonymised before it is fed into the model wherever possible. Logging is set up in line with the BIO: no more than necessary, no less than required. The municipality's IBP is the baseline; we integrate within the existing security architecture, not as a separate silo.
Bias and non-discrimination
Models trained on historical data can reproduce historical inequality. We test for disparate impact across nationality, neighbourhood, age and gender. Models that use sensitive attributes as a proxy in decision-making are not deployed, full stop. When in doubt: a rule-based fallback with a human decision-maker.
Example projects from municipal practice
Three scenarios that illustrate what a municipal AI application looks like in practice, from problem to DPIA to production.
A Woo team that delivers redactions 40% faster
A medium-sized municipality receives structurally more Woo requests (Open Government Act) than it used to under the Wob. The legal team has limited capacity. An NLP pipeline scans the requested documents, identifies personal data, commercially confidential information and draft decision-making documents (Woo art. 5.1 and 5.2), and produces an initial redaction proposal. The lawyer reviews and adjusts it, with no more manual searching for every BSN (citizen service number) or email address.
BSN-free anonymisation for research
A municipality wants to analyse patterns in WMO (Social Support Act) applications to underpin policy. BSNs and name-and-address details are converted through a pseudonymisation pipeline before the data reaches the analytics platform. The analysis team works with keys that cannot be traced back to individual citizens. The raw data stays within the source application, in line with Common Ground.
A multilingual citizen chatbot for the 14+ service desk
A municipality with a diverse population receives enquiries by phone, email, WhatsApp and website. A chatbot draws on the indexed content of the municipal website, product catalogue and FAQs, answers in five languages and escalates to the right department for complex cases. The backend logs every interaction for audit and quality monitoring; no conversation is used for training without explicit review.
A council information system that summarises issues
Ahead of the council meeting, members receive an agenda with dozens of documents. An AI layer generates a summary for each agenda item, covering key points, previous decisions, financial impact and relevant motions. Councillors click through to the full document when needed. For the clerk's office, this means fewer questions beforehand and better quality of debate.
Frequently asked questions about AI in government
Want to use AI in your municipality or government organisation?
Discuss your case with a team that is at home in AI engineering as well as the Awb, the AI Act and BIO. Together we map out where AI can contribute safely and with measurable impact, and where it is better not to use it.
Schedule a call