Municipal Clerk

The Valais municipal clerk is often an entire administration in one person. AI will not replace them: it will radically change what their judgment needs to focus on.

The job today

Valais counted 122 communes as of January 1, 2026¹. Of these, 51 (41.8%) have fewer than 1,000 residents, with a median of 1,305 residents². In these communes, the municipal clerk is often the administrative backbone: sole point of contact, institutional memory, drafter of documents, first-line accountant, and at times de facto legal adviser.

Their day-to-day duties span a wide range:

• Preparing and drafting the minutes of the communal council, often for several meetings a month
• Handling citizen requests: building permits, residency declarations, certificates, complaints
• Monitoring legal compliance: adapting communal regulations to cantonal and federal developments (spatial planning law, cantonal building law, water protection law, etc.)
• Drafting official correspondence: replies to the Canton of Valais, the Confederation, government agencies, and residents
• Managing the residents' register: arrivals, departures, naturalizations, changes of status
• Coordinating commissions: agendas, meeting notices, archiving decisions
• First-level accounting: entering transactions, tracking the budget, preparing year-end closings with the fiduciary accountant

In the smallest communes, all of these tasks rest on a single, often part-time, person.

What AI prepares — and what the clerk validates

Generative AI will mechanize a substantial share of the textual groundwork and formatting. The wording matters here: AI prepares, pre-processes, classifies, and suggests. The commune retains human validation over every legally binding document, and this distinction is a matter of law far more than of drafting style.

Assisted drafting (minutes and routine correspondence). Minutes, produced from an audio transcript or meeting notes, will be generated in seconds, in the proper administrative register. Standardized reply letters will follow the same path: acknowledgments of receipt, requests for additional documents, decision notices. In both cases, the clerk reviews, validates, and signs. Responsibility remains attached to a named individual; that is what distinguishes a public act from a mere convenience document.

Regulatory monitoring. Changes in cantonal and federal law will be flagged automatically, along with the relevant communal regulations and a proposed revised text. AI produces the signal and the draft; the clerk decides on priority according to the local political calendar and ongoing matters.

Pre-processed handling of standard requests. Certificates, address confirmations, routine forms: these workflows will be prepared without manual re-entry, but under the oversight of an identified accountable agent. No legally binding act leaves the office without explicit human validation. For sensitive files (complex building permits, social assistance, land disputes), AI prepares the file; the clerk decides.

Archiving and documentary memory. This may be where the gains prove most lasting. Classification, full-text search across decades of archives, generating summaries from past deliberations: AI becomes an assistant to institutional memory that the clerk can query, provided the data is hosted in a controlled environment (see box below).

Governance and data: the non-negotiable prerequisite

This transformation requires a strict framework, without which no deployment can be conducted responsibly within a Swiss public administration.

Hosting and data. Data processed by an AI tool (minutes, permit files, resident data, correspondence) is subject to LIPDA (the law on public information, data protection, and archiving)³. Absent an explicit legal basis, any transit through foreign servers is excluded. Swiss or cantonal hosting is the baseline requirement.

Traceability. Every document generated by AI must be traceable: who requested it, on what basis, who validated it, and when. Decision logs form part of the commune's legal archive.

Archiving. Cantonal archiving rules apply to AI-produced acts just as they do to manually produced ones. The evidentiary value of an act depends on its documentary integrity, not on how it was produced.

For a commune, the stakes go beyond production speed: the challenge is to produce documents without undermining the legitimacy of the public act.

What rises in importance for judgment

Precisely because AI absorbs the repetitive work, the municipal clerk reverts to what they had never really stopped being: a guarantor of institutional legitimacy.

Contextual interpretation of the law. A statutory provision always applies within a particular political, human, and territorial context. AI can cite the law; it cannot weigh the interests of a neighborhood against those of a project developer who knows everyone involved.

The commune's relational memory. Knowing that a certain family has contested this road alignment for two decades, that the council member in charge of finances prefers cautious wording, that the district prefect appreciates a certain formality: this knowledge cannot be captured in a language model.

Managing sensitive situations. Neighbor disputes, social assistance requests, inheritance disputes over agricultural land: AI prepares the file, the clerk decides or points the way forward.

The legitimacy of the signature. Under Swiss law, certain acts require an identified, accountable natural person. That responsibility cannot be delegated to an algorithm.

Informal mediation. In a commune of 400 residents, half of all problems are resolved before they ever reach the agenda. That is invisible work an AI will never see.

Who keeps the final say?

Composite illustration. A commune in the Haut-Plateau region must notify a project developer — who also happens to be the area's largest employer — of a refused building permit. AI drafts a legally flawless letter. The municipal clerk chooses instead to call the applicant first, explain the two sticking points, and suggest a revision to the file that could lead to an agreement. The refusal letter is never sent. The institutional relationship is preserved. (A fictional situation, composited from cases encountered while advising communes in French-speaking Switzerland.)

Job description, 2030

Job postings for this role will need to include three new competencies. None of them appear in any current initial training program for municipal clerks.

The first is assisted compliance validation: the ability to oversee an automated regulatory monitoring workflow, assess the relevance of the alerts it generates, and prioritize updates according to local political and legal context. Augmented legal judgment, in which programming plays no part whatsoever.

The second is governance of AI-generated acts: reviewing minutes, a letter, or a regulation produced by AI with the eye of an institutional guarantor, catching lapses in register, misinterpretations, and missing context. Ensuring that data-processing conditions comply with LIPDA and cantonal archiving requirements. This validation competency is distinct from drafting competency; it simply did not exist before.

The third is older in principle but renewed in form: digital mediation for residents. Supporting residents — older people, non-French speakers, vulnerable groups — in their interactions with the commune's digitized services. The municipal clerk also becomes the translator between the AI-assisted administration and the resident who does not know its codes.

Territorial anchoring

This transformed role is a direct opportunity for Valais communes to maintain a quality local secretariat in the face of a growing workload, without having to turn to mergers or outsourcing.

Today, pressure toward pooling secretariats is real, driven by arguments about cost and workload. If AI pre-processes a significant share of repetitive tasks, a small commune can preserve a local administrative anchor that guarantees institutional proximity, continuity of memory, and a direct relationship with residents.

Until now, the question was whether a commune was too small to have its own clerk. The question now is whether that clerk is equipped to run the tools needed to guarantee the same quality of service as a town of 5,000 residents — while remaining the indispensable human interface for the territory.

That is precisely the thesis of the Bisse Cognitif applied to local administration: far from replacing human presence on the ground, AI makes it economically viable where it was under threat. What remains to be built is the framework — training, data governance, tools suited to mountain communes — without which this opportunity will stay theoretical.

What decision-makers must do now

For an elected communal official (president, vice-president)

Before 2027, commission a mapping of the secretariat's tasks: which take up the most time, which are purely repetitive, and which require irreplaceable human judgment. This mapping serves as the basis for revaluing the position, justifying investment in training, and defining data governance requirements before any deployment; it is in no way an audit aimed at cutting the position.

For a cantonal official (Communes Service, Department of the Interior)

Lead a continuing-education program specific to the augmented communal secretariat, covering both mastery of the tools and the LIPDA/archiving framework. The cantonal centers of expertise (CFPV, HES-SO Valais) are the right vehicles for this. The central issue is trust in tools that produce legally binding acts grounded in cantonal public law.

For an HR manager at an intercommunal association

Review job profiles and salary scales. A municipal clerk who masters AI tools for administrative productivity and ensures the governance of communal data is no longer doing the same job as their predecessor from 2020. Salary revaluation is the logical counterpart to an expanded scope of responsibility.

---

¹ Canton of Valais, Key Figures 2025. ² Report on Communal Finances 2024, Canton of Valais. ³ Law on Public Information, Data Protection, and Archiving (LIPDA), RS/VS 170.2.

Jérôme Deshaie is the founder of MCVA Consulting SA, an agency specializing in the AI transformation of organizations in Valais, and the author of Bisse Cognitif.