Municipal President

Presiding over a Valais municipality is a militia mandate far more than a job: most presidents hold down another occupation and give up their evenings for it. AI will not govern the municipality: it will pre-process the files so that this inherently limited militia time can focus on the political decision and the responsibility that comes with it.

The job today

Valais counts 122 municipalities as of 1 January 2026, 51 of them with fewer than 1,000 inhabitants. At the head of each stands a president elected by direct suffrage, who leads the municipal executive. In the vast majority of cases the position is part-time, for a modest allowance, alongside a main occupation. It is a militia commitment in the fullest sense: the president reads files in the evening, holds meetings after work, and runs into constituents at the bakery. This chapter therefore speaks of a mandate rather than a profession.

The mandate covers, among other things:

• Preparing and running council meetings: agendas, files, arbitration between departments
• Political review of heavy files: land-use planning, water, energy, schools, natural hazards, infrastructure
• Relations with the canton: consultations, subsidies, compliance, negotiations
• Representation: assemblies, local associations, bourgeoisie, inter-municipal structures
• Running the administration: in tandem with the municipal clerk, staff, budget, organization
• Direct relations with citizens: requests, complaints, informal mediation
• Handling exceptional situations: severe weather, crises, urgent decisions

Files grow more technical every year; militia time does not grow to match. This scissor effect directly threatens the ability of small municipalities to find candidates at all.

What AI is preparing

File preparation, first. A public-inquiry file, a technical report on a water supply system, a two-hundred-page cantonal consultation: AI can produce a faithful summary, with friction points identified and references back to source pages. For a militia representative reading after a day's work, the difference is measured in hours per week. The absolute condition is traceability: a summary whose origin cannot be verified sentence by sentence is worthless as a basis for a public decision.

Then monitoring. Consultation deadlines, calls for projects, federal and cantonal subsidies, legal changes affecting the municipality: these signals arrive scattered today, and small municipalities regularly miss them. Pre-processed monitoring, filtered to the municipality's profile, brings the village president up to the information level of a city with its own staff.

Finally, tracking decisions. What the council decided eight months ago, who was supposed to do what, where the commitment made before the primary assembly stands: a dashboard fed by the minutes gives the executive a continuous institutional memory. Decisions stop getting lost between two terms of office.

None of this decides anything. AI prepares, compares, reminds. The council vote, the arbitration between two legitimate interests, the word given at an assembly: all of that stays exactly where municipal law has placed it. Democratic responsibility cannot be delegated to anyone, and certainly not to a software vendor.

Municipal data: the prerequisite

The president bears the political responsibility for the framework within which the municipality handles data, and that framework is being settled now.

LIPDA. Municipal data (resident registers, building files, social assistance, correspondence) fall under the cantonal data protection act, LIPDA (RS/VS 170.2). Any AI tool touching them must meet requirements of controlled hosting, traceability, and lawful archiving.

No cantonal doctrine yet. Chapter 11 of the essay documents it: there is still no cantonal doctrine on the use of AI in municipal administrations. Until then, each municipality decides alone, with very unequal resources.

The municipal-software moment. Migration to the new municipal management software systems is the opportunity, if there ever was one: it is at the point of migration that hosting, access rights, and data-use clauses with the vendor get negotiated. A municipality that signs without requirements ties itself down for fifteen years.

What matters more for judgment

The political decision. A perfect summary does not say whether to approve something. Weighing tourism development against agricultural zoning, a village school in need of renovation against a road that needs to be made safe, engages values and a vision of the territory; AI can illuminate the consequences of each option, never choose which one is right. The president decides, then defends the choice before the primary assembly.

Arbitrating between interests that all have a face. In a mountain municipality, the backer of a contested project, the objector, and the council member who will present the file have known each other since school. Weighing such situations requires a knowledge of people, families, and local history that no training corpus contains.

A sense of timing. Knowing that a file is ripe, that another must wait until autumn, that the village will only accept this merger of services once it has seen the new waste facility through its first winter: this intelligence about political timing is the silent heart of the mandate. It cannot be bought.

Arbitrating AI deployments. A new role is added: the president becomes the arbiter of what enters the administration, and what does not. Which tool, for which tasks, under what LIPDA framework, under whose responsibility: these decisions bear on the legitimacy of municipal acts and on citizens' trust. Delegating them entirely to the vendor, or even to the municipal clerk alone, would amount to privatizing a political choice.

The public voice. Primary assembly, local press, flood or fire: in the moments that matter, the municipality has a face and a voice. This embodiment is the part of the mandate that digital transformation makes more visible, precisely because everything else becomes tool-driven.

Who keeps the final word?

Composite illustration. A president of a mountain municipality receives, ahead of the council meeting, the generated summary of a land-use file: three objections, classified by the tool as "similar argumentation." He knows the signatories. One of the three is not really aimed at the project: it extends an old irrigation-water dispute between two families. He visits the objector one evening, listens for an hour, and gets the water dispute handled separately with the consortage concerned. Two of the three objections are withdrawn before the meeting. The summary was accurate; the political reading did the rest. (Fictional, composite situation; to be replaced by a real case during the embodiment pass.)

Militia profile 2030

Finding candidates for the presidency will remain difficult. Parties and municipalities searching for them will need to look at three new capabilities.

The first is critical reading of pre-processed files: deciding from generated summaries, demanding traceability back to sources, spotting what is missing, and knowing when a file still deserves to be read in full regardless. A representative who swallows summaries without questioning them shifts real power to whoever configures the tool.

The second is political governance of municipal data: carrying the LIPDA framework at executive level, setting the right requirements for vendors (hosting, reversibility, data use), and treating the municipal-software migration as a political decision rather than a mere IT project. No municipality expects its president to be an IT specialist; all of them need him or her to know what they are signing.

The third is leading change within a militia structure: bringing along a council with diverse sensibilities, a sometimes anxious secretariat, a population that wants to understand. Explaining at the primary assembly what the municipality automates, what it will never let a machine handle, and who answers in case of error, before rumor takes over in its own way.

Territorial roots

The Valais militia system rests on a fragile equation: increasingly technical files entrusted to representatives whose available time does not grow. In the 51 municipalities with fewer than 1,000 inhabitants, this equation is already discouraging candidacies. If pre-processing files makes the mandate sustainable at ten hours a week instead of twenty, it very concretely defends the municipal militia — that is, the Valais form of grassroots democracy.

The president is not alone in the institutional landscape. The bourgeoisies, which own forests, alpages, and heritage assets, have managed commons for centuries and share very concrete files with the municipality (water, forest, buildings); chapter 4 of the essay sees in these community institutions a governance matrix that digital transformation would do well to draw on, and they are natural partners for upcoming deployments. The Association des communes valaisannes, for its part, is the level at which to pool what no municipality of 800 inhabitants can do alone.

This chapter, finally, should be read alongside the first in the series, devoted to the municipal clerk. The president-clerk tandem is the basic unit of Valais municipal administration: the clerk operates the tools day to day, the president answers politically for their use. The two projects (mapping the secretariat's tasks, the executive's governance framework) benefit from being carried out together, within the same term of office.

What the decision-maker must do now

For a sitting president

Inventory by the end of 2026 what has already entered the municipal administration: which tools the secretariat uses, with what data, under what framework. In many municipalities the honest answer is "we don't know exactly." Then set a minimal written framework (approved tools, data barred from export, human validation of official acts), drawing on the task mapping proposed in the Municipal Clerk chapter.

For the Association des communes valaisannes

Pool what 122 executives should not have to reinvent 122 times: a template framework for municipal AI governance, standard contract clauses for the software migration (hosting, reversibility, vendor use of data), and a forum for presidents to exchange experience. Negotiating leverage with vendors is collective, or it does not exist.

For a cantonal official (Service des communes)

Fill the gap in cantonal doctrine noted in chapter 11: publish LIPDA guidelines applicable to municipal AI tools, support the software migration with explicit governance requirements, and offer militia representatives short training, designed for people who sit in the evening after a day's work. Two well-built evenings change the quality of a term's equipment decisions.

---

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