Why Swiss SMEs are adopting automation and AI

SMEs make up the bulk of the Swiss economy: 99.7% of businesses and two-thirds of jobs, according to the Federal Statistical Office. How they relate to automation and AI therefore says a great deal about the country's transformation. Here I rely on the most recent verifiable figures, and on what I see in the SMEs I work with in French-speaking Switzerland.

Adoption is accelerating

AXA's annual study, carried out by the Sotomo institute among 300 SMEs in French- and German-speaking Switzerland, shows a clear rise in a single year:

	2024	2025
SMEs that have built AI into their processes	22%	34%
SMEs that use it occasionally	33%	37%
SMEs that do not use it	45%	29%
SMEs that see AI as an opportunity	35%	45%
SMEs using AI that say they save time	46%	57%

The shift from trying it out to regular use is the real signal. Two years ago, AI was mostly a matter of curiosity. Today, a third of SMEs have built it into the way they work.

Why now?

High labour costs

In Switzerland, an hour of administrative work is expensive. Every copying, chasing or filing task that is removed frees up time with immediate value. It is an argument I hear in almost every first meeting.

Tools that are finally accessible

Platforms such as Make, n8n or Zapier make it possible to automate without heavy development, with subscriptions of a few dozen francs a month. Language models now read most administrative documents correctly. I compare these tools in my article Make vs n8n vs Zapier.

Difficulty recruiting

When an administrative position stays vacant for months, automating part of the workload becomes a pragmatic answer, not an ideological choice.

What SMEs actually automate

According to the AXA study, SMEs that use AI mainly use it for:

• translation (52%), an obvious need in a multilingual country;
• correspondence (47%): emails, letters, standard replies;
• process optimisation (more than a third);
• data analysis (32%).

In the projects I run, the most frequent requests are very concrete: invoice reminders, entering accounting documents, syncing the CRM with invoicing, monthly reporting. They are rarely spectacular projects, but they free up time every week. Prospecting can be automated too: targeting new businesses by sector relies on their activity code, and I explain how to look it up in how to find a Swiss company's NOGA code.

The case of fiduciary firms

Fiduciary firms are among the most affected, because much of their work consists of receiving, reading, filing and entering documents. Automatic extraction of invoice data profoundly changes the way they work. I have a dedicated page on it: AI automation for fiduciary firms (in French).

The obstacles that remain

Data protection

This is the clearest weak spot. Only one in three SMEs using AI has a clear data protection policy in place, and the figure drops to 23% among the smallest businesses.

Yet the revised Swiss Federal Act on Data Protection (FADP), in force since 1 September 2023, applies to processing by AI just as it does to any other processing of personal data. Knowing which data goes to which services, and in which countries, is not optional.

Key point: a tool self-hosted in Switzerland, such as n8n, helps you stay in control. But every AI service the workflow calls also receives data: it must be chosen and configured accordingly.

Lack of time and skills

An SME generally has nobody whose job is automation. The owner sees the potential, but has neither the time to map processes nor the time to learn a new tool. That is often where projects stop before they have started.

Fear of a heavy project

Many business owners picture automation as a big IT project. In reality, a useful first workflow can be built in a few days. Failures more often come from projects that are too ambitious than from ones that are too modest: in its report on RPA, EY estimated that 30 to 50% of initial projects failed, largely because they targeted the wrong processes.

What sets successful SMEs apart

In my experience, four practices make the difference:

1. Start small: a frequent, repetitive process with clear rules, whose benefit is easy to measure.
2. Measure before and after: time spent, volume, errors. Without a baseline, there is no way to prove the gain.
3. Keep a human in the loop on sensitive decisions: disputes, unusual entries, delicate replies.
4. Train the team so it understands and can evolve what has been built. That is the purpose of my n8n and Make training in Geneva.

My view

The figures confirm what I see on the ground: automation and AI have moved beyond the curiosity phase. A third of SMEs already use them every day, and those that get started see a real time saving.

The risk today is no longer missing a revolution. It is automating quickly and badly: without a data protection policy, without documentation, with workflows nobody knows how to take over. The right approach remains modest and methodical.

To understand in practical terms what AI can automate and where to start, read my guide to business automation with AI. And to move on to a first project, you don't necessarily need an automation agency: I explain how I support process automation for SMEs in Geneva.

Useful sources

• ICTjournal, Les PME suisses prennent goût à l'intelligence artificielle (AXA and Sotomo study, October 2025, in French)
• Federal Statistical Office, Small and medium-sized enterprises (in French)
• SME Portal of the Swiss Confederation, New data protection act (nLPD) (in French)
• EY, Get ready for robots (2016)