TLDR: Swiss small and medium-sized enterprises adopted artificial intelligence at speed, yet most adopters lack rules on what staff may enter, making usage rules a human resources task and a hiring criterion.
Artificial intelligence integration in Swiss small and medium-sized enterprises rose from 22% to 34% in one year
The AXA labour market study 2025 on small and medium-sized enterprises (SMEs), carried out by the research institute Sotomo among 300 SMEs in German- and French-speaking Switzerland between 3 and 10 March 2025, found that the share of SMEs that consciously integrate artificial intelligence (AI) into their work processes rose from 22% to 34% in a year. Over the same period, the share of firms with zero AI use fell from 45% to 29%. The State Secretariat for Economic Affairs (SECO) relayed the findings on its SME portal on 5 November 2025 under the title “KI in Schweizer KMU auf dem Vormarsch”, AI on the advance in Swiss SMEs.
Beyond translation (52%) and correspondence (47%), the share of SMEs using AI to optimise work steps climbed from 23% to 34% and the share using it for data analysis rose from 22% to 32%. According to SECO’s summary, targeted advertising reached 24% and customer relationship management (CRM) 20%. Among AI users, 57% report time savings, up from 46%, and 60% see AI as an opportunity against 8% who see a threat. Data analysis and process automation are the uses that touch customer files, pricing and personnel records, which is where governance starts to matter.
The 2026 edition, published by AXA on 15 September 2026 from a survey of 336 SMEs between 11 and 18 March 2026, shows the curve flattening at a high level: 35% of SMEs now integrate AI deliberately, 39% are testing it and 26% are yet to start. As of September 2026, three in four Swiss SMEs therefore work with AI in some form, and the question has moved from adoption to control.
Only one in three AI-using SMEs has set rules on what staff may type into AI tools
According to the AXA study, only a third of SMEs that use AI have set clear rules on which data employees may enter into AI tools. Among small firms with five to nine employees, only 23% have introduced such rules. Kathrin Braunwarth, head of data, technology and innovation at AXA Switzerland, called data protection “central for the safe and responsible use of AI”. In practice, two in three AI-using SMEs leave the decision about what goes into a chatbot to each employee.
| Indicator (AXA SME study) | 2024 | 2025 |
|---|---|---|
| SMEs integrating AI into processes | 22% | 34% |
| AI used to optimise work steps | 23% | 34% |
| AI used for data analysis | 22% | 32% |
| Clear data rules, among SMEs using AI | n/a | one in three (SECO: 34%) |
| Same rules, small firms with five to nine employees | n/a | 23% |
| SMEs where AI created new jobs | n/a | 10% |
| SMEs that reduced staff thanks to AI | n/a | 2% |
The gap carries legal weight under existing Swiss law. The Federal Council decided on 12 February 2025 to ratify the Council of Europe Convention on Artificial Intelligence and to pursue sector-specific amendments, for example in healthcare and transport, as its route to regulation. In the meantime, the Federal Data Protection and Information Commissioner (FDPIC) states that the Federal Act on Data Protection (FADP) is directly applicable to AI and that users must make the purpose, functionality and data sources of AI-based processing transparent, including whether data entered into a chatbot is used to train it.
For staff data the bar is higher still. Article 328b of the Swiss Code of Obligations (CO) lets an employer process employee data only to the extent it concerns suitability for the job or is necessary for the employment contract. An office manager who pastes a salary table into a public chatbot to draft a pay letter, or a founder who uploads candidate curricula vitae to compare them, moves personal data to a third-party processor with an unstated purpose and an empty audit trail. In Edward Galle’s view, a one-page rule closes most of that exposure, and two in three AI-using SMEs still need to write it.
Pharma and medical technology suppliers face a stricter version of the same rule
Switzerland’s life sciences industry runs on small companies. SECO’s SME portal reported in September 2024 that the medical technology (MedTech) sector employs 71,700 people across about 1,400 active companies, with 95% of the industry made up of SMEs. Around them sits a supplier layer of contract manufacturers, testing laboratories, regulatory consultancies and medical writing agencies, many of them small firms, that handle data belonging to pharmaceutical clients.
For these firms, the data a member of staff might paste into a chatbot often includes records governed by good practice (GxP) rules: batch records, deviation reports, stability data, clinical study documents or complaint files. Those records sit under a client’s quality agreement and confidentiality clauses, and their integrity is an inspection topic. The European Medicines Agency and the Pharmaceutical Inspection Co-operation Scheme (PIC/S) opened a joint consultation on 7 July 2025 on a revised Annex 11 for computerised systems and a new Annex 22 on artificial intelligence, which sets requirements for how AI models used in good manufacturing practice (GMP) are selected, trained and validated, and strengthens controls on data integrity, audit trails and electronic signatures. For GMP environments, AI that touches regulated records belongs inside validated, controlled systems.
In Edward Galle’s view, a supplier SME’s AI rule needs one extra line that a generic SME policy omits: a named list of client and GxP data classes that stay inside validated systems and approved enterprise tools, with the quality manager as the person who approves any exception. That line protects the client relationship as much as it protects data. When a pharmaceutical client audits a supplier, a question about staff use of AI extends the usual review of computerised systems, and an SME that can hand over a one-page rule, a list of approved tools and a training record answers it in minutes.
Supplier SMEs in Basel, Vaud and Ticino need people who combine quality assurance with digital literacy: a quality specialist who understands computerised system validation and can assess an AI tool, or a digital project lead who can write a user requirement in GMP language. In Edward Galle’s view, these hybrid profiles are scarce in small firms, where one quality manager often covers several functions, and they sit at the centre of its search work with regulated employers.
Artificial intelligence is changing the skills SMEs hire for faster than it changes headcount
The employment effect so far is small and positive. Only 2% of SMEs say AI time savings allowed them to reduce staff, while 10% say AI led to new jobs. The study authors conclude that the shift affects the skills employers look for more than the number of positions. A third of SMEs with AI experience told AXA that AI tools have changed their requirement profile for employees, and Sotomo director Michael Hermann noted that many companies now look specifically for skilled staff who bring technological competence and keep learning. The 2026 edition keeps the same direction: only 12% of SMEs with AI experience expect to be able to manage with fewer staff in future.
The wider labour market confirms where the demand concentrates. The Adecco Group’s Swiss skills shortage index for 2025, compiled with the University of Zurich’s Stellenmarkt-Monitor Schweiz and released on 27 November 2025, recorded the sharpest fall in open positions in occupations with high automation potential, including commercial, administrative and information and communication technology (ICT) roles, and for the first time in years a surplus of ICT specialists. Netzwoche’s report on the index notes continued demand in finance and mathematical occupations despite high AI exposure, which Adecco linked to gradual AI adoption in SMEs and the need for data analysis and system monitoring skills. Generic information technology (IT) profiles are easier to find; people who can make AI produce reliable numbers inside a small business are harder to find.
Four digital and AI skill sets follow directly from how SMEs use the tools
Each use case in the AXA data maps to a capability that an SME either hires, trains or buys in. The mapping matters because a firm with 30 employees rarely opens four specialist roles; it writes one hybrid job description, and the order of the skills in that description decides who applies. Reporting work in the table below relies on business intelligence (BI) tools, the dashboards and data models that turn raw figures into management reports.
| SME use (AXA 2025) | Skill the role needs |
|---|---|
| Optimising work steps (34%) | Process mapping, low-code automation, prompt and workflow design |
| Data analysis (32%) | Data literacy, spreadsheet and BI modelling, checking AI outputs |
| Targeted advertising (24%) | Marketing automation, content governance, brand-tech tooling |
| Customer relationship management (20%) | CRM administration, data hygiene, revenue operations |
| Work for pharma and MedTech clients | GxP data classes, computerised system validation, quality agreements |
| All uses | FADP basics, data classification, AI usage policy ownership |
The first profile is the process and automation generalist, often an operations or finance employee who documents a workflow before automating it. The second is the data-literate analyst who builds BI reports and checks every AI-generated figure against source data. The third is the marketing and CRM operator, the SME version of the brand-tech and revenue operations (RevOps) specialists that larger companies hire, who keeps customer data clean enough for AI to use. The fourth is a responsibility more than a role: the AI policy owner, who maintains the list of approved tools and the data rules, and who in a regulated supplier works hand in hand with the quality manager. Edward Galle’s analysis of brand-tech talent and of the digital skills commercial teams need describes the larger-company versions of these profiles.
The policy gap is a human resources task, and it belongs in every job description
An AI usage rule works best inside the employment relationship, next to the staff regulations, where every employee meets it on the first day. That places it with human resources (HR), or with whoever plays the HR role in an SME. A workable one-page policy names the approved tools and accounts, the data classes that stay inside approved systems (employee and candidate data, client confidential information, health data, GxP records, unpublished financials), the uses that require a human decision, and the person to ask. The FADP adds two duties that SMEs often overlook: transparency towards the people whose data is processed, and, under Article 21, the right of a person subject to a decision based exclusively on automated processing to request review by a natural person. Any SME that screens applicants with AI therefore needs a human in the loop and a clear route to request that review. Larger employers solve the same problem inside their HR suites, a sequence Edward Galle set out in Workday and SAP SuccessFactors Now Ship HR Agents, and Employee Self-Service Goes First.
| Rule element | Example wording | Legal anchor |
|---|---|---|
| Approved tools | “AI work runs on the tools listed in the annex, through company accounts covered by a data processing agreement.” | FADP, directly applicable to AI (FDPIC) |
| Protected data | “Employee, candidate, health and client confidential data stay inside approved systems.” | CO Art. 328b for staff data |
| Transparency | “Staff tell customers and candidates when AI processes their data, and for what purpose.” | FDPIC guidance on purpose, functionality and data sources |
| Human decision | “A named employee takes and signs every decision about a person, such as a shortlist.” | FADP Art. 21, review by a natural person |
| GxP records (suppliers) | “Batch, deviation and study records stay in validated systems; the quality manager approves any exception.” | PIC/S and European Medicines Agency draft Annex 11 and Annex 22 |
| Owner | “The AI policy owner keeps the tool list current and answers staff questions.” | Staff regulations and onboarding file |
Job descriptions that state the tools in use and the data rules attract candidates who already work that way, and onboarding that includes a short AI briefing closes the gap for everyone else. Performance conversations can then assess AI use against the rule, a theme Edward Galle covered in its piece on AI use as a promotion criterion.
SME owners who write the rule this quarter hire and onboard with a clear standard in 2027
SME owners and managing directors can write the one-page rule this quarter, list the AI tools staff already use, and move work involving personal data onto business accounts with a data processing agreement. When the next vacancy opens, the job description can name the AI and data skills from Exhibit 2 in the order the business actually needs them, with data literacy and process thinking ahead of tool names that change every year.
Quality managers in pharma and MedTech suppliers can add the GxP data classes to the rule, map which client agreements mention AI or automated processing, and keep a short training record for every employee who uses an approved tool. That record answers the client auditor’s question and gives new hires a clear standard from their first week.
HR managers and office managers in growing SMEs can treat the policy as part of the employment file: include it in onboarding, refer to it in the staff regulations, and record which hiring steps use AI so that a candidate’s request for human review has a clear route. Kandidatinnen und Kandidaten targeting Swiss SMEs gain an edge by showing evidence of governed AI use: an automated workflow with its before-and-after time, an analysis whose figures they verified, or a policy they wrote in a previous role.
Edward Galle places digital, brand-tech and RevOps specialists with Swiss employers in regulated industries and supports HR teams through its HR consulting and workforce systems (HR IS) services. Professionals with these hybrid skills can submit a curriculum vitae oder browse current openings.
This article has been reviewed by Haider Alleg, Managing Partner at Kainjoo SA.
References
- AXA Switzerland. AXA KMU-Arbeitsmarktstudie: Künstliche Intelligenz erobert Schweizer KMU. 8 October 2025. https://www.axa.ch/de/ueber-axa/medien/medienmitteilungen/aktuelle-medienmitteilungen/2025/20251008-kmu-arbeitsmarktstudie-2025-ki.html
- SECO, KMU-Portal. KI in Schweizer KMU auf dem Vormarsch. 5 November 2025. https://www.kmu.admin.ch/de/ki-in-schweizer-kmu-auf-dem-vormarsch
- AXA Switzerland. Künstliche Intelligenz: Schweizer KMU erwarten keine fundamentale Veränderung der Arbeitswelt. 15 September 2026. https://www.axa.ch/de/ueber-axa/medien/medienmitteilungen/aktuelle-medienmitteilungen/2026/20260915-kmu-arbeitsmarktstudie-kuenstliche-intelligenz.html
- SECO, SME Portal. Swiss medtech industry continues to grow. 25 September 2024. https://www.kmu.admin.ch/en/swiss-medtech-industry-continues-to-grow
- Pharmaceutical Inspection Co-operation Scheme (PIC/S). Joint stakeholders’ consultation on the revision of Chapter 4, Annex 11 and new Annex 22. 7 July 2025. https://picscheme.org/en/news/joint-stakeholders-consultation-on-the-revision-of-chapter-4
- Swiss Federal Council. AI regulation: Federal Council to ratify Council of Europe Convention. 12 February 2025. https://www.admin.ch/en/nsb?id=104110
- Federal Data Protection and Information Commissioner (FDPIC). Update: Current data protection legislation is directly applicable to AI. 8 May 2025. https://www.edoeb.admin.ch/en/update-current-legislation-directly-applicable-ai
- Swiss Code of Obligations, Article 328b (bilingual edition). Accessed 16 September 2026. https://www.droit-bilingue.ch/de-en/2/22/220-328b-437.html
- Adecco Group Switzerland and Stellenmarkt-Monitor Schweiz, University of Zurich. Fachkräftemangel Index Schweiz 2025: Fachkräftemangel nimmt 2025 das zweite Jahr in Folge spürbar ab. Press release, 27 November 2025. https://www.adeccogroup.com/-/media/project/adecco-group/switzerland/swiss-skills-shortage/2025/files/medienmitteilung_fkm_national_2025.pdf
- Netzwoche. KI drückt die Nachfrage nach IT-Fachkräften. 28 November 2025. https://www.netzwoche.ch/news/2025-11-28/ki-drueckt-die-nachfrage-nach-it-fachkraeften
- Swiss Federal Act on Data Protection (FADP), Article 21. Accessed 16 September 2026. https://prighter.com/resources/laws/swiss-fadp/fadp/articles/article-21