ARTICLE BODY

Artificial intelligence in Swiss SMEs: adoption, opportunities and what to do in 2026

In a nutshell. In 2025 34% of Swiss SMEs consciously use artificial intelligence, up from 22% in 2024. 57% of users report concrete time savings. The main applications are translation, correspondence automation and data analysis. The number-one obstacle is not cost but data management and nFADP compliance: only 34% of SMEs have precise rules on what to feed into AI tools. Switzerland does not yet have an AI law, but the nFADP already applies — and a preliminary draft is due by the end of 2026.

The 2025 figures: AI grows fast in Swiss SMEs

The 2025 AXA+Sotomo study, also published by kmu.admin.ch, offers the most up-to-date picture of AI adoption in Swiss SMEs. In one year, the change has been sharp:

• 34% consciously use AI (it was 22% in 2024, +12 points).
• 37% have already run tests or pilot projects.
• 29% do not use AI in any form (it was 45% in 2024, -16 points).

The 2024 Raiffeisen study on the AI sector adds an interesting figure: 54% of SMEs had already launched pilot projects, but only 9% used it systematically. There is therefore a large band of companies that have explored AI without yet really integrating it into processes.

The overall message is clear: AI is no longer a novelty reserved for large companies. It is spreading quickly in SMEs too, often without a structured strategy — but it is spreading.

How they use it: the most common applications

Among Swiss SMEs that use AI tools, the most common applications in 2025 are (AXA+Sotomo data):

The first two applications — translation and correspondence — are the most accessible, require little infrastructure and deliver visible results immediately. Unsurprisingly, they are where adoption is highest.

The fastest growth is in data analysis and in workflow optimisation: two areas that bring real value to management, not just communication convenience.

The benefits SMEs actually measure

What do those who have already adopted AI tools say? Data from the same 2025 AXA+Sotomo study:

• 57% report concrete time savings (it was 46% in 2024, +11 points).
• 45% see AI as a competitive advantage for their company (+10 points).
• 60% consider it an opportunity for the future.
• 10% record turnover growth thanks to AI.
• Only 2% have reduced staff because of AI.

This last figure is important: the impact on employment is predominantly neutral or positive. AI changes the required skills — some activities are automated, new ones are created — but it is not the job-cutting tool many fear.

The main obstacle: data and compliance

The real brake on adoption is not the cost of tools (increasingly accessible) but data management. Only 34% of Swiss SMEs have defined precise rules on what data can or cannot be fed into AI tools. The situation worsens in smaller companies: among SMEs with 5-9 employees, only 23% have internal regulations on data in AI.

This is where many SMEs expose themselves to real risks: client data, contractual information, employee health data may end up in AI systems without the company having assessed the implications for data protection. And the nFADP already applies.

The other obstacles declared by SMEs are: lack of in-house skills, costs of integration into existing systems, and regulatory uncertainty. The negative perception of AI, on the other hand, has dropped: only 13% see it negatively in 2025, against 20% in 2024.

The Swiss regulatory framework: what already applies today

This is the point often misunderstood by SMEs, so it is worth being clear.

Switzerland does not have a specific law on artificial intelligence in 2026. The Federal Council's approach is sectoral: existing laws are adapted sector by sector, rather than creating a cross-cutting rule like the European AI Act.

What is in progress:

• March 2025: Switzerland has signed the Council of Europe AI Convention, the reference international treaty. The Federal Council launched the ratification process on 12 February 2025.
• By end 2026: a preliminary draft is expected from the Federal Department of Justice (FDJP) on transparency, data protection and non-discrimination in AI, plus a non-binding measure plan from DETEC.
• Switzerland does not automatically apply the EU AI Act: it has an autonomous approach, aligned with European principles but not bound to follow them.

What already applies today: the nFADP. The Federal Data Protection and Information Commissioner (FDPIC) has officially confirmed that the new Federal Act on Data Protection (in force since 1 September 2023) directly applies to the processing of personal data based on AI. The law is formulated in a technology-neutral way and also applies to AI systems.

The concrete obligations that follow:

• Transparency: those who use AI must disclose the purpose, the operation and the data sources.
• Information: users and clients have the right to know if they are interacting with a machine.
• Impact assessment: for high-risk cases (automated decisions about persons, profiling, processing of sensitive data), a prior assessment is required.
• Forbidden applications: large-scale facial recognition, lifestyle surveillance, "social credit" systems are explicitly excluded.

You should not wait for a new law to comply: nFADP compliance is already an obligation for anyone who uses AI tools that process personal data.

AI in accounting and administration: practical use cases

For SMEs looking for the most concrete starting point, the administrative and accounting sector is the one with the best effort-to-benefit ratio:

• OCR and invoice automation: systems that automatically recognise invoices, extract data and post them into accounting without manual entry. Reduces errors and time, with a measurable impact already in the first weeks.
• Correspondence automation: handling of standard emails, confirmations, reminders, document requests — without the team having to reply manually to every repetitive message.
• Automatic document summarisation: contracts, minutes, reports — AI can extract key points and reduce reading and review time.
• Predictive cash flow analysis: tools that use historical accounting data to forecast liquidity and flag potential issues before they become problems.
• Regulatory and tax research: systems that automatically search the updated regulation in response to specific questions.

These are all tools already available and accessible to SMEs, not research projects: many accounting software packages already in use integrate some of these features.

Action plan for an SME wanting to start in 2026

The plan recommended by SECO and the main studies on the topic is gradual — not a revolution, but a progressive integration:

1. Prepare your internal data: before using any AI tool, get your data in order. Disorganised data in produces useless output.
2. Start with simple applications: translation, correspondence management, automatic invoice processing. Low risk, visible benefit immediately.
3. Train the staff: in-depth technical training is not needed — those using the tools just need to understand what they can and cannot input, and how to assess the output.
4. Define clear data rules: what data can be inserted into external AI tools? Client data never, without assessment. Contractual data only with adequate protection measures.
5. Verify nFADP compliance: already applicable. If you use AI tools that process personal data of employees, clients or suppliers, an impact assessment is needed for high-risk cases.
6. Communicate transparently: if you use AI in communications with clients or suppliers, inform them. Not just for legal obligation: transparency builds trust.

How we help

Fidav does not provide AI technology advice, but works every day with the raw material of corporate AI: data. Accounting, financial statements, documents, deadlines, administrative processes — it is all information that, if well organised, can be managed with much greater efficiency, also with the support of the right digital tools.

If you are considering how to digitalise your administrative processes, how to organise accounting so that it can be used as a decision-making tool, or how to manage your data in compliance with the nFADP, we are the reference for the fiduciary and compliance side. Read more on our accounting and financial administration and our corporate and strategic advisory.

Want to understand how to make your SME more efficient and digital? Chat with us on WhatsApp at +41 79 741 02 89 or call +41 91 640 40 20.

FAQ (visible on page + FAQPage schema above)

How many Swiss SMEs use artificial intelligence in 2025? According to the 2025 AXA+Sotomo study published by kmu.admin.ch, 34% of Swiss SMEs consciously use artificial intelligence tools in 2025, up from 22% in 2024. 37% have already run tests or pilot projects. Only 29% do not use AI in any form, against 45% the previous year.

What do Swiss SMEs use AI for? The most widespread applications are automatic translation (52%), correspondence management and internal process automation (47%), data analysis for strategic decisions (32%) and optimisation of work phases (34%, up 11 points vs 2024).

Does Switzerland have a law on artificial intelligence? No, in 2026 Switzerland does not yet have a specific law on AI. The approach is sectoral. Switzerland signed the Council of Europe AI Convention in March 2025. A preliminary draft is expected by the end of 2026. The nFADP (in force since September 2023) already directly applies to AI-based data processing.

Is the use of AI in companies compatible with the Swiss nFADP? Yes, but with precise obligations. The FDPIC confirmed that the nFADP applies directly to AI. Those who use AI must ensure transparency, inform users when they are interacting with a machine and carry out an impact assessment in high-risk cases. Only 34% of Swiss SMEs currently have precise rules on the data to be inserted into AI tools.

Where does a Swiss SME start when it wants to adopt AI in 2026? The first step is to prepare your internal data and start with simple applications: automatic translation, correspondence management, automatic invoice processing (OCR). Then train the staff, define clear rules on which data can be inserted into AI tools, and verify nFADP compliance.