AI is already everywhere in Swiss SMEs.
But are you really making the most of it?

Good news: artificial intelligence has become an everyday working tool in Swiss businesses. According to an EY survey cited by the Confederation's SME portal, 89% of employees use it every day. But there's a gap between using it and truly making the most of it, and the obstacle has a precise name: scattered data and unclear rules.

The good news: AI has properly arrived in the workplace

Adoption is a reality now, not just a promise. The Confederation's SME portal (SECO), drawing on an EY survey of 604 employees, captures an AI that has become part of everyday working life in Switzerland. Built-in tools like Microsoft Copilot and Google Gemini are used by 70% of employees. It's a huge cultural shift, and it's happened in record time.

But we're only just getting started

The enthusiasm is justified; the maturity isn't, not yet. The same study says it clearly: using AI every day doesn't mean it has truly been integrated. 55% of companies have adopted it strategically in one or more areas, but 31% are still at the pilot stage and 14% haven't launched any concrete initiative at all. And above all: only 9% say AI has transformed their business model.

In other words: almost everyone uses it, very few have changed how they actually work because of it. It's the difference between keeping ChatGPT open in another tab and having business processes that genuinely run better thanks to AI. The first is curiosity; the second is competitive advantage. And this is exactly where the next two years will be won or lost.

The number one obstacle? Scattered data

The main barrier isn't technology, it's data. In the EY survey, the most-cited obstacle to integrating AI into business processes is data quality and information silos (20%), ahead of security and data protection (19%) and a lack of skills (18%). In plain terms: AI doesn't perform well when a company's knowledge is scattered across emails, folders and different people's heads.

It's precisely the subject we covered at length in our piece on how an organised company knowledge base, a "company brain," makes automation reliable. The message from the study and the message of our article line up exactly: first you put the knowledge in order, then AI becomes useful. Without that step, the tools stay clever but empty, because they don't actually know the business they're working in.

The hidden risk: vague policies and personal accounts

There's a second problem, quieter but just as serious: how companies govern the use of AI. Internal policies, even in small businesses, are often still vague, swinging between banning it outright and allowing use through personal accounts. And that's a dangerous grey area.

The numbers confirm it: 29% of employees are allowed to use AI through personal accounts (on ChatGPT, for example), and only 3% of companies ban its use completely. That means company data, quotes, contracts, customer information, can end up in ungoverned tools, with no control over what gets shared or where it ends up. It's not a question of banning things: it's a question of clear rules and company-provided tools, instead of a "do it yourself" approach with private accounts.

Compliance matters, and employees feel it too

Sensitivity around data is high, and it's coming from the ground up. For 51% of employees, data sovereignty is a concrete concern: they believe artificial intelligence tools should fully comply with Swiss or European data protection standards. It's no longer just a legal-department matter, it's a widespread expectation among the people using AI every day.

It's the same principle that the EU AI Act and the incoming Swiss rules are converging on: transparency, control over data, accountability. We covered this in detail in our article on the AI Act and its impact on Swiss SMEs. Anyone getting organised now isn't chasing the rules, they'll find they're already meeting them.

In short: how to make adoption safe and effective

The data from the study tells a consistent story. AI is everywhere, but adoption is still at an early stage, and the two main brakes, scattered data and vague rules, are exactly the two things you can act on right away. The key to safe, effective adoption comes down to two moves: organising your company knowledge (the company brain) and setting a clear policy on AI use. The rest, the tools themselves, comes after, and that's the easy part.

And there's one last thing that enthusiasm for AI often makes people forget: better processes don't come from one big project, but from the sum of many small automations. That's exactly the approach we take at Dedalix. We do it without adding new tools to learn, AI is built into the ones the team already uses, and it starts with good data management: we eliminate silos and put the business in a position to unify all its specific knowledge into a format that's useful to AI and the tools that draw on it. That way every automation rests on solid foundations, laying the groundwork for future development that's consistent and ready for ever-wider adoption.

And it's precisely because of all this, data to organise, rules to write, compliance to keep on top of, that it pays to rely on people who've made AI automation their trade. A focused partner speeds up adoption and brings out the benefits even from the simplest automations, because they know where to look and what to avoid. For us at Dedalix, automation with artificial intelligence and everything around it: data, company knowledge, policy, training and compliance, isn't one service among many or a side activity: it's the centre of what we do, every day.

Frequently asked questions