The European Commission has published a Eurobarometer survey on how smaller companies use AI. It is the clearest benchmark yet for a business your size. Based on around 12,826 phone interviews with heads of SMEs and mid-caps employing 1 to 749 people, 42% said they already use AI and algorithmic management, and another 21% plan to. The survey came out on 14 October.
The headline is that your peers are moving. The detail underneath matters more: most of that use is still limited and experimental. Plenty of businesses have tried something. Far fewer have a process that runs on it every day.
That gap is your opening. A small business that picks one process and makes it work properly can still get ahead of competitors who are only dabbling.
Where businesses are actually using AI
The survey shows three areas leading the way: customer service and client relations (47%), monitoring performance and productivity (36%), and logistics, planning and staff scheduling (30%).
These have something in common. They are repetitive, they come in volume, and most cases follow clear rules. A shop answers the same order status questions every day. A clinic fields the same questions about opening hours, prices and booking. A logistics firm rebuilds the same shift plan every week with small changes.
That is where AI earns its keep in a small business. It drafts the reply, sorts the inquiry, or proposes the rota. Your team checks and approves instead of starting from a blank page. The same survey found that 61% of these companies say AI helps them make decisions.
Vendor data points the same way. OpenAI reported on the same day that the number of small-business users of its products in the EU has grown 8x, counting businesses with 200 or fewer associated users. That is one vendor measuring its own customers, so treat it as a signal of direction rather than a market figure.
The three obstacles, and why a small pilot handles them
SMEs named three main reasons for holding back: high investment costs (50%), legal uncertainty (49%) and lack of expertise (48%). All three are real. All three get smaller when you shrink the scope.
Cost. A pilot on one process has a fixed budget and an end date. You pay for a few weeks of one tool and some setup time, and you decide whether to continue based on results.
Legal uncertainty. Reviewing one data flow is manageable. You know which personal data goes in, where it is stored, and who sees the output. Reviewing "AI across the business" is not manageable, which is why many owners stall.
Expertise. Your team learns on a process they already know inside out. The person who answers customer emails every day is the best judge of whether a drafted reply is good. That knowledge transfers to the next project.
Take extra care with monitoring and scheduling
The survey groups AI with "algorithmic management". In practice that means software that tracks how staff work, measures their productivity, or assigns shifts and tasks. Two of the three most common uses above fall into this category.
These tools process personal data about your employees, so GDPR applies in full. You need a clear legal basis, you need to tell staff what is collected and why, and you may need a data protection impact assessment. In some countries, a works council or staff representatives must be consulted before you introduce this kind of system.
The EU AI Act also treats some uses of AI in managing workers, such as allocating tasks or evaluating performance, as higher risk, with extra obligations for the businesses that deploy them. Check where your planned use sits before you switch anything on, not after.
If you want the simplest legal path for a first pilot, start with customer inquiries. Scheduling and performance tools can come next, once you have the review process in place.
What to do this week
- Pick one repeatable process. Write down two or three candidates, such as customer inquiries, appointment booking or the weekly staff rota. Choose the one with the most volume and the clearest rules.
- Name an owner and a measurable goal. One person is responsible for the pilot. Measure the baseline now: how long a typical inquiry takes to answer, or how many hours go into the rota each week. Without a baseline you cannot judge the result.
- Map the data. List what goes into the tool, which of it is personal data, and where the vendor stores it. Ask for the data processing agreement and check whether data stays in the EU.
- Set a budget and a time box. Fix the spend and the number of weeks up front. At the end, decide to keep, adjust or stop, based on the goal you set.
- If staff data is involved, talk first. Before testing any scheduling or monitoring tool, speak to your data protection contact and your team. Explain what the tool does and what it does not do.
The Eurobarometer numbers show that AI in small businesses is now normal, but mostly still at the trial stage. That leaves room for a business that does one thing well: a single process, a named owner, a clear goal and a legal check before launch. Do that once and the next project gets easier, cheaper and faster. If you want help scoping that first pilot, Stolos Digital sets up these systems for European SMBs.