AI in Polish small businesses: an eightfold gap and the barrier of knowledge

In Poland, the share of companies using AI is nearly eight times higher among large enterprises than among small ones. Research shows that small businesses are held back primarily by a lack of knowledge, not by budget constraints. We looked into where this gap comes from and where a small business can start.

According to Eurostat data for 2025, artificial intelligence was used in Poland by 45.8 percent of large enterprises (250 employees or more), 14.8 percent of medium-sized ones (50-249 employees), and only 5.8 percent of small ones (10-49 employees). The EU average is 55 percent, 30.4 percent, and 17 percent respectively. Poland lags behind in every group, but the smaller the company, the wider the gap. Among large companies, we are just under 10 percentage points below the EU average. Small companies use AI three times less often than their EU counterparts.

We compared these figures with the results of Polish studies on the barriers to and effects of implementation, as well as with the experience of a sole proprietorship that built an advantage over agencies using AI. This shows that money is not the main obstacle for small businesses. What they lack above all is knowledge of where to start.

An eightfold difference, and microbusinesses are not included in the statistics

Across the economy as a whole, AI is used, according to Eurostat, by about 8.4 percent of Polish companies employing at least 10 people. The EU average is about 20 percent, and the leaders, Denmark and Finland, reach 42 percent and 37.8 percent respectively. Poland is near the very bottom of the ranking.

This statistic does not include companies employing fewer than 10 people, that is, microenterprises and sole proprietorships. These make up the majority of Polish companies, and no one counts them in this comparison. A broader picture is provided by the European Commission's indicator for 2025: a basic level of digital intensity was reached by 59 percent of Polish small and medium-sized enterprises, compared with 71.4 percent in the EU. The AI gap is therefore part of a more general lag in digitalization.

Why one study says 8 percent and another 75 percent

Figures on AI in companies need to be approached with caution, because studies measure different things. Eurostat reports about 8.4 percent. A report by PARP (the Polish Agency for Enterprise Development) and the Jagiellonian University as part of the Human Capital Balance 3.0 from April 2026 shows that about 23 percent of companies actively use AI. According to a CRIF study from spring 2026, 75 percent of Polish entrepreneurs say they use AI.

The results differ because the groups studied and the definitions are different. Eurostat covers all sectors. PARP studied 1,822 employers from 12 sectors with high potential for AI use. CRIF surveyed 500 business decision-makers in five countries. For some, using AI means writing emails in ChatGPT; for others, it means integration with the company's systems. However, the CRIF study confirms the same direction as Eurostat: in it, medium-sized and large companies used AI nearly twice as often as microenterprises.

Even fewer companies see real effects. According to the Grant Thornton study “AI Adaptation Poland 2026,” known to us from reporting by the PulsHR service, only 7 percent of companies declare full implementation and a measurable impact of AI on results. According to PARP, 46 percent of enterprises are “traditionalists” with low AI readiness, and only 6 percent are “innovation pioneers.”

Money is not the first obstacle

In the Grant Thornton study, the most frequently indicated barrier was insufficient knowledge and employee competencies, cited by 33.9 percent of companies. However, barriers depend on the level. CEOs more often talk about pressure for a quick return on investment and difficulty justifying the expense. Senior management points to data security and regulatory compliance. Middle management talks about skills gaps.

PARP data leads to a similar conclusion. Only 28 percent of companies that use AI or plan to implement it have access to qualified IT staff. Nearly half of companies do nothing to fill skills gaps. In micro and small companies, AI is usually tested and implemented by the owner or the management board alone, that is, by the people who have the least time for it. In medium-sized and large organizations, AI is used more broadly, and employees in higher positions have access to advanced features.

This is the main advantage of corporations. A large company can hire people to handle tool selection and data integration. A small company has an owner, a few subscriptions, and instructional videos at its disposal.

Where to start in a small business

Three practical conclusions emerge from comparing these sources. First, it is worth starting with one process that takes the most time, not with a package of subscriptions. Amelia Szukowska, Vice President of the Foundation of the Coalition for Polish Innovation, gives similar advice. In her opinion, a small company does not need an extensive AI system or a multimillion budget right away, and often improving a single process is enough. Second, before making a purchase, it is necessary to determine how the company will measure the effect. Without this, it will be difficult to assess whether the expense is paying off. Third, competencies can be obtained externally, from a contractor or from public programs.

An example of such a program is WAMA EDIH (EDIH stands for European Digital Innovation Hub), a project run by the Foundation of the Coalition for Polish Innovation. In 2024-2026, it delivered 1,000 services supporting the digitalization of small and medium-sized enterprises with a total value of nearly PLN 20 million, as well as 360 digital maturity assessments. The assessment helps evaluate a company's technologies, processes, data, and competencies before it spends money on tools.

The gap between small and large companies is wide, but its main cause is a lack of knowledge, not a lack of money. For a small business owner, the first step is therefore an assessment and one well-chosen process, not another subscription.