AI in a Polish company — why our market is not a copy of the Western one
Artificial intelligence is talked about today as if it were already everywhere. The data says otherwise: in Polish companies AI is only just arriving, arriving slowly and — most importantly — very unevenly. Understanding that unevenness is key, because a deployment done “the American way,” pasted in without adapting to our market, usually fails to take hold. This text shows what AI adoption in Poland really looks like — on hard data from the national statistics office (GUS) — and what follows from it for a company weighing AI.
How many of us actually use AI
According to GUS, in 2025 8.7% of companies in Poland declared using AI technologies — up from 5.9% a year earlier. The growth is real, but the level is still low: fewer than one company in ten. This is an important reference point, because it corrects the impression built by headlines — AI is not the norm in Polish companies, it is still the exception. Poland also remains clearly below the European average.
The gap that says the most: large versus small
The most interesting figure is not the average, but the spread. Because 8.7% is the resultant of two very different worlds:
- large companies: 42% (33% a year earlier),
- medium: 15.6% (10.4%),
- small: 6.1% (4%).
A large company reaches for AI seven times more often than a small one. This is not chance — it shows where the barrier lies. Large organizations have the budget, people and time to cross the entry threshold. Small and medium ones often do not — even though the benefit could be proportionally greater for them. And it is precisely this group, not the corporations, that is the real field on which AI adoption in Poland will be decided.
Which sectors lead
The sector breakdown is telling too:
- ICT: 36.1% (most often: text and speech generation — 24.9%, machine learning — 16%),
- science: 23.2%,
- industry: 7.8%,
- food sector: 5%.
The pattern is clear: the lead goes to sectors that are themselves close to technology. The rest of the economy — the largest part — is only at the start of the road. That means the greatest untapped potential lies not where AI already is, but where it is not yet.
The barriers: cost and competence
GUS points to two main entry barriers: cost and the competence gap — a lack of people who know how to deploy and maintain AI. This is consistent with what the global cost data shows: companies that did not count the full bill, or had no one to put on the deployment, back out faster. These two barriers — money and knowledge — also explain the gap between large and small: the large have both, the small usually neither.
Why it cannot be a copy of a Western deployment
The Polish market has its own specifics, invisible in American guidebooks:
- Local language and data. Polish documents, Polish terminology, Polish legal realities — an off-the-shelf model, trained mainly on English content, handles them worse than the brochures suggest.
- Company structure. Our market is, to a huge extent, small and medium enterprises — exactly the group hit hardest by the cost and competence barrier. A solution designed for corporations will not work here.
- Data concerns. Some sectors (such as BPO) have real constraints on where data may go — which shifts the calculus toward local solutions.
A deployment that ignores these specifics lands in the statistic of abandoned projects. One that accounts for them — matching scale, cost and method to the realities of a small or medium Polish company — has a chance to take hold.
What to do about it
If you are a small or medium company looking at that 6.1%, the conclusion is not “not yet the time.” It is: the advantage is still there to take, because most of the competition hasn’t moved either. The conditions to do it sensibly are two — exactly the reverse of the barriers in the GUS data:
- A counted, predictable cost — matched to the company’s scale, not to a corporate budget.
- Someone to lead it — because the competence gap is not a reason to give up, but a reason to deploy with someone who knows how. A local solution, matched to Polish data and led by a partner who knows the realities, addresses both barriers at once.
Summary
AI adoption in Poland is low (8.7% per GUS), growing, but very unevenly — large companies use AI seven times more often than small ones, and the barrier is cost and competence, not technology. This is not a market where a pasted-in Western deployment works: the Polish language, Polish data, the dominance of small and medium firms, and privacy concerns all demand a different approach — matched to scale and reality.
For a company that is weighing it up, the conclusion from this data is surprisingly optimistic: since only one large company in six and one small company in sixteen uses AI, the advantage is still on the table. The question is not “will I make it in time,” but “will I do it matched to myself, before the competition does.”
MafiaAI — a team of people and AI agents building tools, websites and solutions for Polish companies: AI matched to local data, scale and realities, led from start to finish. More: t8.pl
Sources: GUS, “Use of AI technologies in enterprises 2025” — via PAP Biznes; Polish Economic Institute (PIE), EY, PwC. All institutional/hard sources.