81% of Italian SMEs use AI. One in four has integrated it.
Between the two numbers there are fifty-six percentage points. In that space sits the majority of Italian companies that have already paid for artificial intelligence and haven't yet collected anything.
81% of Italian SMEs say they use artificial intelligence tools, but only one in four has integrated them into processes. About one in three uses them occasionally. The figure comes from the 2026 Observatory by Sibill and Astraricerche, on 500 Italian SMEs with turnover up to €10 million.
The main barrier isn't cost, named as the primary obstacle by only 6.6% of businesses. It is the skills to use it independently and trust in how data is handled.
What distinguishes the 25% that have integrated: a written process, a baseline measurement, a declared human checkpoint, a written answer on where the data ends up. None of the four is technology.
The number that counts isn't 81%
The Observatory puts it in a single line: «only 1 in 4 has integrated it into its processes on an ongoing basis». The same survey adds that «about 1 in 3 still uses it occasionally and in an unstructured way».
The 81% is cited everywhere, and it should be read for what it is: a measure of uptake. It says almost every Italian company has opened a chat, tried having an email rewritten, asked for a translation. It is a threshold already crossed, and as such it no longer describes a competitive advantage: if eight companies out of ten do it, doing it doesn't set you apart from anyone.
The number that describes a real difference is the other one. One in four has brought those tools inside a process, that is, into a point where work passes every day, with a stable rule and a verifiable outcome. The others have a tool open in a browser tab, used by whoever got curious, differently from person to person.
The difference between the two conditions is one of nature, not degree. A tool used at discretion produces results that depend on who uses it, don't accumulate, can't be measured and disappear when that person changes role. A process, on the other hand, produces the same outcome regardless of who is on shift, and it is the only form in which an improvement stays with the company instead of with the person.
Why it stops right there
The same survey also answers this question, and the answer contradicts the objection we hear most often. Cost is named as the main barrier by only 6.6% of businesses. The real obstacles are two, and both are human: the skills to use it independently, and trust in how data is handled.
This overturns the way most suppliers frame the conversation. If price isn't the problem, a discount solves nothing. If the problem is that nobody in the company knows where to start, and nobody knows where the data going out ends up, then what is missing isn't a cheaper subscription, but someone who writes the process and states where the data goes.
Then there is a figure that makes the picture more interesting, not less: 60% of those using it say they save at least five hours a week per person. Five hours a week, for a single person, is about twenty-five working days a year. So the saving already exists, and it is declared by those who use it occasionally. The question that follows is the only one that matters: if this is what you get without having structured anything, what is left on the table?
The European data says the same thing in different words
It isn't an Italian peculiarity. An OECD survey of more than 5,000 SMEs in seven countries (Austria, Canada, Germany, Ireland, Japan, Korea and the United Kingdom) finds that 65% of SMEs using generative AI report improved staff performance, far more than those who say they used it to scale up (35%), to compete with larger companies (29%) or to increase revenue (26%).
It is a hierarchy worth noticing. The benefit companies recognise first is internal: people work better. What they recognise much less is the effect on turnover. That doesn't mean the effect isn't there: it means that, in most cases, nobody has measured the step from one to the other. The time saved is felt; where that time went isn't.
The same survey adds a detail that works as an operational instruction: the benefits (time saved, quality, job satisfaction) turn out 10% to 40% greater when the employer encourages its use, instead of leaving it to individual initiative. It says nothing about technology. It says a lot about how the company is run.
These surveys are companies' statements, not measurements on their systems. An owner who estimates five hours saved a week is remembering, not timing, and declared estimates tend to be generous towards the choice already made.
That doesn't make them useless, but it makes it wrong to use them as a promise. Nobody can tell you how many hours you would recover starting from these numbers, because those numbers describe other companies. They serve to justify a measurement, not to replace it.
What, in practice, distinguishes the 25% from the rest
It isn't the sophistication of the tools. Companies that have integrated almost always use exactly the same tools as the others. What they have in addition is four things, all boring:
- A written process. They can say, without hesitating, at which point of the work the system steps in, what it receives and what it produces. Whoever isn't integrated answers this question with the name of a tool.
- A baseline measurement. They know the before value: how many hours, how many documents, how many minutes of waiting. Without that number, the improvement is a feeling with an invoice attached.
- A declared human checkpoint. They know precisely what the system can do on its own and what must go through a person. Not out of mistrust: because it is the only way responsibility stays where it should.
- A written answer on data. Where it is processed, who the provider is, whether it can use it to train its own models. It is the survey's second barrier, and it is brought down with a contract, not with a verbal reassurance.
None of these four things is technology. They are all decisions, and they are all taken before anyone builds anything.
The consequence for whoever decides
If your company is in the 81%, you aren't behind: you are where almost everyone is. The gap to close isn't towards those who use AI, because that race has already ended in a draw, but towards the quarter who have put it inside a process, and that gap is covered with work that has nothing technological about it: looking at where work gets stuck, measuring it, and building a single piece where the loss is most evident.
How to look, which four criteria to use to choose the process and what to measure before switching anything on is in the method: which process to automate first.
It is also why, on this site, you won't find packages with a price next to them. Which piece comes first depends on a number we don't have today, and that can't be guessed from outside.
Questions and answers
How many Italian SMEs really use artificial intelligence?
81% say they use it, but only one in four has integrated it into processes on an ongoing basis, and about one in three uses it occasionally and in an unstructured way. The figure comes from the 2026 Observatory by Sibill and Astraricerche, on a sample of 500 Italian SMEs with turnover up to €10 million.
The distinction matters because 81% measures uptake, a threshold almost everyone has already crossed and so it sets nobody apart. 25% measures integration, which is the only one of the two conditions in which an improvement stays with the company instead of with the individual.
What is the main barrier to AI adoption in Italian SMEs?
Not price. In the same survey cost is named as the main barrier by only 6.6% of businesses. The two real obstacles are the skills to use it independently and trust in how data is handled.
The practical consequence concerns whoever sells: if price isn't the problem, a discount solves nothing. What is missing is someone who writes the process and states in writing where the data ends up.
How much time does AI really save a small business?
60% of those using it say they save at least five hours a week per person, which for a single person is worth about twenty-five working days a year. So the saving already exists today, and it is declared even by those who use it occasionally.
It should be read with a precise caution: these are companies' statements, not measurements on their systems. An owner estimating five hours is remembering, not timing. The number serves to justify a measurement in your company, not to replace it.
What distinguishes an SME that has integrated AI from one that merely uses it?
Not the tools, which are almost always the same. Four things, all boring and all decided before building: a written process (they can say at which point of the work the system steps in, what it receives and what it produces), a baseline measurement, a declared human checkpoint, and a written answer on data, that is, where it is processed and whether the provider can use it to train its own models.
Whoever isn't integrated answers the first question with the name of a tool.
Sources
- Sibill · Astraricerche, Osservatorio 2026: AI per le PMI. Sample of 500 Italian SMEs with turnover up to €10 million. The source of: 81% usage, one in four integrated, one in three occasional use, 60% reporting at least 5 hours saved a week per person, cost as a barrier for 6.6%. Quotations in our translation.
- OECD, Generative AI and the SME Workforce. 2024 survey of more than 5,000 SMEs in Austria, Canada, Germany, Ireland, Japan, Korea and the United Kingdom. The source of: the 65% reporting improved staff performance, the comparison with scale (35%), competitiveness (29%) and revenue (26%), and benefits 10% to 40% higher where use is encouraged by the company.
- OECD, Empowering SMEs in the Age of AI. On the gap between using generic tools and integrating them into processes.
Your number, not the sample's.
Twenty questions, five operational dimensions. A score, a chart and three priority interventions, in five minutes. No spam, no data requested beyond what is needed.