What a Fractional Manager must leave behind on AI so it isn't a plug-in: data from Italian SMEs

Ask an SME whether it uses AI and, according to the data, the answer is almost always yes. Ask who decides how it is used and the silence lasts a few seconds too long.

Some look for a Fractional Manager “who knows AI”, the way you would look for someone who knows Excel. The logic is that of the plug-in: add the skill and the problem disappears.

The data says the problem is elsewhere.

What the numbers say about Italian SMEs

• According to the Sibill-Astraricerche survey of 500 companies (April 2026), 81% already use AI tools. Only 1 in 4 has integrated them into its processes on an ongoing basis.
• The most cited barrier is a lack of skills to integrate the tools. Costs weigh in at 6.6%.
• The Osservatorio Innovazione Digitale nelle PMI (Digital Innovation Observatory for SMEs) at Politecnico di Milano finds that 76% of SMEs have not invested in AI. They do not plan to in the short term.
• In the Qonto survey of 1,000 companies and self-employed workers, collected in May 2026, 41.7% of those who use AI say it has changed nothing in their daily work.

Read together: AI is inside, but it is not governed. The tool gets used, the work does not change.

The problem for SMEs is not a lack of AI tools. It is that no one is responsible for deciding how they enter the work.

Why “the manager who knows AI” is not enough

A manager with solid technical skills in AI, placed in the company 2 days a week, brings two things: their knowledge and their time. If they leave only the first behind, everything goes back to how it was when they go.

Tools versus governance

• Tools: choosing the tool, writing the prompts, giving the team a demo.
• Governance: deciding which processes touch AI, which data can go in, who validates the output, what gets measured.

The first can be copied in an afternoon. The second holds over time. It also matters when the regulatory picture changes: we cover that in the piece on the AI Act delay.

Fractional Manager or software vendor?

Sometimes a client mistakes whoever proposes an AI path for a software vendor, and answers that they already have their management software. The difference shows in what is promised. A vendor sells a tool. A Fractional Manager decides with you where AI comes in and where it does not.

A credible Fractional Manager also tells you where AI is not needed. Sometimes the right answer is one more person, or one fewer level of authorisation.

Signs you are only buying tools

• The quote talks about tools and licences, not roles and responsibilities.
• No one asks which data the team uses today.
• The expected result is “using AI”, not a measure (hours, errors, response times).
• At the end of the engagement there is no document that someone else in the company can apply.

What a Fractional Manager should leave behind on AI

1. An internal owner with a name and surname, even part-time.
2. Minimum rules on data: what can go into an AI and what cannot.
3. A way to validate outputs before they reach a client.
4. A measure of what changes, chosen at the start.
5. A review rhythm, because tools and rules change.

We describe this in the article on Fractional Managers and AI: where time is lost across multiple clients. AI frees up time only if someone decides where to put it.

The question to ask before signing

Ask whoever proposes a Fractional Manager on AI just one thing: “What stays in the company when you leave?” If the answer is prompts and tools, you are buying a plug-in. If it is roles, rules and a measure of results, you are buying governance.

If you are a Fractional Manager and want to build this method, start from the For Professionals path. If you want to see where a company stands, there is the AI Governance assessment.