Where should a small business start with AI?

Start with a recurring task that has reasonably stable rules and data you already have. The priority is to learn from the work while keeping the cost of mistakes low. Your most important process is not necessarily the best place to start.

How long do you stay involved?

It depends on the support your business needs to choose and assess projects independently. You should be able to use the method from the first project on the ones that follow. If every decision still depends on us, we have not done our job properly.

What if no one in the business can take responsibility?

Then the project should not start. It needs someone who is accountable for the result and has time to follow it. If no one can take that role, we stop before committing resources.

What if people already use ChatGPT or similar tools?

Start by finding out who uses them and for which tasks. Then make clear which tools are allowed and what data can be entered, and check that the people using them understand those rules.

How is this different from consulting?

Consulting helps you make one decision: which task to address, how to measure it and which project to start. Adoption covers a sequence of projects over time: what comes first, who is accountable in your business and when to stop a project or decide against the next one.

Do you need a dedicated budget?

You need to know what you are willing to spend on the first project and on keeping it running. Planning for running costs is part of the initial decision, before the tool goes into use.

How do you know it is working?

Compare the baseline with the results at the agreed review. If there is no improvement, or you cannot measure one, that is useful information: a reason to stop investing in the project and reconsider whether to start the next one.

AI adoption

Bring AI into your business, one project at a time

A team buys a subscription; an employee starts using a tool. AI can enter your business before you've decided how to use it. We help you decide which projects to start and in what order, prepare data and people, assign responsibility and compare results before and after. That includes stopping what isn't working, even if you've already paid for it.

What AI adoption support involves

It is the work that connects your AI projects over time: deciding where to start and what to leave out, preparing data and people, measuring what changes in everyday work and deciding whether to take on the next project. Consulting addresses one decision and the project that follows. Adoption covers the sequence: how projects are chosen, managed and, when necessary, stopped.

The problems to solve

AI is already in use, but no one is in charge

One person uses an assistant for emails, another for quotes. There is no record of which tools are in use and no clear rules about which customer data can be entered.

Plenty of experiments, no priorities

“Bring AI into our business processes” is an intention, not a decision. When every task has the same priority, experiments pile up without changing how the work gets done.

Unsuccessful projects keep running

A tool you bought but never use keeps costing money. No one wants to drop it because that feels like admitting a mistake, and the next project meets more scepticism.

What we do

We set priorities across projects

We look at the tasks that take the most effort and weigh the work involved against the expected benefit. Together, we decide what to tackle first, what to postpone and what to leave out.

We prepare both the data and the people

Unreliable data or a lack of input from the people doing the work can stall a project. We check the data and involve those people before writing code.

Responsibility stays with your business

Each project needs someone in your business who is accountable for it and knows what to monitor. We work with you to identify that person. Without them, the project depends on us. We do not want to create that dependency.

We agree when to stop

Before work starts, we agree a review date and criteria for continuing, making changes or stopping the project. Having paid for it is no reason to keep it running.

What we do not do

We do not sell a transformation programme

No multi-year plans, no steering committees. One project at a time, with clear criteria for assessing it. If the results do not justify the effort, we change course or stop.

We do not start without a baseline

We measure how the work is done today before making changes. Without that baseline, you cannot tell whether anything has improved or make an informed decision about the next project.

We do not run generic AI training

People learn to use the tool they will work with, on the tasks they actually do. We do not offer courses disconnected from that work.

The work your business still has to do

We support you, but you cannot hand over responsibility for adoption. Your business must decide what to automate, who is accountable and when to stop. If those decisions depend on us, the work stops when you change provider or stop working with us.

Time from the people doing the work

You need to give the people doing the work time to contribute to the project. They know where time is lost and which exceptions are never documented.

Someone who keeps track of the projects

Someone needs to know which tools are in use, what has been decided and what is being measured. They do not have to be a technical specialist, but they must also keep track of which projects were stopped and why.

The discipline to review the numbers

On the agreed date, you need to review the results, especially if they are disappointing. It is easy to put that review off, but you need it to decide whether to continue, make changes or stop.

What to measure and review

Which AI tools are already in use in your company, and who chose them?
Which tasks come up most often, and how many hours do they take each week?
For the first project, what baseline will you use to compare the work before and after?
Who in your business is accountable for the result, and how much time can they actually commit?
When will you review the results, and what would make you stop the project?

Facts and sources

ISTAT reports that 15.7% of Italian businesses with 10–249 people in their workforce used at least one AI technology in 2025, up from 7.7% in 2024.

ISTAT — Imprese e ICT, anno 2025 — 15 dicembre 2025consulted on 2026-09-20

The Politecnico di Milano's Artificial Intelligence Observatory reports that 8% of Italian SMEs had started at least one AI project in 2025: 7% of small businesses and 15% of medium-sized businesses. A further 20% were considering starting one.

Osservatorio Artificial Intelligence, Politecnico di Milano — febbraio 2026consulted on 2026-09-20

Article 4 of the AI Act requires companies using AI systems to take measures to ensure their staff have a sufficient level of AI literacy. It has applied since 2 February 2025 and also covers small companies. It requires neither certified courses nor a minimum number of hours.

Commissione europea — AI Act, articolo 4consulted on 2026-09-20

Questions we often hear

Where should a small business start with AI?: Start with a recurring task that has reasonably stable rules and data you already have. The priority is to learn from the work while keeping the cost of mistakes low. Your most important process is not necessarily the best place to start.
How long do you stay involved?: It depends on the support your business needs to choose and assess projects independently. You should be able to use the method from the first project on the ones that follow. If every decision still depends on us, we have not done our job properly.
What if no one in the business can take responsibility?: Then the project should not start. It needs someone who is accountable for the result and has time to follow it. If no one can take that role, we stop before committing resources.
What if people already use ChatGPT or similar tools?: Start by finding out who uses them and for which tasks. Then make clear which tools are allowed and what data can be entered, and check that the people using them understand those rules.
How is this different from consulting?: Consulting helps you make one decision: which task to address, how to measure it and which project to start. Adoption covers a sequence of projects over time: what comes first, who is accountable in your business and when to stop a project or decide against the next one.
Do you need a dedicated budget?: You need to know what you are willing to spend on the first project and on keeping it running. Planning for running costs is part of the initial decision, before the tool goes into use.
How do you know it is working?: Compare the baseline with the results at the agreed review. If there is no improvement, or you cannot measure one, that is useful information: a reason to stop investing in the project and reconsider whether to start the next one.

Tell us which tools your business already uses

We start with the tools already in your business, who uses them and for which tasks. Together, we decide which project to tackle first, what to leave as it is and whether you are ready to start.