000
For a micro-business, the right starting point comes down to three decisions: one use case to tackle, one paid general-purpose tool, and two or three people trained before buying anything else.
Read the full guide
AI in Business 2026: Complete SME Guide (Strategy, ROI, Adoption)
Answer a few questions and get a personalized assessment with recommendations tailored to your industry.
Assess my AI maturityRelated articles
Dive deeper with these complementary articles.
AI in business means handing software systems the tasks that used to require skilled human time: writing, analysis, sorting, customer replies. In 2026, the question for an SME is no longer whether to experiment, but which two or three uses to pick, measure and sustain.
For a micro-business, the safest starting point comes down to three decisions: pick one use case, take one paid general-purpose tool, and train two or three people before buying anything else. The starting budget is counted in tens of euros per month, not thousands. For the SME-wide picture, see the guide to AI in business; here we look at a team of fewer than 10 people where every euro counts.
The right first project is a repetitive, text-based task that already eats time every week. Not an "AI project". The usual candidates in a small team: answering quote requests, writing meeting minutes, preparing LinkedIn posts or the newsletter, summarizing long documents, rewording letters.
A simple selection test: list five tasks that come back each week and note, for each, the time spent, the tolerance for error and the sensitivity of the data. Keep the one that takes the most time, tolerates human review and does not involve health data, payroll or trade secrets. If you hesitate between several, the method in the AI audit for SMEs can be shrunk to half a day in a small structure.
Illustrative example: a 4-person consulting firm spends 3 hours a week writing commercial proposals. That single use case is enough as a starting point. The rest can wait.
A reasonable starting budget for a micro-business has three lines: a subscription, training time, and testing time. The table below is a template; amounts are calculation assumptions to replace with real prices, which vendors change regularly.
| Item | Calculation assumption | Yearly order of magnitude |
|---|---|---|
| General-purpose tool subscriptions | 5 seats at 20 EUR per month | 1,200 EUR |
| Short training (hands-on start) | 1 to 2 days for 2 to 3 people | to be quoted depending on provider and funding |
| Testing and review time | 2 h per week for 4 weeks | internal time |
| Automation and development | postponed | 0 EUR at the start |
The last line matters most: anything involving integration, automation or custom development is decided after the first measured gain, not before. For a broader view of costs, see how much AI costs an SME in 2026.
A paid general-purpose assistant covers most micro-business needs, and it comes before any specialized tool. Two reasons: free versions usually come with usage limits and data-handling terms less suited to professional use, and a general tool is reusable across ten tasks while a specialized one covers just one.
Selection criteria, in order: data-handling terms (are your exchanges used to train the model or not), hosting and GDPR compliance, ease of use for non-technical staff, and compatibility with what you already use (email, office suite). Price comes last, because the gap between two equivalent plans weighs little against the cost of time lost on a bad choice.
Do not multiply licenses: start with the two or three people who will really run the trial, and extend later. An unused license is the first source of waste seen in small teams.
One hour of well-targeted training pays off more than an extra subscription. What separates a disappointing result from a useful one is the quality of the instructions given to the tool (the prompt), the context supplied and the way the answer is reviewed. This is learned in a few hours of guided practice on your own cases.
On obligations, Article 4 of Regulation (EU) 2024/1689 (AI Act) asks organizations using AI systems to ensure a sufficient level of AI literacy among the people involved. Company size does not remove that logic: short, documented training is a simple way to address it. Details are in our analysis of AI training for micro-businesses; to compare formats, see also operational AI training.
Three written rules are enough to start: which data must never be entered, which tools are allowed, and who reviews before sending. Without them, each employee improvises, and you get the phenomenon described in the risks of Shadow AI: tools used outside any framework, with client data in them.
A minimal template fits on one page:
This page can later grow into a proper company AI policy. In a micro-business, a short policy that is applied beats a ten-page document nobody reads.
A 30-day test with a single indicator (time saved) is enough to decide what comes next. Before starting, time the chosen task over two or three real runs. After four weeks, measure again.
An illustrative calculation, with assumptions to adapt: if 5 people save 2 hours a week over 46 weeks, that is 460 hours. At a loaded hourly cost of 30 EUR (assumption), the value is 13,800 EUR against 1,200 EUR of subscriptions. Such a result is never guaranteed; it depends on the task, the quality of review and real adoption by the team. If the measured gain is lower than the cost, that is useful information, not a failure. The full method is in calculating AI ROI in an SME; the term is also defined in the glossary: AI ROI.
Training can often be covered in whole or in part, but the rules change: check before signing. For a micro-business, the starting contact is your OPCO (the French sector training fund), which states the criteria, caps and priority topics currently in force. The FNE-Formation scheme has seen significant changes; its current status must be confirmed with your OPCO. The status update is in our article on FNE-Formation and AI, and the steps to follow are in the OPCO funding guide.
Two habits to keep: request coverage before the training starts, and choose a Qualiopi-certified provider, the usual condition for public and pooled training funding in France.
The most frequent mistakes are buying too many tools, training too late and measuring nothing. They appear in AI deployment mistakes in SMEs, with a few variants specific to small teams:
To locate your starting point before spending, the article on company AI maturity offers a simple grid, and the 5 steps of AI adoption give the logical next move.
Can a micro-business start with AI on a zero budget? Yes, to test: free versions help you understand the uses. They do not always suit sustained professional use, especially regarding data-handling terms. As soon as client documents are involved, move to a professional paid plan.
How much should I plan to start? For a serious test, count subscriptions for a few seats and a short training. Vendor prices change fast: ask for current prices and cost out 4 weeks before committing to an annual subscription.
Do I need to hire or appoint someone internally? No hiring at the start. Appoint a referent with dedicated time who tests, documents what works and trains the others. In a micro-business, half a day a week is enough during the test phase.
Does the AI Act apply to a micro-business? Regulation (EU) 2024/1689 applies according to the organization's role and uses, not size alone. Article 4 on AI literacy targets organizations using these systems. Short, traceable training is a good starting point; for details, see the AI Act guide for SMEs.
Which task should I pick first? The one that comes back every week, takes more than two hours, is easy to review and involves no sensitive data: quotes, minutes, letters, communication content.
When should I move to automation? When a manual use has proven its gain over 30 days and repeats identically. Automating too early freezes a poorly understood process.
GrowthPerf is a Qualiopi-certified training provider specialized in AI, no-code and automation for SMEs and associations. For a micro-business, we start from your real tasks: a short discussion to choose the first use case, hands-on training on your own documents, then a measurement review at 30 days. We also help build the funding application with your OPCO when it is possible.
For what comes next, go back to the guide to AI in business for SMEs, or see our AI training for micro-businesses and the operational AI program. To know whether an audit makes sense before you start, see the definition of an AI audit. Contact us about the Micro-business Pack: a free 30-minute first call to define your first use case.