France's Osez l'IA Plan: What AI Support Can Your SME Claim?

What is the Osez l'IA plan launched in July 2025?

Osez l'IA (literally 'dare to use AI') is a plan launched by the French government in July 2025, backed by 200 million euros to speed up AI adoption across businesses. The stated goal is ambitious: 80 % of SMEs using AI by 2030. The starting point is a familiar one: large companies are adopting AI fast, while many small and mid-sized firms stay on the sidelines for lack of time, skills or budget.

The plan is more than a headline: it relies on concrete schemes run by France Num and Bpifrance, built around three levers: diagnostics, training and funding. One point worth stating upfront: this article is an independent editorial guide. Neting is not an official ambassador of the plan, and you need no intermediary to access the support.

Who is the Osez l'IA plan for?

The plan primarily targets French micro-businesses and SMEs across every sector: retail, trades, services, industry, professional services. You do not need to be a tech company: the schemes are designed for firms that have not yet taken the leap. Exact criteria (headcount, age, location) vary from one scheme to another, so check each one before applying.

What concrete support is available: diagnostics, training, funding?

The support falls into three families, delivered mainly by France Num and Bpifrance.

The smart move is to start with the diagnostic. It stops you buying the wrong tool and often becomes the backbone of your funding application.

How do you check eligibility and build an application?

You can check eligibility directly on the France Num and Bpifrance websites, with no mandatory intermediary. A typical path takes five steps:

1. List the open schemes on francenum.gouv.fr and bpifrance.fr and note their access criteria.
2. Gather your basics: company registration number, headcount, key financials, a description of your activity.
3. Run a diagnostic to establish an objective starting point.
4. Scope a precise project with a deliverable, a timeline and a costed quote.
5. Submit the application online and allow for processing time.

An application lands better when it describes a concrete use case, such as automating replies to customer enquiries, rather than a vague ambition to 'do AI'.

Which AI use cases should an SME start with?

Start with high-volume repetitive tasks: customer replies, quotes and documents. Three scenarios consistently deliver:

• Automated customer replies: an AI messaging agent answers 24/7 on WhatsApp, Instagram or your website chat, qualifies the request and books appointments. Solutions such as Zazapp start at 44 euros per month.
• Quotes: AI drafts a quote from an incoming request, which you review before sending.
• Documents: sorting, extracting and summarising invoices, contracts or emails, saving hours every week.

To dig deeper into these scenarios, read our guide on automating your business with AI.

How long before you see a first result?

A well-scoped first AI use case can be in production within 2 to 6 weeks. The key is a tight perimeter: one process, one channel, measurable success criteria. By contrast, an 'AI everywhere' project with no clear priority stalls and demoralises teams. Start small, measure, then expand: that is also the logic behind the public schemes, which favour structured projects over fuzzy experiments.

How can Neting support you with these schemes?

Neting is not an official ambassador of the Osez l'IA plan: our role is that of an independent provider who knows the ground. Through the Boost IA programme, you get a free AI diagnostic, then a first use case in production within 2 to 6 weeks. We also help you build your France Num and Bpifrance applications: project scoping, costing, detailed quotes. With more than 20 years of expertise in development and project management, the team answers every quote request within 24 hours.

Which mistakes should you avoid before starting?

The first mistake is waiting for a grant before starting to think. Diagnostics and scoping cost little or nothing: do them now, and your application will only be stronger. Other common traps: picking a tool before defining the need, overlooking GDPR when handling customer data, and scoping too broadly. A single use case, delivered in a few weeks and measured, beats a transformation plan that never starts.