Many leaders want to start with AI but hesitate to pay for a diagnostic out of pocket. Bpifrance, the French public investment bank, offers support schemes designed to remove this first barrier, notably around data and AI (often referred to as Diag Data IA). This article explains the spirit of the scheme and the steps to follow, without ever replacing the official conditions.

Rates, amounts and eligibility criteria change regularly. Treat all figures below as indicative and verify the exact state of the scheme on bpifrance.fr before applying.

What is the Diag Data IA?

In principle, it is a short engagement: an approved expert spends a few days in your company to assess your data and AI maturity, identify relevant use cases and draw up a roadmap. Part of the cost of this work is covered, with the remainder borne by the company. Bpifrance aims to kick-start the journey, not to fund the full AI project.

Who is eligible?

The scheme mainly targets SMBs. An important recent point: indicatively, mid-cap companies (ETI) are no longer eligible since 2026, as the scope has been refocused on SMBs. Criteria generally cover company size, age and financial situation. These conditions change: do not rely on an article, verify your eligibility directly on bpifrance.fr or with an advisor.

  • Main target: SMBs (mid-caps are no longer eligible since 2026, indicatively).
  • Usual criteria: headcount and turnover within SMB thresholds, company registered in France, sound financial situation.
  • To check case by case: minimum age, sector, and compliance with social and tax obligations.

What it funds (and what it does not)

The scheme funds the diagnostic and the roadmap, that is, the scoping phase. Indicatively, the support covers a significant share of the cost of the engagement, up to a ceiling, with the balance paid by you. What it generally does not fund: building the AI solution itself, software subscriptions and maintenance. The exact percentage and ceiling must be checked on the official page, as they are subject to change.

A funded diagnostic gives you something valuable: an external, costed view before you spend. That is exactly the moment to demand an ROI calculation, not a list of promises.

How to apply

The process is online. Broadly: you create an account on the Bpifrance platform, check your eligibility, submit your request, then the engagement is carried out by an approved provider. Prepare a clear description of your business, your processes and what you expect from AI in advance: the more precise your request, the more useful the diagnostic.

  • Check eligibility and the current existence of the scheme on bpifrance.fr.
  • Create an account and build the file (company overview, challenges, objectives).
  • Submit the request and wait for validation.
  • Run the diagnostic with the approved expert and receive the roadmap.
  • Decide, based on the diagnostic, which projects to launch and in what order.

Why an ROI diagnostic first

A funding scheme does not replace economic common sense. The risk of a free or subsidised diagnostic is to produce a nice report with no follow-up. For it to be useful, demand a costed output: priority use cases, estimated gain, indicative cost and payback. This is our conviction: we challenge the project before we build it. An AI diagnostic is only worth it if it tells you clearly where the return on investment is, and where it is not.

In short

The AI diagnostic funded by Bpifrance (Diag Data IA) is a good way for an SMB to remove the cost barrier of the first step. Keep two reflexes: verify the exact, current conditions on bpifrance.fr, as eligibility and rates change (mid-caps are no longer eligible since 2026); and demand an ROI-oriented diagnostic that leads to costed projects rather than intentions.