Behind the technological enthusiasm around AI, a more fundamental question emerges, particularly in SMEs and mid-sized companies: is the company really ready to exploit data and AI?
In just a few months, artificial intelligence has established itself as a strategic subject in companies. Automation, process optimization, conversational assistants, AI agents: the promises are numerous and the use cases are multiplying.
But behind the technological enthusiasm, a more fundamental question emerges, particularly in SMEs and mid-sized companies: is the company really ready to exploit data and AI?
For several years, we have supported and analyzed hundreds of organizations in the field. And the observation is always the same: Data & AI maturity cannot be decreed. It is measured… and above all, it is built over time.
Data maturity still low in SMEs
Data from the Business Data & AI Maturity Observatory highlights a real gap between the “dream” of performance expressed by SME/ETI managers and the reality of taking action, still hampered by a lack of structuring and resources.
According to the results of the survey conducted in 2025, 88% of managers consider data to be a key performance lever, but only 19% say they are really capable of exploiting it.
At the same time, almost one in two companies continues to manage their activity using Excel or Google Sheets, and the obstacles encountered are, unsurprisingly, still the same:
- a lack of organization
- a skills gap
- a lack of knowledge
- a lack of strategic vision
In other words, managers are well aware of the potential of data and AI to improve their performance.
Even though I still see a lot of companies that still start from the tool, managers often start from their business challenges or expected results, and that’s a good thing.
But what is missing is the mastery of the prerequisites: those which make it possible to structure the sustainable exploitation of data and AI in the service of the company.
Without these foundations, initiatives move forward… but struggle to sustain themselves over time.
A generation of leaders at a crossroads
On the ground, the observation is almost always the same.
The managers of SMEs and mid-sized companies are today faced with structuring trade-offs, in an uncertain and constantly changing environment.
It’s difficult to decide and plan for the long term when current events push us to question our choices… almost every month.
On the one hand:
- a strong desire to accelerate on AI
- the fear of missing the train
- an immediate competitiveness issue
On the other:
- questions of digital sovereignty
- cybersecurity issues
- regulatory uncertainties
- environmental and societal concerns
This is exactly what we observe on a daily basis alongside the SMEs/ETIs that we support.
Leaders move forward, test, sometimes urgently… but without always having the necessary benchmarks to structure their approach.
Result: pressure to act, in an environment where everything is accelerating, but where frameworks still need to be built.
2025: the year of tests… sometimes in disorder
Faced with this pressure, many companies have chosen to act quickly.
2025 has clearly been a year of experimentation.
Nearly one in three companies say they have already implemented an AI solution, and one in two plan to do so between 2025 and 2026.
But in reality, what we mainly observe is:
- the purchase and deployment of licenses (ChatGPT, Copilot, Mistral, Gemini, etc.) for employees
- individual uses
- little overall strategy
We are not yet talking about organizational transformation. We are mainly talking about experimentation, and the establishment of an initial framework aimed at supporting the current dynamic: not slowing down the expectations of employees, nor the “desires to be” of managers.
The classic trap: start with the tool
This is undoubtedly the most common error.
Businesses start with the tools…
even before having clarified:
- their business challenges
- their priorities
- their level of maturity
Result :
- scattered uses
- teams that advance in a heterogeneous manner
- human brakes that appear
- and a passage to the ladder which blocks
AI works…until it’s no longer enough
First of all, the results are there.
Companies save time on:
- content writing
- meeting minutes
- document analysis (emails, PDFs, orders, etc.)
But as soon as we seek to industrialize or create lasting value, a limit appears.
The real subject is not AI… it’s data
Indeed, it is often at this moment of “truth” that companies are overtaken by reality.
In the field, the observation comes up regularly: after having tested and experimented, they quickly come up against the quality of their data, their difficulty of exploitation, and above all the impossibility of scaling up.
What works in test mode does not hold up over time, nor on a company scale. And it is precisely at this moment that everything is at stake.
We can’t do AI without data. And we don’t create data without having thought of the foundations.
“Try to build a house without having a good architect’s plan and foundations in place, you may regret it! »
Without structuring, projects remain limited.
AI acts above all as an amplifier.
Used well, on solid foundations, it increases performance tenfold.
But without prior structuring, it only amplifies existing weaknesses.
Three levels of use… and a lot of confusion
In the field, we distinguish three levels of use of AI:
- Individual AI
- Business/project AI
- Organizational AI
Today, the majority of SMEs and mid-sized companies still operate between the first two levels.
Very few companies have actually undertaken a global transformation. And this is not a problem in itself: a phase of experimentation and learning is necessary.
But this phase cannot last forever. There comes a time when the company must accelerate, structure its approach and provide clear answers to its employees, who are often skeptical about choices perceived as uncertain, or even potentially risky for their jobs or the future of the organization.
The real obstacles are human, not technological
Contrary to popular belief, blockages are not technical.
The main obstacles are organizational, cultural and fundamentally human.
With very concrete questions:
- Who is managing the subject internally?
- what skills do we need?
- how to get teams on board?
Without forgetting broader issues:
- What will our environmental impact be?
- What is the company’s social responsibility?
- Do we want to create jobs or reduce our workforce?
- Ultimately, what is the place of humans in the organization?
2026: slow down to speed up
After a year 2025 marked by experimentation, a new phase is opening.
2026 could well be the year of structuring.
With much less “we need to do AI” and more “where can we create value?” and “how to structure all of this sustainably?”
The companies that succeed will not be the ones that move the fastest. But those who will have been the most pragmatic and methodical.
A simple question… but essential
The question is no longer whether to do AI.
But rather: “Are you building a transformation… or just stacking tools?” »
Before wanting to accelerate, you still need to know where you are.
My advice: start by taking stock of the situation.
Clarify your use cases, your business challenges and your priorities.
Lay solid foundations, train your teams, then accelerate in a controlled manner.
