Your maturity level
You find out where your company stands on a five-step scale, from the very first trials to AI that truly serves your strategy.
AI Assessments
What is your company’s AI maturity level?
Your teams may already be using AI, but do you really know how, with which tools and with what results? Rest assured, most of the business leaders we meet ask themselves the same question. To answer it, we designed a free 20-question assessment, based on situations you face every day. It places your company on five levels of AI maturity, dimension by dimension, and tells you where to start.
Where are you on the AI maturity curve?
A 20-question evidence-based diagnostic mapped to the five stages of AI maturity — Experimentation, Adoption, Adaptation, Optimisation, Innovation — across the five dimensions every organization needs to move forward in. Each question asks about observable behavior, not self-perception.
How to answer: Pick the option that best describes what would actually happen today — not what's on a slide deck or what you'd like to be true. The diagnostic is most useful when answers reflect observable reality.
- People
- Process
- Data
- Tools
- Strategy
- 20 questions
- 5 dimensions · 4 each
- ~8 minutes
- Anchored in real scenarios
- Stage 1 → 5
- Per dimension + overall
- Save & share
- PDF or copyable summary
All answers stay in your browser. Refreshing keeps your progress.
Adapted from Spark AI Maturity Model™ (All rights reserved)
The result
What the assessment reveals
Your strengths and your blockers
The assessment looks at your company from five angles: your teams, your ways of working, your data, your tools and your strategy. You see at a glance what moves you forward, and what holds you back.
Your next step
A clear, realistic priority, adapted to your situation, to share with your team so you can move forward in the coming weeks.
Definition
What is a company’s AI maturity?
AI maturity describes how your company has taken ownership of artificial intelligence. It depends less on the tools you have bought than on how you use them: who uses them, for what, with which rules, on which data, and with what results.
In many companies, AI arrived unannounced. Nearly one in two French people now use generative AI, just three years after it appeared[1], while only 18% of companies with 10 or more employees use at least one AI technology[2]. Your employees have therefore often started before you, each in their own way. Measuring your maturity lets you take stock of these uses, give them a framework and invest where they will really pay off.
The 5 levels
The 5 levels of AI maturity, from first trial to innovation
Our assessment is based on five levels, adapted from the Spark AI Maturity Model™. Don’t see your result as a grade: each level is a normal stage of the journey, one we have seen many companies get through. For each, here is how to recognise it, what to watch out for, and the priority we recommend.
Level 1. Experimentation
A few curious employees try out AI tools on their own initiative, often in their free version. There is still no common rule, budget or shared goal. It is a very good start, but your data may already be circulating in tools nobody controls.
Our advice: take stock of what is already being used, set a first simple rule on data, then choose two concrete use cases together.
Level 2. Adoption
The company provides tools and several teams use them. But use remains uneven and often relies on a few motivated people. The day they change roles, progress stops, and sometimes a subscription remains that nobody dares cancel, just in case someone uses it.
Our advice: train your teams on their real tasks, and appoint a champion in each department to keep good practices alive.
Level 3. Adaptation
AI is starting to change the way you work: a workflow is automated, a tool is connected to your data, the first gains are measured. This is often when it is tempting to multiply pilot projects without ever rolling them out across the company.
Our advice: measure the results precisely, secure your data, then extend what works to other teams.
Level 4. Optimisation
Several uses are part of everyday work, tracked by indicators and governed by rules everyone knows. Your company has done the hardest part. Now you need to keep an eye on costs, which can creep up quietly, and on tools that end up overlapping.
Our advice: manage your uses as a coherent whole, with their costs, results and priorities.
Level 5. Innovation
AI is part of your strategy: it improves your offers, creates new services or changes the way you serve your customers. At this level, the main danger is letting vigilance on security and compliance slip as uses multiply.
Our advice: keep exploring new uses, while always staying in control of the risks.
The 5 dimensions
The 5 dimensions to assess
Within the same company, AI does not move forward at the same pace everywhere. Your teams may well use powerful tools while your data is still scattered across several software products. The weakest point then holds back everything else, and that is generally where we advise you to start. Here are the five dimensions we look at in the assessment, each with a question to ask yourself today.
| Dimension | What it measures | The question to ask yourself |
|---|---|---|
| People | Your teams’ skills, their willingness to use AI and the support provided | Would a new employee know, in their first week, which AI tool to use and how? |
| Process | The real place of AI in the way you work | If your AI tools were removed tomorrow, which processes would really be disrupted? |
| Data | The reliability, accessibility and protection of your information | Is your customer data stored in one place, up to date and protected? |
| Tools | How well the tools provided fit, are known and are actually used | Do you know which AI tools your teams really use, free ones included? |
| Strategy | The link between AI and your goals, and who owns the topic | Who in your company decides AI priorities? |
Method
How do you assess your company’s AI maturity?
There is no need to spend weeks on it. A reliable assessment mainly takes a little honesty with yourself, and the view of several people. Here is the approach we recommend to our clients.
- Start by taking the assessment, thinking about what really happens in your company today. The questions start from concrete situations, which helps you stay close to reality.
- Then offer it to other members of your leadership team, each separately, and compare your results. Surprises are frequent, and often instructive: it is not unusual for management to see AI everywhere, while teams are still looking for it.
- Take the time to check your answers against concrete evidence, such as a tool that is actually used, a written rule or a measured gain.
- Identify the dimension that lags furthest behind, and make it your first priority.
- Retake the assessment a few months after each major step, then at least once a year, to measure your progress with the same method.
In France
Where do French companies stand?
Do you feel you are moving more slowly than you hoped? You are far from alone. AI use is growing in France, but very unevenly depending on company size[2]:
| Company size | Share using at least one AI technology in 2025 |
|---|---|
| 10 to 49 employees | 15% |
| 50 to 249 employees | 31%, compared with 15% a year earlier |
| 250 employees or more | 58% |
Source: Insee, companies with 10 or more employees, excluding the agricultural, financial and insurance sectors.
The most recent figures confirm this acceleration: in 2026, 40% of very small businesses and SMEs use AI, 14 points more than a year earlier, and as many as 53% of SMEs do[3]. Only 19% of them, however, pay for a solution. Many uses therefore still rely on free tools, chosen by each person on their own, something we see very often in the field.
The blockers
What holds companies back
When companies are asked what holds back their use of AI, the same answers often come up:
22%
consider themselves able to make proper use of their data
Very small businesses, SMEs and mid-sized companies surveyed · Observatoire data et IA[4]
In the companies we support, we often find these three blockers. First, it is hard to see what AI could be used for day to day. Then, once its value is understood, teams lack the time or skills to get started. Finally, when a project starts, data is often scattered across several tools, or incomplete. Most French companies are therefore still at maturity levels 1 or 2. These blockers are not inevitable, though. Once clearly identified, they are often lifted within a few months, and the assessment helps you spot the one that affects you most.
In the field
The obstacles that slow your progress
Throughout our projects, we often come across the same difficulties, whatever the size of the company. Perhaps you will recognise some of them:
- Employees each using their own free AI tool, without the company knowing which one or for what
- A pilot project that convinced everyone in a meeting, and has remained a test for months, for lack of an owner or a budget
- Subscriptions paid every month, without anyone really knowing who uses them
- Promising initiatives driven by a single person, which stop the day they change roles
Some of these obstacles also raise a real security issue: a contract pasted into a consumer assistant or a customer file analysed in an unknown tool means data leaving the company without anyone knowing. A tool generated in a few hours with AI, then put online without checking who can access the information, also becomes an easy target, all the more so as attackers now use AI to find these flaws.
90 days
Moving up a level in 90 days
Knowing your level is a first step. The next is to move forward gradually, in the right order and with your teams. To help you get started, here is a possible roadmap over three months.
| Period | What you do | What you get |
|---|---|---|
| The first month | You appoint someone to own the topic, write a simple rule on data use and choose two concrete use cases | A shared direction and governed uses |
| The second month | You provide a suitable tool to the teams concerned, train them on their real tasks and appoint a champion in each department | Uses that spread beyond a few volunteers |
| The third month | You measure the time saved on the two use cases, extend what works and retake the assessment | Concrete results, and a clear view of the next step |
A point to watch if your company has at least 50 employees: remember to inform and consult your CSE (the French works council) before rolling out a new tool. Several courts have already suspended AI projects launched without this step[5], and involving your staff representatives from the start also makes it easier to get everyone on board.
Your turn
Your turn! Where does your company stand?
8 minutes is the time it takes to drink a coffee, and it is all you need to find out where your company stands and where to focus your efforts in the coming months. You can answer alone, or offer the assessment to several members of your team to compare your views. And if you would like to talk it over, one of our experts can take the time to go through your results with you, in a free 30-minute conversation.
Free and no sign-up · Your answers stay in your browser · A result to export as PDF or copy, to discuss with your team
FAQ
Everything you need to know about AI maturity
Can’t find your question?
Go through your results with one of our experts, in a free 30-minute conversation.
Most maturity models, in AI as elsewhere, are inspired by CMMI, a framework originally created to assess software development practices. It distinguishes five levels: Initial, Managed, Defined, Quantitatively Managed and Optimising. AI maturity models follow the same logic: you start from improvised practices and move, step by step, towards organised practices that are measured and then continuously improved.
This question often refers to the scale presented by OpenAI in 2024 to measure the progress of AI itself[6]:
- chatbots, able to converse in natural language,
- reasoners, able to solve problems like a human,
- agents, able to act on the user’s behalf,
- innovators, able to help invent,
- organisations, able to do the work of an entire organisation.
Be careful not to confuse the two: this scale describes the capabilities of the technology, whereas AI maturity describes how your company uses it.
Four types of AI are often distinguished:
- reactive machines, which respond to a situation with no memory of the past, like the first chess programs,
- limited-memory AI, which relies on learned or recent data, like today’s conversational assistants,
- theory of mind, an AI that would understand human emotions and intentions, still at the research stage,
- self-awareness, an AI conscious of itself, which today belongs to science fiction.
Young adults. Among generative AI users, 51% of 18-24-year-olds use it every day, compared with 27% of 40-59-year-olds and 17% of over-60s[7]. For a company, this is a signal worth considering: your young recruits often arrive with well-established habits. It is best to support them from day one, and why not rely on their ease to bring the rest of the team along.
Several free assessments now let you take stock of AI use in your company. They do not all measure the same thing, nor take the same time:
| Assessment | Format | What you get |
|---|---|---|
| Kreante | 20 questions based on concrete situations, about 8 minutes, no sign-up | Your level on 5 stages, overall and for 5 dimensions, with a priority and a result you can export as PDF |
| Silkhom | Express 10-question test, about 3 minutes | An immediate score and first ideas for action |
| Tensoria | 12 questions, about 5 minutes, no sign-up | A level out of 4, a score on 4 axes and 3 recommended actions |
| Skillevos | Questionnaire of about 10 minutes | A score out of 100, a level out of 5 and a chart on 6 dimensions, with a full report sent by email |
| Koneetiv | Online diagnostic of about 3 minutes | A score on 6 axes |
| Observatoire de la maturité data et IA | A few minutes’ questionnaire, part of an annual survey | One of 5 profiles, on your data and AI maturity |
| Anand Candassamy | Grid of 20 closed questions, to fill in yourself | A level out of 4, depending on the number of positive answers |
To go further than an online assessment, the French state also offers, as part of the “Osez l’IA” plan, a data and AI diagnostic carried out by a Bpifrance-approved expert over 8 days. It is 40% funded by France 2030, leaving a cost of €6,000 excl. VAT for the company[8].
None of these assessments is official. Public frameworks, such as NIST’s AI risk management framework[9] or the ISO/IEC 42001 standard[10], define good practices but do not offer a maturity scale. Our advice: choose the assessment that speaks to you most, then stick with it from one evaluation to the next to measure your progress on a comparable basis.
Data maturity measures how a company collects, organises and uses its data. AI maturity measures how it uses artificial intelligence. The two are closely linked, because an AI is only as good as the data it relies on. It is a situation we often come across: recent AI tools, but scattered or incomplete data that keeps them from delivering their full potential.
Yes. Maturity does not depend on the size of the company or the amount invested, but on how AI is used and governed. A 20-employee company can perfectly well reach level 4 with a few well-chosen uses, clear rules and tracked results. Its small size is often even an advantage: decisions are made faster, and everyone quickly sees the benefits.
No. The assessment is a self-evaluation: it gives you a first snapshot of your situation based on your answers. An audit goes further, with interviews with your teams and analysis of your tools and data. The assessment remains an excellent starting point, and helps you know whether an audit would be useful to you, and on which topics.
If your company has at least 50 employees, remember to also consult your CSE (the French works council) before rolling out a new tool[5].
Not necessarily, especially at the first levels. Your teams know your business better than anyone, and up to level 3, we advise training them on their real tasks and appointing internal champions instead. A specialist profile, in data or AI, becomes useful when several uses work day to day and need to be managed, connected to each other and kept under cost control.
The European regulation on artificial intelligence requires companies that use AI systems to take measures so that their staff have a sufficient level of AI literacy[11]. Since its revision in 2026, this is an obligation of means: no precise level is required for each employee, but the company must act, according to each person’s use[12]. Your result on the People dimension is a good starting point for defining these actions.
Sources
- [1]Arcep, Baromètre du numérique, 2026 edition (presentation, in French)
- [2]Insee, Les technologies de l’information et de la communication dans les entreprises en 2025 (in French)
- [3]French Directorate-General for Enterprise, Baromètre France Num 2026 (in French)
- [4]Observatoire de la maturité data et IA des entreprises (in French)
- [5]Éditions Tissot, Déployer un outil IA en entreprise : faut-il consulter le CSE ? (in French)
- [6]Axios, OpenAI nears “reasoning”-capable AI, 15 July 2024
- [7]Arcep, Baromètre du numérique, 2026 edition (infographic, in French)
- [8]French Directorate-General for Enterprise, Le plan national Osez l’IA (in French)
- [9]NIST, AI Risk Management Framework
- [10]ISO, ISO/IEC 42001: Artificial intelligence management systems
- [11]EUR-Lex, Regulation (EU) 2024/1689 on artificial intelligence, Article 4
- [12]EUR-Lex, Regulation (EU) 2026/1744 amending the Artificial Intelligence Act
Written by Jorge del Carpio, CEO and co-founder of Kreante. Since it was founded in 2020, the agency has delivered 265+ projects in 35 countries.
Published Updated