AI doesn't replace jobs.
It changes tasks.
Your executive committee expects gains; your teams and works council expect facts. Skillberg shows, job by job and task by task, what AI changes in your company, and gives you what you need to decide together.
30 minutes on video with Skillberg, no commitment. No personal data required.
The questions you get asked about AI, and what you'll be able to answer.
- The executive team, at an offsite
Where should AI start for us?
With the tasks AI can do, that teams want to hand over, and where the gain is worth the effort. The map of your company's jobs brings them to light, and the choice is prepared together, on a shared basis.
Choose where to start - The finance team, in an executive committee
How much time will AI really save us, and where?
An estimate job by job, calculated from your job descriptions and headcount. You know where every hour comes from, and every assumption can be replaced with yours.
Quantify the time returned - The person in charge of AI, before the committee
Our pilots work. Why does nothing scale?
Often, use cases were chosen from a demo, far from the teams' real work. Starting from the tasks teams themselves want to hand over to AI, adoption has a much better chance of following.
Scale up - A member of the works council, in session
Will AI cut our jobs?
What changes, what stays and what is learned, job by job, on facts everyone can check. The tool describes jobs and rates no one.
Talk about AI with the works council - Someone from accounting, at the coffee machine
Will my job still exist in three years?
Their job will evolve, and you can show them how: the tasks AI will lighten, those where their judgment remains essential, and the skills that will help them move forward.
Prepare workforce planning
AI is already at work in your teams. Rarely where you expect it.
Your executive committee wants a figure, the works council expects facts, and every week brings contradictory announcements. The most solid studies paint a more nuanced picture, and a far more useful one for acting. The studies, explained simply (in French).
of the companies studied have staff already using a personal AI tool for work, whether or not it was organized.
MIT NANDA, 2025 · preliminary report
of jobs see AI used for at least a quarter of their tasks. For three quarters of their tasks, that is the case for only 4% of them.
Anthropic Economic Index, 2025 · US data
Change happens below the surface.
A job title tells you neither which tasks AI can take on nor which skills are gaining value. Skillberg maps this hidden part, using its own model of 158,000 interlinked skills.
What the org chart shows: job titles and headcount per job.
What Skillberg reveals: the tasks in each job, the skills they draw on, what AI can take on, the time returned to teams, the skills to develop and the pathways between jobs.
It's at task level that you see what AI really changes.
Deciding role by role risks handing AI the very tasks through which your juniors learn their job. Take the payroll officer job: AI can enter pay data received by email, but supporting someone on long-term leave remains human work. A new task appears too: checking what AI has produced. Up close, every job becomes readable, and every choice can be discussed on concrete grounds. How Skillberg runs this analysis
From your job descriptions to shared decisions, in three steps.
- Step 1
You share your job descriptions
As they exist today, with headcount per job. No personal data: the analysis covers jobs, never people.
- Step 2
We analyze every job, task by task
Our model breaks each role down into tasks and skills, then assesses what AI can do there in light of the most robust studies. Two people on our team review every assessment.
- Step 3
Decisions are built together, tool in hand
We present the results to your executive committee, then the tool stays in your hands. Leadership, managers, teams and social partners rely on the same facts to decide what comes next.
It all starts with a pilot on a few jobs, to judge on the evidence before extending the approach. The pilot in detail · Where the figures come from
Skillberg Intelligence, in under two minutes.
Martin Vielvoye, founder of Skillberg, walks through the approach: from your job descriptions to the task-by-task diagnosis, then to your executive committee's decisions.
What does the tool look like? You set a limit, it calculates the rest.
Here is a fictional company of 210 people. By moving the slider, you choose how far AI can go: the time freed up, the skills to develop and the budget are recalculated instantly.
You set the limit. The tool never gives AI more than what was measured on your tasks.
Time returned to teams, not positions cut: the tool reasons in equivalents to reassign and train: the substance of your workforce planning (GEPP).
Explore the full demoFictional company, calculated with the tool's rules. The full demo in seven steps

Who's behind Skillberg?
« The AI shift is first and foremost a question of skills. And good decisions start with a good map. »
Alongside him, a research and engineering team based in Lille, building its own skills model since 2022. Finalist of the ManpowerGroup challenge at VivaTech 2026, among 200 startups.
Dr Axel Potier, AI researcher · Guillaume Brebion, Data Engineer · Nicolas Van-Duysen, AI engineer · Tristan Riboulet, AI engineer · Thomas Busin, AI engineer · Merveille Arikungoma, Web developer
Our team and our storyWhat people ask us before getting started.
Who sees the results?
The results describe jobs, never people: there is nothing personal to see. We present them to your executive committee, then the tool stays in your hands, and it's up to you to decide who to share it with, from managers to the works council.
Will our teams accept it?
More than you might think. According to Stanford University (Shao et al., 2026), which surveyed 1,500 workers in the United States, respondents are in favor of automating 46% of their tasks. The tool therefore starts from what your teams want to hand over to AI, gathered through anonymous questionnaires.
How much time do we need to ask of our teams?
Not much. To start, your job descriptions and headcount per job are enough: HR gathers them. Teams then answer an anonymous questionnaire on the tasks they would like to hand over to AI.
Where do the tool's figures come from?
First from our own model, the result of four years of research: a map of 158,000 skills linked to each other and to jobs, organized into skill trees, which our team has been building since 2022 from the major frameworks (France Travail's ROME, ESCO, O*NET). This model breaks every role in your company down into tasks and skills. We then apply the most robust public studies (Stanford, Anthropic, MIT, LaborIA from the French Ministry of Labour), and two people review every assessment. Your data completes the picture: job descriptions, headcount, time spent per task. When data is missing, we flag it, and every assumption can be replaced with yours. The details are on the Method page.
How much does a pilot cost, and how long does it take?
Price and duration depend on the number of jobs covered. The first 30-minute conversation lets us sketch that scope with you, with no commitment.
Thirty minutes to ask the right questions.
And know where to start.
30 minutes on video with Skillberg. You tell us about your situation, we show you the tool on a fictional company, then together we sketch what a pilot would produce for you. No commitment, no personal data required.
Not right now? Read the AI Act guide for CEOs and CHROs (in French) or write to contact@skillberg.app