We do not automate to take over jobs. We automate because we cannot find anyone.

There is a sentence I have never once heard in a meeting. Not once. It is "I would like to automate so I can let someone go." What I hear instead is always the same thing, in different words: "I cannot manage to recruit", "my secretary is drowning", "I turned down three jobs last month for lack of hands". The public debate talks about machines taking jobs. In the workshops where I spend my days, the problem is exactly the reverse: there is work, and there is no one to do it.

Is the labour shortage a feeling or a fact?

It is a fact, and it is measured every year. The 2026 Labour Needs survey (BMO) by France Travail, the French national employment service, records 2.275 million recruitment plans in France, and 43.8% of them are judged difficult by employers themselves. Almost one recruitment in two. That figure is down 6.3 points on 2025, and I would rather say so straight away than hide it: the pressure is easing a little. But when you look at the trades I am talking about here, the easing is theoretical.

In construction, the Construction Sector Skills Observatory uses the same data and arrives at a 65% rate of recruitments judged difficult, out of nearly 140,000 plans. Two hires in three that get stuck. In detail, 9,540 plumbers and heating engineers are being sought this year. Not 9,540 jobs threatened by a machine. Nine thousand five hundred and forty people being looked for and not found.

People do not call me because there are too many people. They call me because there is no one.

What does that look like from inside a small business?

It looks like a woman staying late in the evening to enter invoices that no one else knows how to enter. A boss doing his paperwork on Sunday because the week is taken up by jobs on site. A handful-strong company that has been looking for an admin hire for months without finding one. Meanwhile, the work does not disappear. It gets spread across the people who are already there, and the people who are already there are overwhelmed.

Here is what actually happens when I install something at these places. The plumber whose day I described went from two hours of paperwork every evening to a quarter of an hour a week. He did not lay anyone off. There was no one to lay off: he was on his own. What he got back was his evenings. At the garage where one call in three used to go unanswered, the smart phone system did not take over any job, because that job never existed and never had. The machine does not take someone's place. It fills an empty one.

Will AI never cut a single job?

No, and I am not going to spin you a story to sell you something. There are trades where the repeatable share pays the bills today, and those are going to feel it. I wrote a whole article on this, about what really decides whether your job is exposed, and its answer is not reassuring everywhere. In large corporations, with entire departments dedicated to repeatable output, the question of jobs being taken over is a real one. I do not work there. I work with businesses of two to twenty people, where every person does four jobs at once, and where the only thing that exists in excess is the work itself.

There is an honesty I owe myself in all this: if tomorrow an owner-manager calls me saying they want to automate to get rid of their assistant, I will know what to answer, and it will not be a quote. This is not cheap moralising, it is my working line. A small business that sacks its admin person for a robot ends up with a robot that cannot call back the worried customer on a Sunday evening, and no one left to do it in its place.

That is why everything I do at AlphaCor starts from the same place: time, not headcount. I look at where the hours of a team too small for the load it carries are going, and I give some of them back. Here, AI replaces no one. It gives time back to the professionals who lack it.