This weekend, one of your colleagues asked an AI where to go on vacation. Another had it compare two offers before buying. A third had it rephrase a message they couldn’t manage to write. None of the three had been trained.

AI seems to be settling quickly into our private lives. What about at work?

What AI has already changed in companies

Three things have become part of everyday life.

  • Searching has been replaced by asking. We no longer type keywords, we describe a situation. It’s the most mundane change, and the deepest.
  • Understanding no longer requires prior expertise. A quote, a contract, a user manual, an administrative form: deciphering is no longer an obstacle.
  • Getting started no longer costs anything. No more blank page, no more first sentence to find. You start, then you adjust.

At work, AI is still a better Google

Search and summaries: the two dominant uses of generative AI

91% of French employees say they use AI at work. But only 27% use it daily, and 3% in an advanced way, meaning they use several tools on complex tasks (EY Work Reimagined 2025, December 2025).

Now let’s look at what AI is actually used for: the two dominant uses at work are searching for information (49%) and summarizing documents (39%) (EY, 2025).

These are exactly the two things a search engine already did, only less well. A tool able to transform a job is being used as a slightly more polite search bar.

And it isn’t unwillingness. It’s an inherited reflex: with a search engine, you ask a short question and get a generic answer, because it knows neither your documents, nor your templates, nor your way of working. Your employees query AI with twenty years of these reflexes. Nobody has shown them anything else.

AI is Excel in the 1980s

The arrival of AI looks a lot like the arrival of Excel.

At first, most people used it to add up three numbers. Useful, but trivial. The ones who really got ahead were those who learned to write their own formulas, in other words to adapt the tool to their job. The sales rep’s tracking sheet, the financial controller’s model, the logistics manager’s macro: each of them shaped their own tool.

AI is exactly the same. Using it to ask vague questions is sticking to adding up three numbers. The real shift comes when everyone adapts it to their actual work: their tasks, their documents, their way of doing things.

The 3% of advanced users are the ones writing their own formulas. They can gain the equivalent of a day and a half of productivity per week (EY, 2025).

Employee AI adoption: what it changes day to day

Advanced uses: the end of blank-page syndrome

Writing a project brief, an interview guide or an internal announcement used to cost an hour before you even knew whether the idea held up. That start-up cost has disappeared.

What it concretely changes in a day: decisions are no longer made on intentions, but on drafts. An HR manager prepares three versions of an announcement about a reorganization and has the leadership team decide on actual text, not on a principle. The projects that kept being postponed for lack of time to get them started finally get going.

Productivity and AI: taking on the complex tasks nobody had time for

Dozens of useful tasks simply weren’t being done, for lack of time: rereading all the annual review reports, comparing three vendor offers line by line, analyzing the open comments of an employee survey. They become doable in an hour.

The time saved on existing tasks frees up room for the projects that kept getting pushed back for lack of bandwidth. A team that wins back five hours a week on its day-to-day work doesn’t spend five more hours doing the same thing. It finally opens the file that has been gathering dust for eighteen months.

The takeaway? AI doesn’t just save time on existing work, it makes possible what wasn’t being done.

How do you move from basic to advanced uses of AI at work?

This step can’t be decreed by the leadership team. It is built job by job, from each person’s actual work.

  • Start from your own tasks, never from demos. Showing what the tool can do is like giving a lecture on Excel functions: nobody remembers it. What works is starting from the tasks the person does every week and would rather stop doing alone. The anti-model is the catalog of use cases sent to the whole company, which nobody opens.
  • Give AI the context a search engine never had. Your documents, your templates, your way of writing. As long as you give it none of that, the result can only be generic, which explains why so many users are disappointed.
  • Capture what works. A useful Excel formula ended up in a shared file. A good instruction given to an AI stays in one person’s history. Then disappears. If advanced uses don’t spread, this is the first reason: there is nowhere to store them. The bare minimum: a folder, an owner, a regular review.

This is precisely the shift from throwaway prompts to a tool shaped for a job. To go further, read our article on the difference between an AI skill, an AI agent and an AI assistant.

Conclusion

Access to AI is no longer the issue. What remains to be done is building AI literacy: moving from individual, opportunistic uses to a collective practice, with the right tools, anchored in business processes.

You don’t reach this next level with one more license. You reach it with training anchored in real work, and the effect is clear: among trained employees, 90% say they use AI, compared with 47% on average. Today, only 23% of employees have been trained (KPMG / ViaVoice, June 2026).

Your teams don’t need to be convinced. They need to learn to write their own formulas.

FAQ: your questions about everyday AI at work

How do employees use AI at work?

91% of French employees say they use AI at work, but only 27% do so daily, and 3% in an advanced way, using several tools on complex tasks (EY Work Reimagined 2025). The two dominant uses are searching for information (49%) and summarizing documents (39%): AI is still mostly used as an improved search engine.

What does AI change in daily work?

It removes the cost of getting started: no more blank page, and decisions are made on drafts rather than on intentions. It makes doable the useful tasks that were never done for lack of time, such as rereading all the annual review reports or analyzing the open comments of an employee survey. Advanced users can gain the equivalent of a day and a half of productivity per week (EY, 2025).

How do you move from basic to advanced use of AI?

This step is taken job by job, starting from the real work. Start from the tasks each person does every week rather than from demos, give AI the context a search engine never had (your documents, your templates, your way of writing), and store what works in a shared place, with an owner and a regular review. Training anchored in real work makes the difference: 90% of trained employees use AI, compared with 47% on average (KPMG / ViaVoice, June 2026).