They count
Those of Copilot or ChatGPT count logins and active licences.
Event · Adopt AI 2026
December 3 and 4 at the Grand Palais, on the French Tech Grand Paris pavilion. We present the audit of the real use of AI, run with our own software, Idun Cockpit: who uses Microsoft 365 Copilot, ChatGPT or Claude and for what, which licences are paid for and unused, and what each use costs and returns, measured on the company's own data.

Find us on the pavilion
Not coming to the show? The same meeting by video call
At the stand, the Idun Cockpit demo shows what the AI tools produced and cost over the period, adoption by team and where it drops off, the value and cost of each use, and the recommended workstreams.
Those of Copilot or ChatGPT count logins and active licences.
The use cases that recur from one team to the next, the ones nobody had declared, the teams where usage drops and what makes it drop, the assistants built twice.
We start from the usage exports your tools already produce, with nothing to install and no code written on your side, and we interview your teams: the interviews bring what the logs do not say, the intent and the context. Results come within 15 days.
What your teams really do with AI.
What each use costs and returns, with paid licences set against real usage.
What to scale first, what to stop, which teams to train and on what.
Your data stays in France, hosted by Scaleway in Paris, or on your own infrastructure if you prefer, and no one is ever named in the results.
Copilot, ChatGPT, Claude or their own assistants, without being able to say what it brings them. The question often comes up before a licence renewal or a budget review.
critical use cases uncovered by the audit
SNCF Immobilierreal activation: 33 users out of 39 paid seats
InskipAI diagnostics delivered in a year
Idun GroupWe show you Idun Cockpit.
We look with you at what your tools already export.
You leave with what it would take to get the same reading in your company.
After the show, we send it to you, built on a fictional company close to yours.
CEO and co-founder, formerly at Airbus.
CPO and co-founder, five years at LVMH, two of them on the generative AI platform of its perfume and cosmetics brands.
Business development.
The programme gives one CEO Stage track to measuring impact: "AI & Performance — Measuring impact, proving value". It starts from three questions: why individual productivity gains do not show up in the P&L, how to define AI success at the level of the organisation, and who in the company answers for its return on investment.
Times and access conditions are in the official programme
Five questions to ask at any stand
Can you tell who uses AI and for what, or only who logged in?
The admin consoles, in partHow many paid licences go unused, and in which teams?
The admin consoles, in partWhich use cases exist in our company without anyone having declared them?
A reading of real usageWho should be trained, and on what, given how people use AI today?
A reading of real usageDoes what we deployed return what we expected? If not, what is in the way?
A reading of real usageIt is built on the company's real usage. We measure it in time and translate it into cost, at the company's own rate. No gain is promised before it has been measured.
Spent on each task, with and without AI.
Matched to specific use cases.
Those of the leadership team, to know what each use serves.
An AI error, set against the human error rate on the same task.
What changing nothing costs.

Pick a thirty-minute slot on December 3 or 4.