AI Training
Your teams already use AI each in their own corner, and the energy leaks away. Three fixed-price tracks: the leadership team, the technical teams, middle management and the business teams.
In most of the companies we see, AI is already in the building. Someone uses it for meeting notes, someone else for code, and nobody brings it up in a meeting. Each person relearns alone what a colleague worked out last week, the practices never converge, and as much energy leaks away as gets produced.
So this is not a course on how to use ChatGPT. The subject is not the tool, it is the shared method that is missing: when to use it, when to leave it alone, how to phrase a request, what to verify, and the point at which a human re-reads and signs.
Leadership team. Five days, three of them on site, two seats, €15,000 excluding tax. Not a line of code: the work is what a leader actually spends the day doing. Seats can go to three or four, but that gets settled after the framing session, when the programme is written and it is clear who has to be in the room.
Technical teams. Eight days, four of them on site, four to six seats, €25,000 excluding tax, on your own repository. Follow-up is cut into two-hour blocks rather than whole days, because a team that ships cannot free up a full day in one go.
Middle management and business teams. Same format as the technical track — eight days, four on site, two-hour blocks, four to six seats — €29,000 excluding tax, on your own files rather than on code.
In all three cases the follow-up is used up within two months of the last on-site day: after that the momentum has gone and the sessions turn back into a conversation.
It is not a prompt catalogue. A list of formulas copied into a shared document ages in three months and teaches nobody anything.
It is not a conference with a demo either. Thirty people watching a screen produces enthusiasm on the day and no change at all on the Monday. We work in very small groups, on your real files, and we would rather run two cohorts than fill a room.
A shared, written method your teams replay without us: the cases where the tool genuinely saves time, how to phrase a request, the checks to run before anything is circulated, and the point where a human re-reads and puts their name to it.
The tool does not replace the craft. It removes the re-typing, the digging through archives and the summarising, which is to say the hours nobody ever wanted to bill.
- Leadership team: 5 days, 3 of them on site, 2 seats, not a line of code
- Technical teams: 8 days, 4 of them on site, 4 to 6 seats, on your own repository
- Middle management and business teams: 8 days, same format, 4 to 6 seats
- Extra leadership seats (3 or 4) settled after the framing session
- Follow-up used up within two months of the last on-site day
- Usage charter and access split decided with you
- Written report and slide deck delivered within 5 days
- Editor licenses (Cursor, Claude…) and travel sit outside the package
Prepare a board meeting in an hour instead of an evening, from your own documents.
Read a thirty-page memo and come out with the decisions it actually calls for.
Prepare a vendor decision: the questions to ask, the answers to cross-check.
Set the company usage charter, and be able to defend it in front of your teams.
Work on your own repository, not on a demo project.
Review a diff knowing what the tool understood and what it misread.
Write up an incident while it is still fresh.
Stop re-typing what already exists somewhere else in the company.
Pick a client file back up without re-reading all of it: history, commitments made, open points.
Write a specification the technical team can read without rewriting it.
Prepare a steering meeting from the real project material.
Work through the queue of repetitive requests without losing the week to it.
Before day 1, we find where the hours go
A framing session with you and a few short interviews to spot the tasks that genuinely cost time: re-typing, digging through archives, formatting, summarising. The programme is written after that diagnosis, not before.
The programme is written on your files
Your documents, your reports, your repository. Nothing is demonstrated on a sample dataset, so what comes out of the sessions is reusable as-is the next day.
The on-site days produce artifacts
Three for the leadership team, four for the technical and business tracks, held within a month so the momentum holds. Every day leaves something behind: written methods, checkpoints, the usage charter and the access split decided with you. Written report and slide deck delivered within five days.
Follow-up so you can replay it without us
Two days for the leadership team; two-hour blocks for the technical and business tracks, because a team that ships cannot free up a whole day. In all three cases the follow-up is used up within two months of the last on-site day. We check that the method holds once we are no longer in the room.
Prices are fixed packages, excluding tax, per track and per cohort. Two things sit outside: the editor licenses (Cursor, Claude and the rest), taken out in your own name and yours to keep, and travel costs for the on-site days.
The seat count is the one stated: two for the leadership team, four to six for the technical and business tracks. Extending the leadership track to three or four seats is possible, but it gets settled after the framing session, when the programme is written and it is clear who has to be in the room.
The follow-up is used up within two months of the last on-site day.
Since 2 February 2025, Article 4 of European regulation 2024/1689 on artificial intelligence requires providers and deployers of AI systems to take measures ensuring a sufficient level of AI literacy among the people who use those systems on their behalf. The text sets no programme, no duration and no level to reach: it is an obligation of means, to be sized against how the company actually uses these tools.
We turn it into an operational module, weighted towards the leadership track: which uses are allowed and which are not, which data never gets pasted anywhere, when a human re-read is mandatory, and who carries responsibility for what leaves the company. It ends with the usage charter, written with you.
We cite the regulation as a working framework. This is not legal advice and does not replace your own counsel.
French industrial group (multi-site): fractional CIO — weekly operations steering, infrastructure reliability, cloud and AI framing, board committees.
French leader in refurbished-smartphone rental: migration assessment based on reading the actual code, funded by AWS MAP, then a priced modernization path (managed AWS services, domain-by-domain migration).
How is this different from a generic ChatGPT training?
A generic course shows a tool on generic examples, and everyone leaves with notes they will never open again. Here the programme is written after looking at where your hours go, the exercises run on your own documents or your own repository, and what comes out is usable the next day. Two to six people depending on the track, not a room.
How much does AI training cost?
Three fixed-price tracks, excluding tax. Leadership team: €15,000 for five days and two seats. Technical teams: €25,000 for eight days and four to six seats. Middle management and business teams: €29,000 for eight days in the same format as the technical track, and four to six seats. Editor licenses and travel sit outside the package.
How are the days split?
The leadership track runs over five days: three on site, held within one month so the momentum holds, then two follow-up days. The technical and business tracks run over eight days: four on site, then follow-up cut into two-hour blocks. In all three cases the follow-up is used up within two months of the last on-site day.
We are not a technical team, is this for us?
Yes, twice over. The leadership track has not a line of code in it: it works on what a leader actually spends the day doing, preparing a board meeting, reading a long memo, preparing a vendor decision, setting the company usage charter. And the middle management and business track takes the format of the technical one, run on your own files rather than on code. The technical track itself runs on your own repository.
Why cap the number of seats?
Because dilution is the first cause of failure for this kind of training. Two seats on the leadership track, four to six on the technical and business tracks: beyond that the facilitator can no longer unblock each person on their own work. The leadership track can go to three or four seats, but that gets settled after the framing session, not at signature. We would rather run two cohorts.
Are AI licenses included in the price?
No. Editor licenses (Cursor, Claude and the rest) are taken out in your own name and stay yours: you keep control of the accounts and of what goes through them. Travel costs for the on-site days sit outside the package too. Access is needed during the days themselves, since the work runs on your own material.
What is left once the engagement ends?
A written report, the slide deck, and the methods recorded during the engagement, replayable by your teams. For a technical track, a shared meta-repository built with you and owned by you.
Does this replace people?
It is neither what we sell nor what we observe. The tool does not do the job: it removes the re-typing, the searching and the formatting. Judgement, trade-offs and the signature stay human, and nothing leaves the company without a person having re-read it.
Are we required to train our teams on AI?
Since 2 February 2025, Article 4 of European regulation 2024/1689 requires providers and deployers of AI systems to take measures ensuring a sufficient level of AI literacy among the people using them on their behalf. It is an obligation of means: no programme, duration or level is imposed. We cover it as a module of the training, as a working framework and not as legal advice.
Let's talk about your situation.
A 30-minute call with the founder, no strings attached. Reply within 24 business hours.
Your details are used only to answer your enquiry: never sold, never used for cold outreach. Privacy policy.