Lycée Athena
Four days on your real processes. Not a case study: a system built by them, on their data, which stays in the company.
Four days · seven to eight managers per edition · at your premises
Every week a new tool arrives, a demonstration, an experiment started somewhere in the company. Management follows from a distance and approves the spending.
Then comes the moment to decide: which process to change first, where the person keeps control, who answers if the system gets it wrong. The most frequent answer is that they don’t know.
They already use AI. They don’t govern it.
Where it gets stuck
Mapping a process so that an AI system can carry it is a different craft from using an online assistant. Almost nobody has ever taken them through doing it on a real case.
At which point in the process a person must decide, who signs, who answers if the system gets it wrong. Management must have the answer, not the technical department.
The hours that are freed up are counted before spending, not after. Without that calculation every AI project becomes an act of faith.
The method
Each day delivers a result, and that result is the raw material of the next day. Anyone who skips a step notices straight away.
| Day | What happens in the classroom | What comes out |
|---|---|---|
| Day 1 Understand | The process is mapped with the people who do it every day: phases, roles, exceptions, critical points, data coming in and data going out. Analogue work, on the wall. | The process map. It is the backbone of everything that comes after. |
| Day 2 Build | The map becomes a system that runs: procedures, agents, data, connections. The team corrects and confirms piece by piece, and sees what changes with each correction. | A system that really runs, not a presentation of what could be done. |
| Day 3 Take ownership | The team rewrites the procedures in its own words, decides where a person’s approval is needed and runs the process from start to finish. The errors to be corrected are prepared by us, they don’t happen at random. | The operating procedures and the human approval points, chosen by them. |
| Day 4 Demonstrate | Tests on the difficult cases, numbers before and after, dashboard. Then the presentation to whoever in the company will have to say yes. Not a presentation: an acceptance test. | The document with which you ask to go into production, numbers included. |
The team builds the system. We stand alongside.
What remains
Two things come out: a documented and tested system, and the papers to show how it was decided. They are the same tools we use in systems in production.
Two editions
First edition
The managers brought their own processes into the classroom, not a textbook case. In four days they built the systems themselves, one per process, with the human approval points decided by them.
In July they presented them to top management. No objections to the method, and the priority set was to take the lab’s two processes into production.
The request came from them: make the lab recurring, not a one-off event.
From the lab to production
Contract hot-dip galvanising. The real problems came into the classroom: knowing where an order stands without having to ask on the phone, and bringing order to the production schedules.
What was born in the lab then went into production. The in-house technical lead took charge of it, with us alongside and not in their place.
Management learns to direct, the technical lead learns to build.
How it works
Above that number the discussion stops working and the classroom goes back to being an audience. The four days are held at your premises or at a network partner, and the schedule is agreed: consecutive, or one a week.
The process is yours. Which data you work on is decided beforehand.
The filter
Three questions
It is the most frequent objection, and whoever raises it is right. What remains is not the tool: it is knowing how to design the process and knowing what it takes to run it at your end. The first thing you discover in the classroom is precisely that list, and it comes before any spending.
No. In the classroom you direct agents and procedures, you don’t write code. Knowing the process counts for more than knowing the technology: that is why managers go into the classroom.
The document that comes out of the fourth day is already the specification of a project: what is needed, who needs to be involved, in what order. Whoever decides to take it into production goes through Metis, with the same people inside. Whoever decides nothing still has a system in hand and a team that knows how to read it.
The commitment
Fee per participant, calibrated on how many people come into the classroom.
The quote is made after the call: it depends on how many processes you bring into the classroom and how many people follow them. In the call we also decide whether the lab is the right thing now, or whether it is better to start somewhere else.
Thirty minutes to work out which processes to bring into the classroom, how many people to involve and when to do it. If the lab isn’t the right thing for you now, we tell you.