What we measure
For each team: how many people use their assistant each week, and the time spent on one or two tasks chosen with you, measured at the start, then 30 days after onboarding.
ChatGPT, Claude, Copilot or Le Chat, tuned to the work of each of your teams, with its rules of use and a gain measured at 30 days.
An AI assistant writes, summarises or searches when someone asks it to. That person checks the answer and decides. With an AI Assistant Setup, each team gets its own assistant, tuned to its work.
Your teams already use ChatGPT or Copilot, each on their own, often with personal accounts and no written rules.
You already pay for AI licences, and few people really use them.
The answers are correct, but generic: the tool does not know your company, your line of business or your tone of voice.
You want your teams to save time in the coming weeks, with no IT project and nothing to connect to your tools.
You can open the accounts yourselves. What makes the difference is everything else: the right tool for your company, your reference documents (what your assistants need to know, in writing), a setup checked against each team’s real cases, and rules everyone understands. We put it all in place in two to three weeks, then measure the gain at 30 days; everything we write belongs to you.
Three ways to hand work over to AI. What changes: who starts the work, and who decides.
A first draft, a summary, a reply: the person who asked reads it, corrects it, then uses it.
Each time an email arrives, or at a set time: nobody has to start it, and there is nothing to decide along the way. The result goes straight into your tool.
It takes charge of an entire process, decisions included, and lets you approve anything that commits your company.
The three often work together: your teams write with their assistant, and the work that recurs every week is handed to a Task Automation or an agent. The Flash Diagnostic tells you which one to start with.
In two to three weeks: the tool chosen and set up, one assistant per team, clear rules, then the gain measured at 30 days.
The tool chosen, and why, in writing: what you already have, your data constraints, the cost of licences. It always includes a European option.
Business accounts, privacy settings in place, and the right permissions: everyone sees what they should see.
What your assistants need to know about your company, stored in your own account, outside the AI tool.
Tuned to its work through the instructions given to the AI, then checked against the team’s real cases.
What you can hand over to AI, and what never goes into it.
On the chosen tool and its assistant, with the team’s own examples.
The adjustments, then the measurement: who uses it each week, and the time saved on one or two tasks.
Your reference documents, the instructions given to the AI and the charter, in your own account (except the tool’s licences).
Basic scope: the analysis, the setup and one team’s assistant. Each additional team gets its own assistant and its own onboarding session.
We start from what you already have: with Microsoft 365, it is often Copilot; with Google Workspace, Gemini; otherwise, ChatGPT, Claude or Le Chat. For your case, the selection note compares what each tool does for your teams, where it processes your data and what its licences cost each month.
Le Chat, from the French provider Mistral AI, is always included. And because we resell no licences, our advice does not depend on any provider: you make the call.
Your reference documents belong to you: you can switch tools without losing them.
They are stored in your own account: in a GitHub repository where possible (an online space that keeps every version), otherwise in your company drive. The AI tool only receives a copy. What goes into them is decided with you, based on what is already written down.
Two to three weeks up to the last onboarding session, then the 30-day review.
The base price covers the analysis, the choice of tool, your reference documents, the AI use charter and a first team’s assistant. Each additional team has its own price, known in advance.
Analysis, setup and one assistant, from€1,900 excl. VAT Per additional assistant+ €790 excl. VAT
Tool licences not included: you buy them in your own name, from the provider.
One assistant per team: the first is included, and each one after that has its own price, known in advance. You count your teams yourself; their size does not change our price, only the cost of your licences.
Indicative prices, excluding VAT, for the basic scope described. Your price is agreed with you in the quote, before any work starts: it is guaranteed, with no surprises.
For each team: how many people use their assistant each week, and the time spent on one or two tasks chosen with you, measured at the start, then 30 days after onboarding.
Where your data is processed depends on the chosen tool: the selection note states it for each tool compared, before you decide.
ChatGPT, Claude, Copilot and Gemini come from American providers; Le Chat comes from the French provider Mistral AI. It is included in every selection note.
The AI use charter sets this out for all your teams. Your reference documents contain no personal data, and nothing is loaded into the tool without your agreement.
This is a deliberate choice: with an off-the-shelf tool, your data is processed by its provider. Our three sovereignty levels do not apply here; they cover what we put into service ourselves, an automation, an agent or a Company Brain. The three levels, explained
The 30-day review shows the useful next step, or that none is needed.
Who decides, who approves, and how you stay in control of your agents.
Where your data goes, and how to require that it stays in France.
Personal data handled by the rules, from the design stage.
Complying with the European AI regulation, and being able to prove it.
The schemes that can reduce what you pay, and the application we prepare with you.
Our free tool to choose an AI model by task, price and sovereignty.
Free and with no commitment. The Flash Diagnostic tells you whether assistants are enough for your teams, or whether a task would be better off running by itself.