The questions directors actually ask us
The same ones come up in every first call, and several of them call for an answer few companies are willing to put in writing.
What exactly does Edicore build?
The artificial intelligence that belongs to one company: a system fed by its history, its files, the software it already runs on, and the way its director makes decisions. It is not a generic assistant installed on your premises. It is made of your material, and it would not work for anyone else.
We do not replace your tools with ours, and we do not hand over a turnkey project whose upkeep then becomes your problem.
What does it cost?
You do not pay for the build: we carry it, for the pilot and for the extensions that follow. From the first day of the pilot you pay a monthly operating fee: subscriptions specific to your company, hosting, artificial intelligence models, project management. Our remuneration is a share of the gains the AI actually delivers, measured with you rather than claimed by us. If the pilot delivers nothing, it ends on schedule and the fee ends with it.
The amount of the fee depends on actual operating costs, and the share of the gains has no reason to be the same for an industrial mid-cap and for a services group. Both are agreed before we start, and revised afterwards whenever both parties agree to it.
Does this AI belong to us?
No. It lives through our collaboration, and its use ends if that collaboration ends.
This is not a penalty clause, it is a technical consequence. A company AI is not a deliverable you install and forget. Your offering shifts, your teams change, your tools evolve — without constant upkeep it starts answering beside the point within months. Left alone, it would stop serving you anyway.
Your material, on the other hand, stays yours: nothing that goes into your AI feeds a product sold on elsewhere. If the collaboration ends, you get all your data back, structured and in a format readable without us, along with the documentation of what was built.
Where does our data go?
It serves your company and no one else. Your files, your exchanges and the way you arbitrate do not feed a product resold to other businesses: what we build inside your company has value only inside your company.
It is hosted on professional infrastructure located in the European Union, with restricted access. It serves no other client and trains no artificial intelligence model: the models are used through professional services whose terms rule out any other use. A confidentiality agreement can be signed before the diagnosis.
This is the same discretion we commit to. The companies we work with appear nowhere, and never will.
How long before it is of any use?
A first useful version in weeks rather than quarters, on a deliberately narrow scope, chosen because it matters to you and not because it makes for a good demo.
Building the whole thing, however, is counted in years and in five stages: understanding the company, building, driving adoption, measuring, and growing it. The first stage is the longest, and it is the one no fixed-scope engagement ever takes the time to do.
How does a pilot run?
One to three workstreams, chosen together from those the diagnosis brought to light, on three criteria: speed of implementation, availability of the data, measurable gain. The scope is fixed and does not move along the way; whatever comes up is noted for later.
The first week records the starting point of each workstream, in your own tools and with your designated contact. A first usable version is put in the teams’ hands during the pilot, to be corrected in use. A thirty-minute check-in each week with your contact, a monthly steering meeting with you. At the end, the end point is recorded by the same method: the gap is the result of the pilot, and it is the basis for the terms of what follows.
What can it be used for, concretely?
It depends on the company, and the diagnosis is what tells. By way of illustration, in a services company: preparing schedules and covering absences, reconciling hours worked with contracts before invoicing, a draft quote written from standard terms, tracking complaints through to closure, a monthly report per client generated from activity.
None of these examples is an achievement to date: they are targets a pilot can aim at, each with an indicator recorded before and after, in your own tools.
How is this different from a consulting firm?
A supplier bills days. It earns the same whether the project succeeds or not, and it leaves once the order is closed.
We commit for the long term and our remuneration is indexed on real gains. We stay once the tool is in service — the part of the work nobody wants to take on, and the part that decides whether the AI is still useful two years later.
How is this different from an off-the-shelf AI tool?
Your competitors can subscribe to the same tools you do, on the same day. An AI made of your own way of working is not something anyone can buy: it would not exist without you.
That is also why it gets adopted. A generic tool is abandoned within weeks — not out of resistance to change, but because it understands nothing about the work your teams actually do.
Do you take a stake in our company?
No. We do not enter your capital and we do not ask for shares. The relationship is a multi-year collaboration contract, and nothing else.
Is this going to cut jobs?
We do not promise to cut jobs, and we do not sell it as an argument. What the AI takes on is the repetitive work nobody ever enjoyed doing, and the information your teams spend their time hunting down.
The subject that interests us more is knowledge that evaporates: what your company knows lives inside people’s heads, and people leave. Your AI captures that tacit knowledge and makes it transferable.
Do you have client references to show?
None, and that is a commitment rather than a gap. What we see inside a company belongs to that company.
What we show instead are our own ventures: Full Moon Africa, Communauté Provence Énergies, VisitCasamance, Ambassadeurs de Casamance and Ambassadeurs de Provence. We created them and we run them. They are not client references, they are our own ventures, and they are where the way of working we are proposing to you has been tested for years.
Who do you work with?
Directors of large SMEs and mid-cap companies, in France and in West Africa.
The method needs time at the outset and genuine access to the company: it assumes a director who decides, not a committee arbitrating a tender.
Where do you operate?
Paris, Aix-en-Provence, Nice, Dakar and Cap Skirring.
A commercial presence: the diagnosis is conducted remotely, and we come to you when the pilot requires it, with no offices or staff on the ground.
Who is behind Edicore?
Marc and Bruno. Marc spent twenty-five years growing companies, from the field to the board: building a company’s AI first requires being able to read that company. Bruno designs and develops the platforms in the portfolio and builds the agents.
Edicore is family-run. We choose few partners, and we stay.
How do we start?
With a thirty-minute video call: a conversation about your company, about what is costing you time, and about what an AI that knew you could change. No demo, no quote.
If the conversation warrants it, it opens onto a diagnosis, offered free of charge and conducted remotely: a preparatory questionnaire, then a one- to three-hour interview with you, completed at your request by two or three managers. You leave with a map of your processes and data, and two or three candidate workstreams for a pilot. If we see nothing serious to do, we will tell you on the call.
Everything else is discussed case by case. It is the first subject of the call.