About AC-Création

Supporting leaders from decision through to delivery

We help business leaders make decisions that match their resources and their ambitions. We support from decision-making through to delivery. From the sole trader to the manufacturer with more than 100 employees.

Let's talk about your business

Our approach

Decide first, build afterwards

An AI idea deserves to be turned into a decision that can be examined, discussed and tested.

See the support step by step

How far does the AI go, and where does it stop? What job can it prepare, and which decision stays yours? These questions help you delimit a task rather than hand over a whole process to a machine, and to decide on an informed basis.

Our offer runs from decision-making through to delivery: we build a custom-built AI for one precise job, then install it on your premises — on a server in your company, a custom-built machine or an off-the-shelf model such as the NVIDIA DGX Spark — or we host it if that is what suits. Installation on your premises is the decisive advantage when your data is sensitive; it does not go without saying and does not always mean a dedicated machine. When several tasks, documents and validations have to work together, your AI organises itself around a workspace adapted to your teams.

The support serves to define the rules of use: what your AI can do, when to ask for a review and how to report an error. The training starts from the situations users encounter.

Choosing with the stakes in mind

Deciding with the stakes in mind

An AI can be wrong, and its use changes trades, jobs and ways of working. We talk about it without angelism and without fear, before deciding.

  • An identifiable human validation

    Sensitive decisions stay under human responsibility. You have to define who reviews a result, on which elements and before which action.

  • Access rights suited to the need

    The documents your AI can reach, the rights of action and the place of installation are discussed project by project. Your AI installed on your premises, on your own server, is the most direct way of keeping your data. A general promise of confidentiality does not replace these choices.

  • Results to evaluate

    A demonstration is not enough to establish the usefulness of a tool. Representative situations, acceptance criteria and the errors observed have to guide what comes next, taking account of the costs, the dependency created and the frugality of the uses.

Project examples

A few uses to understand the scope

Checking information at scale, preparing the screening of applications, preparing an answer to a customer: these types of projects illustrate the separation between preparation and decision.

These examples describe types of projects, the envisaged scope and the limits of each use. They are neither deployment reports nor measured results.

First conversation

Start with a decision to take

Describe what takes time, requires repeated checks or blocks a decision. We will discuss the scope, your resources, the place of the AI and what it involves for your teams.

Describe my need How far the AI goes: our landmarks for deciding