Support
From decision-making through to delivery.
We help business leaders make decisions that match their resources and their ambitions. We stay by your side from the first conversation through to the concrete set-up: listening and framing, helping you choose, building, installing — on your premises included — then training and supporting your teams.
Who we work with
From the sole trader to the manufacturer with more than 100 employees.
A two-person company and a factory of a hundred have neither the same resources, nor the same urgencies, nor the same constraints. The approach is the same; the size of the project, the resources committed and the pace adapt to your situation. You do not need to be ready: helping you get there is our job.
- Listen and frame
We start from your real work: what takes time, what blocks, what stays fragile. We clarify the need, the available data and the risks, then we agree with you on a work scope and on the criteria that will judge what comes next.
- Help you choose
Before any investment, we help you decide: what to hand over to the AI, what to leave to your teams, and at what level of resources. Our role is to propose a project that matches your resources and your ambitions — not the largest possible, the right one for you. You can also decide to wait: we will say so.
- Build
We design and set up your AI, for one precise job: your AI that prepares your files, your monitoring AI, your customer follow-up AI, your AI for helping prepare quotes. Every deliverable, every access and every limit is agreed before we start.
See what an AI does by profession- Install on your premises or host
We look at three possible formats for your AI with you: installation at your premises, hosting by AC-Création, or installation on a rented server with a hosting provider. We compare what each entails according to your uses, data and resources.
Understand both options- Train and support the teams
We train the people who will use the AI on their own working situations, and the referents who will look after it day to day. Support continues after set-up: monitoring errors, actual use and costs, with changes tested before they are adopted.
Method
Validate a use case before extending it.
A successful demonstration is not enough. We also look for errors, incomplete cases and the effects on your team's work, before we ask you to decide.
Define the expected work
We choose a process with your business referent and set the expected result, the exclusions, the usable data and the conditions for success.
Try on a small scale
We test the setup on representative cases that you authorise. Your team checks the results and notes what still needs a correction or a human intervention.
Deciding on an informed basis
We review the tests, the access rights, the recovery procedures and the real costs. You decide whether the agreed criteria are met, or whether the project must be adjusted, reduced or stopped.
Organise what comes next
We install, hand over the documentation and train the users. Follow-up covers errors, actual use and costs; every change is subject to new tests.
Training and support
Knowing how to use the AI, and also how to contradict it.
The learning takes place on your own working situations. Teams learn to check an answer, to spot a missing source and to take back over when the AI steps outside its scope.
It is not a chat bubble: it is an AI that does a precise job. Like any worker, it is sometimes wrong — and sometimes confidently wrong. Learning to correct it is fully part of the training.
Depending on the assignment, you receive training material, a user guide and instructions for the referents: managing access, reporting an incident and tracking changes. We agree on the people concerned and on the responsibilities before the session. Training replaces neither your internal rules nor the review of data processing.
Human control
Sensitive decisions keep a named owner.
We distinguish the actions the AI can prepare from those it can execute. A message to a customer, a change to a record or a decision that concerns a person may require explicit approval.
The AI can produce an incorrect answer, even a convincing one. The checks, the possibilities of stopping and the way back to the usual process are part of the framing. The level of automation depends on the risks and on the results of the trial.
Practical questions
Before you start.
Do you already have to know what you want?
No. Most of the leaders we meet know what takes time or what blocks, without yet knowing what to hand over to the AI. Listening and framing serve precisely that purpose.
What do you need for the first conversation?
Describe a recurring task, the people concerned, the tools used and the difficulty encountered. Do not send confidential documents at this stage; we will agree a way of sharing them if their analysis becomes necessary.
How are the budget and the schedule set?
They depend on the scope, the available access, the connections to your tools and the tests needed. The proposal distinguishes the deliverables, the prerequisites and any recurring costs of use, hosting or maintenance, so that the decision matches your resources.
Let us start with one concrete task.
Explain what takes time or blocks your teams. We can then discuss a first scope and the points to check before you commit — and weigh together whether the project is worth it.