Skip to content

Automating is not transforming.

AI does not just speed up what you already do. It changes what your company is capable of doing.

What AI really changes.

Most companies come to AI through the back door: an assistant here, an automated task there. It saves time. But it leaves the essentials untouched.

AI makes it possible to decide with information you could not use before, to offer services you could not provide, to give your experts back the time their expertise really demands, to rethink an entire line of work rather than optimise one step of it. Sometimes it calls into question the very way the company creates value.

That potential cannot be decreed, and it cannot be bought off the shelf. It is built on a close understanding of your company: what it has to achieve, what makes it valuable, what holds it back. That is where we start. Always.

Adopting AI is not enough. You also have to stay in control of it.

Transforming in depth also means exposing yourself. Eighteen months on, the tool does the work, the results look good, and nobody remembers how things used to be done. Nobody really knows what each case handled costs, or what will happen the day the model, the prices or the terms of access change. The company has adopted AI. Without noticing, it has also lost control.

We do not see these two demands as opposites: one without the other leads nowhere. Caution that prevents transformation is not control: it is standing still. We hold both ends.

Technology for its own sake does not interest us. AI transforms nothing by itself: it is the choices made with it that transform a company. And those choices must remain yours.

A transformation you do not control is not a success: it is a dependency.

‘Human-centric’? Prove it.

You have read that phrase everywhere. It means nothing until someone says who decides what. We do.

With us, people set a system’s objectives; the system does not quietly redefine people’s work. AI widens what your teams can analyse, understand, create and do. Your teams keep the right and the means to challenge a result, to arbitrate, to take back control. And we make sure they understand their tools well enough to exercise that right for real.

For every action entrusted to AI, we set down in black and white its level of autonomy, on a scale of four settings.

This dial is often the real strategic decision. It is rarely taken consciously. With us, it is. That is what the human-in-the-loop principle we apply means in practice.

The autonomy dial.

  1. Level 1

    It suggests

    Who decides
    A person, every time.
    Example
    AI proposes three replies to a customer; the adviser chooses and writes.
    Oversight
    Full
  2. Level 2

    It prepares

    Who decides
    A person, before any action.
    Example
    AI prepares a quote; the salesperson reviews it, corrects it and sends it.
    Oversight
    Systematic approval
  3. Level 3

    It executes

    Who decides
    AI, following a scenario written in advance.
    Example
    AI sorts and routes incoming requests; a sample is checked every week.
    Oversight
    Spot checks
  4. Level 4

    It orchestrates

    Who decides
    AI chooses what comes next, within a set scope.
    Example
    For a request already covered by a contract, AI checks availability, creates the logistics tasks and prepares the confirmation, in whatever order it sees fit.
    Oversight
    Scope, alerts and the right to stop it

Let’s talk about your organisation and your needs. Not about technology.

Contact us