The firm
NEURIXIS is a French AI consulting and engineering firm. Two engineers, a quantified method, and the obligation to prove what we claim. This page explains where the name comes from, how we work, and what we do not do.
The brand
NEURIXIS holds two roots. Neuro, the living thing that learns. Axis, the line that measures. The name sets out both halves of the job: build systems that learn, and hold them to a measured axis. Where the two cross, there is the X.
Value sleeps in the dark of your data. That is the forest, the deep green of our identity: dense, unlit, full of things that matter and that nobody looks at. NEURIXIS draws one yellow line through it, sharp and single, exactly where it counts. What the line touches gets proven. What cannot be measured stays a promise.
The logo says the same thing. Inside the X, the rising diagonal, in yellow, carries the flow and the value. It passes in front of the falling one, which carries the structure. The order is not decoration: proof comes before the pitch.
Everything else follows from that: one yellow line per screen, never two. An accent that spreads stops pointing at anything. True of a poster, true of a portfolio of AI projects.
AI you can measure.
The manifesto
A rule that costs nothing commits nobody. Here are ours, with the price.
Every engagement opens with a business case: volume handled, time spent, error rate, unit cost, projected inference cost. The price is that our first commit lands two weeks later than a supplier who starts coding on day one. And that some projects die during scoping, before costing a penny of development. That is the point.
Before we start, we write down what will count as success: which indicator, measured how, over what scope, by when. That criterion can be turned against us, which is precisely why it exists. A project that misses its target is observed, not narrated.
If the analysis turns up no use case that pays back within eighteen months, we tell you and we do not sell the next phase. We lose the work. We keep the one thing that brings a client back: the certainty that our recommendation does not depend on our order book.
Your repository, your CI, your review rules, your release procedures. Not a platform of ours parked alongside, which you would keep for the length of the contract. The start is slower: we have to learn your conventions and get your access. In return, what we deliver does not die the day we leave.
Operating documentation, skills transfer, your engineers taking the code over: the exit is a deliverable, not an accident. It denies us the annuity an integrator builds by making itself indispensable. We would rather have a client who comes back because they had a choice.
You talk to the two engineers doing the work. No engagement director, no layer that rephrases. Nobody is there to soften bad news, so you get it raw. And we cannot run ten projects at once: we turn down more than we take.
If it isn’t measured, we don’t sell it.
The team
The company is owned and run by its two founders. They scope the work, write the code and sign the recommendations.
President
Data scientist. Several years at Airbus Helicopters on AI projects in industrial settings, from scoping to production. He advises leadership on project strategy: which use cases to launch, in what order, for what return.
Chief Technology Officer
Data scientist, trained in the demands of industrial environments at Saunier Duval. He owns the technical side end to end: real data, production and monitoring, at the reliability level an industrial site requires.
The limits
Better said before the first meeting than at the third.
We work in France and neighbouring countries, in French and in English. A simultaneous rollout across twelve subsidiaries and five time zones needs a structure we do not have, and we will not subcontract one to look bigger.
We do not place consultants on day rates to fill a resourcing plan. We take bounded engagements, with a deliverable and a success criterion. If what you need is seats filled, a large services firm will do it better and cheaper.
We do not sell an AI roadmap that nobody executes. We scope because we build afterwards, and the scoping is worth what it makes buildable. If we could not implement a recommendation ourselves, we do not make it.
Two engineers is a chosen size. It rules out volume and forces us to select: we only take subjects where our presence changes the outcome. Against a large firm we do not compete on coverage. We compete on technical depth, and on the fact that the person who answers you is the person who delivers.
