About

Ten years making AI systems work inside real companies.

My name is Anthony Crowther-Alwyn. I spent ten years in large French companies before founding Praedic, and I now put that experience to work for organisations with no data team and no time to waste.

Anthony Crowther-Alwyn

Founder

Anthony Crowther-Alwyn

"A model that cannot be explained to the people doing the job will never be adopted."
Based in
Templeuve-en-Pévèle
Works
Hauts-de-France, and remotely anywhere in France

I started out in classic data science, the world of forecasting models and decision systems, in companies where a mistake is counted in lorries, in empty shelves or in customers hanging up.

What I learned there fits in one sentence: the obstacle is almost never the technology, it is knowing what to build first.

Since March 2026 I have been applying that to generative artificial intelligence, which has put within reach of twenty-person firms the kind of work that used to need a whole team. The useful role is no longer knowing how, it is knowing what to do, in what order, and how far to take it.

  1. 01 2015 - 2021

    Inside the large companies

    Data science at or for Cofidis, Mondial Relay, Leclerc and Chronopost. That is where I saw what actually stalls a project: rarely the model, almost always the data everyone assumed was available, the decision nobody will own, and the tool the teams never take up.

  2. 02 2022 - 2025

    AI systems put into production

    Press sales forecasting for a media group, a platform that analyses every call coming into a contact centre, support agents for a software publisher, stock optimisation for a distributor, invoice checking inside a financial audit. The common thread: all of it went live, and all of it was measured.

  3. 03 Since March 2026

    Generative AI adoption in smaller companies

    Adoption projects at my main client, a 20-person consultancy, and at a remote monitoring company. Acquisition, sales automation, internal tools, data pipelines. The results shown on this site come from that work.

Convictions

Four things I am reasonably sure of.

The problem is rarely the technology

The models available today already go past what most companies of this size need. What is missing is someone who spends a week looking at where the time actually goes.

A system that does not run is worth nothing

A demonstration impresses a board for twenty minutes. A system used every day changes a company. They are not the same job, and the second one takes longer.

A model nobody can explain will never be adopted

That holds for a replenishment algorithm as much as for an assistant that drafts a proposal. If the person using it cannot see where the suggestion came from, they work around it.

Dependence is a risk, not a business model

The code is with you, the documentation too, and handover is planned from the start. A supplier you cannot do without always ends up costing more than it brings in.

Next step

Where would you start?

Ten days to map where your teams lose time, put a figure on each lever and name the first system to build. If AI is not the right answer, I will tell you so.

Diagnostic

Étape 1

10 days

€9,500 excl. VAT · fixed fee, one area of the business

  • A map of where the time actually goes in the area audited
  • Each lever costed in hours freed, ranked by effort and by gain
  • The data note: what leaves your systems, and where it is processed
  • A three-month plan, naming the first system to build