About AIssential

You work directly with Julien Pierre Martin. A PhD in AI, he has been building tools since 2000 alongside the people who use them, in logistics, animated film, investigative journalism and law.
Background and distinctions
- Sorbonne Université
- Inria
- ICIJ
- Brown Institute for Media Innovation
- Prix CNIL-Inria (distinction du jury)
- Prix DILA Open Law
- Supported by the CNC
Who's Behind It
He started in 3D animation, on the feature film Kaena: The Prophecy, then on series. There he designed, built and kept evolving production pipelines, with the director, the heads of department and the producers. A production brings together hundreds of people and up to eight trades, for months and sometimes years. It produces very large volumes of very varied data, and each trade depends on the work of the others. A mistake there costs dearly: it can pull dozens of people back onto the job for weeks. He later joined a studio where he led up to twelve people, including AI engineers, and his programme won support from the CNC, the French national film centre.
His PhD at Inria starts from a simple observation. An engineer describes a problem very well piece by piece, with the rules of the trade, but the tools that solve hard problems make them restate everything in one block. So he designed a language in which engineers write their know-how the way they think it, and the machine finds the solution. The work was tested in the logistics industry, in the European Net-WMS project, on loading goods under the rules of each warehouse. He was already putting a trade's know-how into software.
At Imaginatio, he led R&D and then operations for a young innovative company offering an augmented legal word processor. While a lawyer reads or writes, the cited texts appear in the margin, in the version in force. The method is patented.
At the ICIJ, the consortium of journalists behind the Panama Papers, he was head of innovation. He developed the first version of DataShare, which lets journalists search their documents without handing them to anyone, and cross-check them with those of their peers inside a confidential virtual network. The project is a laureate of the Brown Institute for Media Innovation (Stanford / Columbia), and the paper describing its networked version was distinguished by the CNIL-Inria privacy prize jury in 2021.
The thread never changed. We understand the whole trade, with everyone who takes part in it, before putting software into it. We then build a tool that holds in production, because the people who use it have no time to fix it. Software takes the what follows a rule, and the matters of judgment stay with the trade. That is what we do in your business, one piece of work at a time: AI where it pays off, your expertise where it counts most.
What We Know How to Do
- Writing down a trade's know-how. His PhD designed a language in which engineers write their know-how the way they think it, and the machine finds the solution. It was put to the test on loading goods in warehouses.
- Tools for experts. At Imaginatio, a legal word processor that shows the cited texts in the margin, in the version in force. The method is patented, and the company won the DILA Open Law prize.
- Artificial intelligence. An AI specialist for twenty years, he has led AI engineers. He knows how to choose an AI tool and judge it on your files, not on its demo. We know what each method can do and what it misses: where AI pays off, and where a plain calculation is enough.
- Decision support. Scheduling, allocating cases, choosing under constraints: a part of the work that calculation solves without resorting to AI.
- Software architecture. Production pipelines that hundreds of people and up to eight trades depend on, for years, where a mistake costs dearly.
- Security and confidentiality. DataShare lets ICIJ journalists search their documents without handing them to anyone. Its networked version was distinguished by the CNIL-Inria privacy prize jury.
- Operations management. He ran R&D and then operations at Imaginatio, and led up to twelve people in a studio. That is how we leave you a way of working that runs without us.
What We Do
So we start with a single process. We choose with you the one that sits at the heart of your company's biggest stakes. We write the line between what follows a rule and matters of judgment, we measure what that process costs you today, and you leave with both — even if you stop there. Your people stay on all their files: we take away only what no longer teaches them anything. The three reference files cost €4,500 excl. VAT; the software is quoted at their end.
Why one process at a time
AI is discussed everywhere. Far fewer can say what it changed in their own operation. Among French companies of 50 to 249 people, 31% use any AI at all (INSEE, 11,000 firms). The tools work very well, and each person already uses them for their part. It is the whole file, and what each person does in it, that has never changed.
A more powerful AI would change nothing. Your procedure is written down. What your teams look at when they open a file is not. They see at once the missing document, or the price that will not hold on this job. As long as that knowledge stays in people's heads, AI cannot use it.
Not everything is solved by tools that write, such as ChatGPT. Part of the work is a plain calculation. Another part is constrained optimisation, such as scheduling or case allocation. That is AI too, an older kind, and it was the subject of Julien Pierre Martin's thesis. For each task, we use the cheapest method that holds up, and we keep AI for where it pays. Of the AI use cases put in place, only 28% fully succeed and deliver the expected return (Gartner, 782 leaders). Putting AI everywhere is not a method.
Twenty minutes, free. You will know whether it can be done in your firm.
Counsel — our self-serve decision aid
Counsel is the software side of the same method, and you can use it without talking to us. Weighing a specific AI decision? Counsel argues a position on your call, with every claim quoted from a real source — never generated. 14 days free, no card; then €99/mo Solo for every decision you bring.
We publish public verdicts on the calls leadership teams are weighing now, so you can see exactly what Counsel produces before you get in touch.
What Counsel reads
Counsel reads 800+ trusted sources — research institutions, industry publications, independent blogs and media, including arXiv, Hugging Face, Google DeepMind, OpenAI, MIT, Stanford and The Gradient — across research papers, industry analysis, tutorials, open-source releases, podcasts and expert commentary. You set your role and interests on the free tier; on Pro, you share the decisions on your plate and each one is argued back at you as new evidence lands.
The Company
AIssential is a product of MaGMa, a French technology company registered in Évry, France (SIRET: 81968546200015).
For legal and corporate information, see our Legal Notice.
Get in Touch
Do you have a process at the heart of your company's biggest stakes? That is the conversation we are looking for.
Visit our Contact page or email us directly at contact@aissential.tech.