WikiAI

In three years, artificial intelligence moved from the laboratory into daily professional life. The vocabulary followed: RAG, agents, MCP, fine-tuning, quantization. Every week brings one more model, one more tool, one more acronym.

The documentation has not followed. What exists is split between research papers unreadable by non-specialists, posts published by the vendors themselves — judge and jury at once — and discussion threads whose shelf life is measured in days. Most of it is in English, and much of it is already out of date by the time you read it.

The result: a student, a teacher or an executive who wants to understand what a large language model actually is, what the EU AI Act changes, or what the word "agent" really covers has no neutral, reliable place to start.

Every page exists in French and in English, under two distinct paths. French is not a late translation of the English: both versions are treated as content in their own right. It is a direct answer to a very real imbalance — most quality AI documentation exists today only in English.

Quality rests on a process drawn from academia and from code review: a submitted page goes to a reviewer assigned by expertise, who approves it, requests changes or declines it, before publication.

A mentoring programme supports new contributors through their first pages.

WikiAI is published and operated by NOVYUP, an independent publisher of digital platforms.