Gallica / BnF — The French national library as a data partner for AI

Overview

National libraries hold centuries of digitised knowledge that AI systems either ignore or scrape badly. Gallica, the digital library of the Bibliothèque nationale de France, serves millions of digitised documents — press archives, books, maps, manuscripts, images — none of it designed with AI agents in mind.

With the BnF, Alien Intelligence builds three things:

• MCP access to the collections — catalogue and full-text search, document and page-level reading, IIIF imagery, exposed to AI agents as a typed tool surface.
• AI-extracted corpuses — turning raw digitised collections into useful, well-described corpuses (historical French press is the first ground).
• Agentic services for the BnF’s B2B clients — putting those corpuses and tools to work for the institutions and companies the library serves.

My role

I lead the partnership for Alien Intelligence: the work with the BnF teams, the agent-facing tool surface over Gallica’s APIs, and the first corpus extractions.

Where it stands

A first agent-facing access layer is operational and corpus work is under way. A full write-up follows as services open to the BnF’s clients. The same principles drive the OpenAIRE MCP and LDS work.