We explore what AI truly changes.
Artificial intelligence is a major technological rupture. It is already transforming the way software is produced. Tomorrow, it will reshape how we interact with machines, tools, and organizations more broadly.
Lesur.AI is a French lab, based between Caen and Paris, working to make this technology useful, effective, and valuable while keeping humans at the heart of operations.
Reinventing software operations
We believe AI requires a deep reinvention of the operational model behind software.
Alongside faster coding, this means rethinking roles, workflows, memory, supervision, trust, and how humans collaborate with increasingly autonomous systems.
Our work starts with software and will extend to other fields.
Building patiently
We build products, protocols, and methods to progressively tame this new technological layer.
Our work remains intentionally discreet for now. We want AI to be reliable, understandable, and useful to the people operating it.
Products
Hivemind
One shared memory for your AI agents: they notice what others are doing, inherit what others have learned, and understand projects together. Open-source, self-hosted, Apache-2.0.
Transkryptor
Turn meetings, courses and trainings into clean transcripts and ready-to-send reports. Open-source, self-hosted, GPL-3.0.
Stealth Project
A new way to interact with models, tools, and knowledge.
Stealth Project
An environment for rethinking software production with autonomous agents, human supervision, and more structured operations.
Research
Preprints, working papers and technical notes from the lab, published in open access with their limits stated.
Hivemind FIDELIS Benchmark: Measuring the Fidelity of Consolidated Agent Memories Guillaume Lesur · Christophe Lesur Read the paper All publicationsBlog
Read the release notes for Hivemind and Transkryptor and articles about our research.
Why we are starting a blog Read the post All posts