Labs
The name is not decorative: deloslabs was born in a lab — and remains one.
Our R&D has explored artificial intelligence and data for three decades. What we build there ends up in products — and, whenever we can, in open source.
It all began with a small lab set up by Ivan Becerril and Jérôme Vollet — an inference engine on one side, cognitive science and data mining on the other. That lab was the cradle of deloslabs.
The standard hasn't changed: prove it, rather than pitch it.
We build on open standards, and we give back to the commons what we can.
Our data-driven application framework, open-sourced since 2006 — the foundation our products rest on.
github.com/deloslabs →Our thesis fits in one sentence: data makes intelligence, and vice versa. Buzzee's contact-centric model — the richest relational data there is — was designed, from the outset, as fuel for the finest artificial intelligences.
AI is no recent pivot here. It runs from the founders' first inference engine and their roots in cognitive science and data mining, through to a dedicated R&D team — led by a Chief Data & AI Officer, with researchers drawn from Polytechnique and the ENS — and to foundational work such as our paper on conversational meta-agents.
Where the research becomes a product. Two years of development and no fewer than five cooperating AIs record, transcribe and analyze calls in real time — fixed or mobile, on any operator or handset — turning each one into an intelligent summary: key points, appointments, tasks.
It is growing into a true augmented assistant: calling back the opportunities behind unknown numbers, reaching a contact across call, SMS and email to book a meeting, and reading stress or urgency in a caller's voice in real time.
And it is built the way we believe software should be: data and models hosted in France's sovereign cloud, compliant with the GDPR, the AI Act and the NIS2 directive, never shared with third parties.