IA en un minutoNewsBusiness and products
Mistral AI buys SkipLabs, its fifth acquisition of 2026
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
SkipLabs, a French startup with offices in Paris and London, now belongs to Mistral AI for a sum that has not been made public. Mistral thus reaches five acquisitions this year, after Koyeb, Emmi AI, Pimento and New Material. Language is the name of the group its six engineers will set up to check the code produced by agents that program.
Mistral AI, the French generative AI company, has taken over SkipLabs, another French startup whose tools aim to make code written by artificial intelligence come out faster, more reliable and safer. Maddyness broke the deal on November 5, and how much Mistral paid has not been made public.
That makes five purchases for Mistral in 2026. Before this one came Koyeb (cloud) in February, Emmi AI (industrial AI) in May, Pimento (advertising) in September and New Material (user experience for AI) in October.
SkipLabs was set up six years ago and is based in Paris and London. Its founder, Julien Verlaguet, worked as an engineer at Meta for almost nine years. According to Le Monde Informatique, its staff is six engineers, who will make up a new team inside Mistral called Language. Their job will be to design verification systems for AI agents that code on their own, so they can spot their own mistakes and fix what they have written.
Before the deal, SkipLabs had built Skip, an open-source framework for reactive programming, and Skipper, an agent that produces back-end services from plain-language instructions. In 2025 it raised 8 million dollars in its first round, with Amplify Partners and Tapestry VC among the backers.
Mistral has not said which products will use this technology or when they will arrive.
Why it matters · analysis and opinion
Generating code is no longer the hard part; the hard part is being able to trust it. This purchase aims right at that point. Agents that write programs without supervision produce a lot, and fast, but a bug nobody catches can end up in a service customers use. Mistral wants to build the checks into the generation process itself, instead of leaving them for the end, and to do that it is bringing in a team that already worked on verification. The deal also confirms how it has grown this year: buying small, specialized companies to cover more fronts. For anyone already using coding agents, the practical takeaway is to keep tests and human review in place until these tools reach the products.
Source: Maddyness · Written with the help of AI: how we make the news
