Mistral strengthens its code agents with the acquisition of SkipLabs
Mistral AI has acquired SkipLabs and recruited its six engineers into a new team called Language. The goal is to better verify code generated by AI agents, with no detailed timeline or product yet.
Mistral AI has acquired the French startup SkipLabs and will integrate its six engineers into a new team called Language. The operation aims to strengthen verification mechanisms for code generated by AI agents, a topic that has become central as these tools move from suggestion to execution. The acquisition amount was not disclosed. At this stage, the company does not specify the products involved or the deployment timeline.
Verifying machine-produced code, not just writing it
This is the core issue: a programming agent can autonomously generate code, but it must also know how to detect its errors and correct its output. Mistral therefore wants to add more controls in the generation chain, with mechanisms capable of quickly integrating into different stages of the process.
This need is not abstract. Traditional development tools already rely on compilers, static analyzers, and tests to spot errors. But in an AI-driven environment, this feedback must be precise enough for an agent to use it without human intervention. The developer reads an error message and decides; the machine, however, needs a much clearer signal. Otherwise, it stalls. Or worse, it insists.
SkipLabs arrived with building blocks already useful to Mistral
Before the acquisition, SkipLabs worked on several programming-related technologies: compilers, typing systems, static analysis, and incremental computation. The startup had also developed Skip, an open-source reactive programming framework, that is, an approach that updates certain processes based on data changes without recalculating everything. It also offered Skipper, an AI agent designed to generate back-end services from natural language instructions.
This last component is particularly important in the current context. Producing code that “looks right” is not enough; it must also be correct and usable in a real software project. SkipLabs’ tools already provided type-checking functions, useful for controlling the consistency of generated code. Mistral thus gains very targeted expertise at the junction between generation and validation.
An acquisition that fits into a series of broader purchases
The acquisition of SkipLabs adds to other external growth operations carried out by Mistral AI, which has already acquired Koyeb, specialized in cloud infrastructure, Emmi AI, focused on industrial simulation, Pimento, positioned in marketing, and New Material, dedicated to product design. Meanwhile, SkipLabs had raised $8 million in seed funding in 2025, notably from Amplify Partners and Tapestry VC.
This sequence says something about Mistral’s strategy: enriching its models with technical building blocks very close to concrete uses, rather than just stacking generalist capabilities. The generative AI market is no longer played only on model size, but also on their ability to produce something reliable, especially when it comes to code.
Key points
- Mistral AI acquires SkipLabs and adds six engineers.
- The new team is called Language.
- SkipLabs raised $8M in seed funding in 2025.
- The acquisition amount was not disclosed.
- Mistral cites uses for verifying generated code.
By the numbers
- 6 engineers — integrated at Mistral AI, according to the announcement.
- $8M — raised by SkipLabs in seed funding in 2025.
- 4 acquisitions — Koyeb, Emmi AI, Pimento, and New Material, cited by Mistral.