Klarent raises €7.13M seed: the agentic QA startup that lets AI write tests — but never run them
Klarent — formerly fore ai — has closed a €7.13M seed led by Mosaic Ventures. Its agents use LLMs to author test code but execute it deterministically, betting QA becomes the bottleneck as AI writes more of the world's software.
Zurich-based Klarent has closed a €7.13 million seed round to scale its agentic software-testing platform. The round was led by London-based Mosaic Ventures, with participation from Moonfire Ventures and angel investors including Jürgen Galler, a former director at Google, Harris Barton and AngelInvest Ventures.
The pitch is an answer to a question the whole industry is now asking: as AI coding agents write ever more of the world’s software, who keeps that code honest? Klarent’s answer is what it calls “deterministic autonomy.” Its agents use large language models to author test code, but the tests themselves execute in a deterministic, repeatable way — combining, in the company’s framing, the adaptability of AI with the rigour that production engineering demands. A human-in-the-loop verification layer sits on top, and customers pay only for verified outcomes.
From LLM evals to QA itself
Klarent — formerly fore ai — was founded in 2023 by two former Google engineers: CEO Asheem Panakkat, who spent more than a decade at Google leading teams on Shopping and Lens, and CTO Momchil Ivanov, a staff software engineer focused on information retrieval. The pair began by building evaluations for LLMs, then realised the bigger opportunity was quality assurance itself.
“At Google, Momo and I could see two things coming at once: LLMs were getting better fast, and the amount of software being created was about to explode,” Panakkat said. “As AI writes more of the world’s code, testing stops being a cost centre and becomes the thing that determines whether any of that code can be trusted.”
Enterprise traction and what the money buys
The approach has already won over enterprise clients in Europe. JD Sports uses Klarent’s AI agents to automate multi-system workflows across its e-commerce pipeline — automation the retailer said its own teams could not replicate. Swiss media group NZZ and mobility company Sixt are also on the customer list, according to the funding announcement.
The funding will bankroll Klarent’s entry into the United States, building on its European momentum. The company also plans to grow its engineering team and broaden cross-platform testing capabilities across web and mobile — and it is pushing into financial services, pitching banks and insurers, where the cost of a broken software release can be catastrophically high.
Why it matters
The round crystallises a thesis investors keep repeating: code generation is scaling faster than our ability to test it. “With AI, code generation is scaling faster than our ability to test it, so QA will become the key bottleneck,” said Chandar Lal, partner at Mosaic Ventures. “What convinced us was their insistence on determinism – using AI to write tests, not to run them – which is why enterprises are trusting Klarent in production.”
It is a deliberate contrast with the wave of AI testing tools that let models both write and execute tests — a closed loop that can hide its own failures. Klarent’s bet is that in testing, the one place you don’t want probabilistic behaviour is the execution step. If AI-written code keeps compounding, the audit layer may end up mattering more than the codegen itself. A second funding summary confirms the round details and Klarent’s US-expansion plans.