Klarent raises €7.13M seed to test the code AI writes

Klarent, a Zurich startup founded in 2023 by two former Google engineers, has raised a €7.13M seed round led by London’s Mosaic Ventures to scale its agentic software testing platform and expand into the US, with Moonfire Ventures, AngelInvest Ventures and angels including former Google director Jürgen Galler joining.

Everyone is letting AI write code this year – and almost nobody is fully sure the output works. That gap between generation and verification is exactly where this round lands. Below, I lay out the round, the Google alumni behind it, and why testing is quietly becoming the choke point of AI-written software.

A London cheque for Zurich’s test writers

Klarent announced a €7.13M seed round on 9 October 2026, led by Mosaic Ventures and joined by Moonfire Ventures, AngelInvest Ventures and a group of angels including Jürgen Galler, the former Google director, and Harris Barton.

The money has a direction. The company says the funds will go to entering the US market, growing the engineering team and broadening its cross-platform testing coverage for web and mobile applications.

The customer list is already more concrete than most seed-stage decks: Klarent tests software for European enterprises including JD Sports, NZZ and Sixt, and names banks and insurers as its next targets – buyers for whom a failed deploy is a regulatory event, not an inconvenience.

From fore ai to Klarent, Google alumni who stayed in Zurich

Founded in 2023 under the name fore ai, Klarent is the work of Asheem Panakkat, who spent 12 years at Google leading engineering teams on Shopping and Maps, and Momchil Ivanov, a staff engineer there for close to eight years, per the company’s own telling. Zurich hosts one of Google’s largest engineering bases outside the US, and the pattern of its alumni building where they already live – rather than relocating to San Francisco – is one we keep seeing on this desk.

The product is an autonomous testing platform for web, iOS and Android. You describe a test in plain English; a multi-agent system maps the application, plans the test, writes the code, verifies it, runs it and alerts you when something breaks, with tests that self-heal as the UI changes, per the company. The design choice that matters: Klarent’s agents write deterministic test code rather than keeping a language model in the execution loop – AI adaptability at authoring time, repeatable results at runtime. And for the banks and insurers it is courting, the company ships SOC 2 and ISO 27001 compliance and will deploy on-premises.

Automation testing compounds at 14.5% a year

The segment has a measured size. The global automation testing market stood at an estimated $28.1bn in 2023 and is projected to reach $55.26bn by 2028, a 14.5% CAGR (MarketsandMarkets, February 2024). The fact worth holding onto: that forecast was written before agentic coding tools pushed AI-generated code into production at scale. Every line of machine-written code still needs to be verified by something.

For scale against our own records: within developer tooling, we logged OllyGarden‘s €3.55M round into telemetry quality this same week, while at the other end of the seed spectrum Noah raised a $38M seed for stablecoin payments. A €7.13M Swiss seed with a named London lead sits in the fundable middle – enough to open a US office without betting the company on it.

When machines write the code, checking it becomes the business

So what does this round signal? The reality is that code generation has outrun code verification: engineering teams everywhere are approving more machine-written changes than any human review process was designed to absorb. The good news is that verification is a European-shaped problem. It rewards exactly what our ecosystem is structurally good at – reliability engineering, compliance, selling into regulated industries – and the talent is already here, walking out of Google Zurich, ETH and EPFL with precisely these skills.

What we are seeing is the second order of the AI boom getting funded: after two years of cheques for the agents that write code, investors are starting to back the agents that check it. Two things are worth watching from here: whether more London funds follow Mosaic into Swiss and German seeds, and whether Klarent’s US push keeps its engineering anchored in Zurich. Our fundraising data will tell.

AI will keep writing more of Europe’s software. The layer that proves it works can be built here – let’s make sure it is!