Ex-Google Founders Raise $7.7M to Fix AI's Biggest Blind Spot

Jan de Vries · 2026-10-09

Two former Google engineers just raised $7.7 million to solve a problem most of us haven't even noticed yet. And honestly? It might be one of the smartest bets in tech right now. Klarent, a Zurich-based startup formerly known as fore ai, just closed a Seed round led by London's Mosaic Ventures. Moonfire Ventures and a handful of heavyweight angels joined in, including Jürgen Galler (a former Google Director), Harris Barton, and AngelInvest Ventures. But here's what makes this interesting beyond the dollar figure. ### The Problem Nobody's Talking About AI can write code faster than ever. Like, scary fast. But here's the catch—who's checking if any of it actually works? That's the gap Asheem Panakkat and Momchil Ivanov spotted while working at Google. Panakkat spent over a decade leading teams on Shopping and Lens. Ivanov was a Staff Software Engineer focused on information retrieval. Together, they watched large language models get smarter and realized something uncomfortable: the systems meant to verify all that AI-generated code were nowhere near ready. As Panakkat puts it: "How do you keep these coding agents honest?" Fair question. And one that enterprises are starting to take very seriously. ### Their Fix: Let AI Write Tests, Not Run Them Here's where Klarent gets clever. Instead of letting AI run tests directly (which can be unpredictable), their agents use LLMs only to *write* the test code. Once written, that code runs deterministically—same input, same result, every single time. It's like having a brilliant assistant draft your checklist, but you still control the execution. There's also a human-in-the-loop layer, so customers only pay for verified outcomes. No fuzzy AI magic. Just rigorous, repeatable engineering. Chandar Lal from Mosaic Ventures explained why this mattered: "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." ### Real Clients, Real Results This isn't theoretical. JD Sports, Swiss media group NZZ, and mobility company Sixt already use the platform. At JD Sports, Klarent's AI agents automate multi-system workflows across the e-commerce pipeline—something the retailer admitted its own teams couldn't replicate. That's a strong endorsement. ### What's Next: America The fresh capital will fuel Klarent's push into the United States, building on momentum from European enterprise clients. They're also expanding their engineering team, broadening cross-platform testing across web and mobile, and moving into financial services—where a broken software release can cost millions. Smart move. Banks and insurance firms can't afford guesswork. As AI writes more of the world's code, testing stops being a cost center. It becomes the thing that determines whether any of that code can be trusted. Klarent's betting on that shift—and with $7.7 million behind them, they're positioned to prove it.