AI can now generate software faster than many engineering teams can verify it. Klarent is building for the bottleneck that appears after the code is written: proving that it actually works.
The Zurich-based startup has raised $8 million in Seed funding, led by Mosaic Ventures, with participation from Moonfire Ventures and angel investors including former Google director Jürgen Galler, Harris Barton and AngelInvest Ventures. Klarent says the round will primarily fund its US expansion, while also growing engineering and extending the platform across web and mobile testing.
Founded in 2023 as fore ai, Klarent was created by former Google engineers Asheem Panakkat, CEO, and Momchil Ivanov, CTO. Panakkat spent 12 years at Google working across products including Shopping and Maps, while Ivanov spent nearly eight years there as a Staff Software Engineer focused on information retrieval. The company adopted the Klarent name in August 2026.
Its technical bet is unusually specific: use AI to write and adapt tests, but not to improvise their execution.
Klarent says its LLM-powered agents generate deterministic test code, which is then executed repeatedly in a conventional, reproducible way. Its architecture divides the workflow between six agents: Explorer maps an application’s screens and states; Planner designs the test; Coder creates the test script; Verifier checks plans and code; Runner executes the approved suite; and Notifier sends the result back to the team.
That distinction matters because purely agentic browser automation can vary between runs. Klarent’s approach keeps AI in the authoring and adaptation layer while execution remains deterministic. The company also describes a human-in-the-loop service that verifies results, and says customers pay for verified outcomes.
The product is broader than a prompt-to-test generator. Klarent’s documentation covers test creation, test management, self-healing, test data, browser states, debugging and CI/CD integrations including GitHub Actions, GitLab, Jenkins, Azure DevOps and Tekton. Its web automation is built on Playwright, while its mobile product supports native, hybrid and cross-platform iOS and Android applications.
Enterprise adoption is beginning to provide evidence beyond the architecture. Klarent names JD Sports, Neue Zürcher Zeitung and Sixt among customers. In Klarent’s published NZZ case study, the publisher reports 98% of smoke tests automated, 75% faster smoke testing and more than half of tests created by product and business teams. Another customer, OnlineFuels, is reported to run 130+ automated tests in under four minutes, four times more frequently than before. These remain company-published customer results rather than independent benchmarks.
Klarent is also positioning for enterprise environments where test data cannot simply leave the network. The company states that it is ISO 27001 certified and SOC 2 compliant, encrypts data in transit and at rest, and supports cloud, private-cloud and on-premises deployment.
Its hiring already reflects the Seed strategy, with current openings spanning US and UK sales, infrastructure and scaling, and ML quality engineering.
The next question is bigger than whether AI can generate another test script. It is whether deterministic, agent-authored QA can scale quickly enough to become the trust layer for software increasingly written by other AI agents.
