AI coding tools have changed how fast software gets built. But the QA process has not caught up. Engineers still write and maintain test scripts manually. QA testers still click through products to verify that features work. Holmes, a Ghent-based autonomous QA platform, is built to close that gap by learning how users move through a product and testing those flows continuously, without anyone writing a single test script.
The problem is real and growing. As tools like Cursor, GitHub Copilot, and Codex accelerate how fast code gets written, the bottleneck shifts to quality assurance. Teams end up choosing between shipping fast and accepting bugs, or slowing down to test properly. Holmes removes that tradeoff.
How Holmes Actually Works:
Holmes does not rely on predefined scripts or manual test maintenance. It observes real user behavior, learns the full journey a user takes through the product, and generates tests based on those actual flows. As the product changes, the tests update automatically. No one has to rewrite them.
Five specialized AI agents handle happy paths, edge cases, responsive layouts, accessibility, and error recovery. The platform covers the user-facing flows that matter most: sign-up, login, checkout, search, navigation, and forms. When an issue is detected, Holmes surfaces it directly inside the CI/CD pipeline and the developer’s existing workflow, so there is no separate testing environment to manage or context to switch to. The bug shows up where the engineer is already working.
The Founders Behind it:
The three co-founders bring direct, relevant experience. Robin Praet and Robbrecht Delrue co-founded Smartendr, an ordering solution for the hospitality industry, which OrderBilly acquired in September 2025. Building reliable software for non-technical users in high-pressure environments gave them a clear view of what breaks when QA is deprioritized.
Sofie Buyse was the first employee at Henchman, an AI tool for legal professionals that LexisNexis acquired for €136.1 million in June 2024. As product manager during Henchman’s growth, she experienced the QA problem firsthand. Skilled QA testers are expensive and hard to find, so testing falls on developers and product managers who are already stretched thin. That experience directly shaped what Holmes became.
Funding and Early Traction:
In May 2026, Holmes raised €1.1 million in pre-seed funding to accelerate platform development and expand its product and engineering teams. The round was led by Syndicate One, with participation from Aikido founders Roeland Delrue and Willem Delbare, Showpad co-founder Louis Jonckheere, and serial entrepreneur Thomas Van Overbeke. Investment funds NewSchool.vc, RDY Capital, and 100IN also joined.
Holmes is currently working with 30 design partners who are actively shaping the product, with broader access rolling out in the coming months. The advisory group includes Dieter Wachters, Senior Director of Engineering at Collibra, and Haroen Vermylen, among other experienced operators from the Ghent tech ecosystem.
Who Gets the Most Value:
Holmes is built for software teams already using AI to write code faster and need their QA process to match that pace. If your team spends real time writing and maintaining test scripts, or if bugs regularly reach users because testing gets deprioritized, automated software testing at this level is a practical solution to explore.
Product managers who have inherited QA responsibilities on top of their core work will find the autonomous model useful. The platform is equally relevant for early-stage startups building their QA process for the first time and for larger product teams that have hit the limits of manual testing.
Teams that want to ship fast without sacrificing reliability now have a tool built specifically for that.
