Momentic, a Y Combinator W24 startup founded by Wei-Wei and Jeff (both formerly at Robinhood, Retool and other dev-tool companies), is launching Mo, an designed to autonomously test software applications. Rather than writing and maintaining traditional test scripts with hardcoded selectors, Mo uses an -based agent to explore app interfaces, plan test paths, and identify bugs in real time.
Mo differs from earlier approaches that tried to generate Playwright or Selenium tests via AI. Momentic's founders argue those tools force developers to hardcode implementation details, making maintenance costly. Instead, Mo builds a self-organizing graph of test actions and adapts as the app changes, trading some precision for better recall—the agent finds more real bugs. The startup prices it at $2.67 per bug caught, lower than Claude Code and Devin.
The founders say modern LLMs are strong at code generation but weak at visual understanding and tend to write redundant tests. Momentic built Mo on three years of testing research and millions of real-world test traces.
This is an AI-written summary. The original reporting belongs to momentic.ai. How we write these · Spot a mistake?