Asymptote Labs, a New York startup, raises a $5.5 million seed round led by Tower Research Ventures to develop security capabilities for AI agent “runtime” behavior. The company commercializes Beacon, a tool it uses to record what AI agents do, and plans to expand a paid platform aimed at preventing or blocking risky agent actions.

Across outlets, the core details remain consistent: funding amount, the lead investor, and the product direction. SecurityWeek frames the news around “agent runtime security,” focusing on Beacon as the basis for a commercialization effort. The Next Web emphasizes that Beacon is free and open-source for recording agent activity, while the paid offering is positioned for monitoring and control. The Next Web also highlights intended use of the funding for engineering, hiring, and sales. Both accounts describe the round as a step toward making AI agent behavior more observable and safer during operation.