Onyx Raises $113M for the AI-Agent Control Layer
Onyx Security raised $113 million to help enterprises discover AI agents, observe what they do and stop actions that violate policy.
Enterprises are deploying AI agents across their software stacks, but many still lack a clear view of which agents exist or what they are doing.
Onyx Security has raised $113 million in Series B funding to build a control layer for that problem.
What happened
The round was led by Bessemer Venture Partners and brings Onyx’s total funding to $153 million.
Its platform is designed to discover the AI agents used inside a company, monitor their actions and intervene when activity conflicts with company policy.
Why it matters
An AI agent can move across applications, retrieve sensitive information and take actions on a user’s behalf. That creates a security problem that looks different from protecting a laptop or checking whether a human login is genuine.
Companies need to answer basic questions: Which agent acted? Who authorised it? What data did it access? Was the action within policy? Can it be stopped?
Onyx is betting that these questions justify a dedicated category of security software rather than another feature hidden inside existing dashboards.
The bigger picture
The agent-security stack is forming quickly. Some companies focus on identity and temporary credentials. Others enforce permissions before software runs. Onyx is concentrating on discovery, observation and policy control across the agent lifecycle.
Onyx says its revenue quadrupled within four months of emerging from stealth, but that claim and the platform’s effectiveness have not been independently verified. The notable signal is how much capital is now chasing AI-agent governance. The market is moving from “can agents work?” to “how do we keep them under control?”