The company says its technology checks AI-agent actions before they take effect and can stop or correct unsafe behavior.

Onyx Security raised $113 million to expand AI agent controls that it says can monitor what agents do and intervene before an unsafe action takes effect.

Bessemer Venture Partners led the Series B round, bringing Onyx’s total funding to $153 million since the company was founded two years ago. Cyberstarts, TCV, Conviction, FirstMark, Vintage, QuantumLight and G Squared participated in the round.

Taking control of AI agents

Onyx said the funding will help expand their platform, which is designed to control AI agents operating across company systems. The software identifies agents, monitors their activity, and checks proposed actions against company rules.

If an action violates those rules, Onyx says the platform can block it or redirect the agent before the action takes effect. The company is positioning that ability as an added safeguard for agents who can use business software, access sensitive information, and act with limited human direction.

Monitoring 1.1 million agents

Onyx said in an accompanying company blog post that its platform monitors more than 1.1 million agents and inspects more than 66 million AI sessions across its enterprise customers.

Onyx said it works with dozens of large companies, including Fortune 500 businesses in banking, technology, insurance and energy. Its announcement included a customer statement from Revolut and cited a June integration with Anthropic.

The company also said revenue quadrupled during the four months after it emerged from stealth. It did not disclose its starting or current revenue.

Onyx said it will use the new funding to improve its technology and expand sales in the United States and other markets.