Zenity raised $125 million to intercept AI agents before they act.

Zenity raised $125 million in a round led by Norwest, with SoftBank Vision Fund 2, Qumra, Hitachi Ventures and LG Technology Ventures joining prior backers Vertex, Third Point, DTCP and Intel Capital, per the August 3 announcement.

The product doesn't audit agents after something goes wrong. It sits between an enterprise AI agent and the action it's about to take, then allows, modifies or blocks that action before it executes. That's an interception layer, not a scanner, and it's a different job than the CASB and SSPM tools most security teams already have on contract.

Revenue tripled each of the past two years, and Zenity says it's tracking to triple again this year across a workforce that's now past 230 people. SoftBank Corp is listed as a customer, separate from SoftBank Vision Fund 2's stake on the cap table.

The take

A $125 million round for a pre-execution interception layer is a signal about how enterprises are categorizing this problem. They're treating agent governance as its own budget line, not a feature they expect their existing SaaS-security or CASB vendor to bolt on for free. That's a bad bet for any incumbent still planning to ship agent controls as a minor release note. Watch whether CASB and SSPM vendors respond by building their own interception layers or by trying to buy one.