Manifold
Detection and response for AI agents on employee endpoints: maps what each agent touches, baselines its behavior, and stops the ones that go rogue.
About Manifold
Manifold watches what AI agents do on employee machines rather than what they say.
It records agent actions such as tool calls, command execution, file access, and connections to MCP servers and databases, builds a behavioral baseline per agent, and flags deviations as they happen.
Security teams get a live map of which agents run where and what they can reach, a risk view of privilege paths and delegation chains, and controls to quarantine or terminate a rogue agent.
It runs on existing infrastructure, with no proxy or gateway in the traffic path. The founders previously built LLM Guard at Laiyer AI, which Protect AI acquired in 2024.
Investors
Work at Manifold? Link to this listing
Optional badge for your site or README. It changes nothing about the listing, its order or its review. Other formats and a dark version: press kit.
<a href="https://securitystartups.com/companies/manifold-security"><img src="https://securitystartups.com/badge/listed.svg" alt="Listed on Security Startups" width="184" height="44"></a>