Manifold is a cybersecurity company focused on runtime security for AI agents operating across enterprise endpoints and applications. The platform is designed to provide visibility and protection for autonomous AI agents, which traditional endpoint detection and response (EDR) tools are not built to monitor. The core platform is deployed agentlessly and provides runtime visibility into AI agent behavior, covering: - Discovery of AI agents running across endpoints and first-party applications - Monitoring of MCP (Model Context Protocol) servers, tools, and resources invoked by agents - Risk and exposure assessment for agent behaviors such as unauthorized system access and improper data exposure - Detection and response capabilities for anomalous or risky agent activity Manifold targets enterprise security teams, particularly CISOs and IT security professionals, who face challenges with employees connecting AI tools to internal systems without IT approval and with AI coding agents accessing codebases and production systems outside of traditional security controls. The platform addresses the gap between existing security tooling and the non-deterministic, dynamic nature of modern AI agents. It integrates with existing observability toolchains and protocols, aiming to provide oversight without impeding AI agent workflows. Manifold has received backing from investors including Costanoa, Rain Capital, Cherry Ventures, and mtf.vc, and has garnered endorsements from former security leaders at CrowdStrike, Google, Meta, Uber, and Cloudflare.