Manifold Raises $8M to Secure Autonomous AI Agents on Enterprise Endpoints

Why It Matters

Securing autonomous AI agents at the endpoint closes a critical gap as enterprises scale agentic AI, protecting production systems from hidden, runtime threats.

Key Takeaways

• $8M seed round led by Costanoa Ventures.
• Platform monitors AI agent actions on endpoints.
• No additional infrastructure or proxy required.
• Detects anomalies via real‑time behavioral baselines.
• Backed by former Uber and Google security leaders.

Pulse Analysis

The rapid adoption of autonomous AI agents—from code‑generation assistants to workflow orchestrators—has shifted the security perimeter from traditional user‑centric controls to the runtime behavior of software entities. While existing AI safeguards focus on model inputs, outputs, or API traffic, they often miss the actions an agent takes once it lands on a corporate workstation. This blind spot creates opportunities for privilege escalation, data exfiltration, and supply‑chain attacks that can bypass conventional detection tools, prompting a need for dedicated endpoint visibility.

Manifold’s platform addresses that blind spot by instrumenting the endpoint itself and streaming granular telemetry such as API calls, file reads, and command executions performed by AI agents. Because the solution operates without requiring new agents, proxies, or network taps, it can be deployed across heterogeneous environments with minimal overhead. Real‑time analysis builds behavioral baselines for each agent, flagging deviations that indicate malicious intent or misconfiguration. The approach also maps relationships between agents, tools, and internal services, giving security teams a clear picture of permission usage and potential attack paths.

The $8 million seed round underscores investor confidence that endpoint‑focused AI security will become a core layer of enterprise defense. Backers including Costanoa Ventures, Cherry Ventures, and former security chiefs from Uber and Google bring both capital and credibility, positioning Manifold to set standards for the nascent “agentic security” category. As organizations scale AI‑driven development pipelines, the ability to monitor and contain autonomous agents at the device level will likely become a compliance requirement, driving broader market adoption and spawning complementary solutions.