Artificial intelligence detection and response platform startup Manifold announced today that it has raised $8 million in funding to fuel development of its platform.

The platform is designed to protect companies from the risks of expanding autonomous AI usage across enterprise endpoints and enable enterprise employees to securely embrace agentic AI at large scale. Manifold is seeking to address a gap in existing security models, which primarily focus on user activity or model inputs and outputs but do not track the runtime actions of agents that can execute commands, access systems and interact with external tools.

The company’s platform centers on observing agent activity directly on endpoints, where agents operate with access to local resources, development environments and connected infrastructure. It works by capturing agent behavior on endpoints as it occurs, without requiring new infrastructure or proxy layers.

It records the actions agents take, including application programming interface calls, file access, command execution and interactions with external services such as Model Context Protocol servers and databases. The telemetry is analyzed in real time to establish behavioral baselines and detect deviations to allow security teams to monitor agent activity beyond the inference layer where traditional AI security controls are applied.

Manifold’s system provides visibility into the relationships between agents, tools and systems and maps how agents connect to internal and external resources to enable teams to identify which agents are active, what permissions they use and how they interact with production systems and development pipelines. It also supports anomaly detection by comparing observed behavior against established norms, helping surface unexpected actions such as unusual access patterns or unauthorized system interactions.

The technology is intended for environments where AI agents operate on employee endpoints, including coding agents with access to source code, production systems and continuous integration and continuous delivery pipelines.

The company was founded by Neal Swaelens, Oleksandr Yaremchuk and Michael McKenna, who formerly worked on AI security technologies, including the LLM Guard project developed at Laiyer AI. Following Laiyer AI’s acquisition by Protect AI Inc. in January 2024, the founders focused on extending security coverage to agent-based systems

The seed funding round was led by Costanoa Ventures. Also participating were Cherry Ventures Management GmbH, Rain Capital, Modern Technical Fund, former Uber Technologies Inc. Chief Security Officer Joe Sullivan and former Google DeepMind Chief Information Security Officer Vijay Bolina.

“There’s an open window to define the category for agentic security now, but it won’t be open long,” said John Cowgill, a partner at Costanoa Ventures. “Endpoint agent security is the next major layer of enterprise infrastructure. This team already built foundational AI security tooling and deployed it at massive scale. They know precisely where the previous generation fails. We believe they will own what comes next.”

Image: SiliconANGLE/Ideogram