Skip to content

Manifold

San Diego startup providing AI detection-and-response (AIDR) that monitors autonomous AI agents at runtime on enterprise endpoints.

Visit Website ↗ + Add to Compare Claim This Company
47/100Emerging / Unranked

Overview

Manifold builds an AI Detection and Response (AIDR) and governance platform that gives security teams runtime visibility into what autonomous AI agents actually do on enterprise endpoints — the tools they call, the systems they access, and the actions they take, including connections to MCP servers, databases and external systems. The product flags anomalies when agent behavior drifts from expected patterns, aiming to close a gap left by traditional EDR tools that were not built to interpret agent-driven activity.

Manifold was founded in 2025 by Neal Swaelens, Oleksandr Yaremchuk and Michael McKenna, the team behind Laiyer AI and its open-source LLM Guard project, which was acquired by Protect AI in January 2024. The company, headquartered in San Diego with about 17 employees, raised an $8 million seed round in March 2026 led by Costanoa Ventures, with participation from Cherry Ventures, Rain Capital and Modern Technical Fund, and angel backing from former Uber CSO Joe Sullivan and former Google DeepMind CISO Vijay Bolina.

Innovation Matrix Assessment

Innovation Velocity 5/10

Founding team has prior exit experience (LLM Guard/Protect AI acquisition) and moved from founding to a funded, named product within roughly a year.

Operational Value 6/10

Targets a real and growing operational gap — legacy EDR does not interpret agent-driven endpoint activity — with credible security-industry angel backers validating the thesis.

Market Momentum 3/10

One $8M seed round, ~17 employees, no publicly named enterprise customers yet disclosed.

Category Disruption 4/10

Applies detection-and-response thinking to a new class of endpoint activity (agents) rather than inventing a wholly new security discipline; multiple well-funded competitors target the same problem.

Real-World Efficacy 3/10

No independent third-party testing or named customer results disclosed publicly.

Enduring Relevance 7/10

Agentic AI on enterprise endpoints is a rapidly growing exposure, and the underlying monitoring need is likely durable over the next several years.

Why CISOs Should Care

Gives SOC teams a way to see and control what AI agents are actually doing on managed endpoints, an activity class current EDR tools were not designed to interpret.

What Makes It Different

Team's prior open-source LLM Guard project gives them applied experience in LLM-specific threat detection, now extended to runtime agent behavior rather than static model scanning.

The Matrix Verdict

47/100 — EMERGING / UNRANKED

Incremental Innovator: a credible, previously-exited founding team and notable security-industry angel validation, but still pre-revenue-disclosure with no independent efficacy evidence.

Editorial Note: Claims vs. Verified Findings

Funding, investors, founder background and headcount are independently confirmed by SecurityWeek, SiliconANGLE and the San Diego Business Journal; effectiveness claims are vendor-stated.

Sources