Geordie is an agent-native security and governance platform that helps enterprises safely adopt AI agents by acting as what its team calls air traffic control for a company's agents. As organizations rush to deploy agents across many tools and frameworks, Geordie's thesis is that security and governance teams lack visibility into what agents exist, what they can do, and whether their behavior aligns with policy, a gap that grows more dangerous as agents gain autonomy.

The platform is vendor-agnostic. It discovers deployed agents across all tools and frameworks, then maps their behavior and configurations to continuously assess risk rather than relying on point-in-time checks. At the core is Beam, Geordie's real-time risk-mitigation engine, which contextually guides agent decisions as they happen, keeping agents functional and aligned with enterprise policy without disrupting operations.

Geordie was founded in early 2025 by a team of cybersecurity veterans: CEO Henry Comfort, former COO Americas at Darktrace; Hanah-Marie Darley, former director of security and AI strategy at Darktrace; and Benji Weber, a former senior director of engineering at Snyk. That pedigree, spanning AI-driven security and developer tooling, shapes Geordie's focus on governing agents in real production environments.

The company emerged from stealth with a $6.5 million seed round co-led by Ten Eleven Ventures and General Catalyst, with participation from leading angel investors. It later raised a $30 million Series A to expand its platform as enterprises scale agent deployments. Geordie also won the RSAC Innovation Sandbox, a prominent industry recognition for emerging security companies.

Geordie competes in the rapidly forming agentic-AI governance and security category alongside companies building control planes and trust layers for agents. Its differentiation lies in continuous, vendor-agnostic discovery and risk mapping paired with real-time decision guidance, positioning it as a governance substrate for enterprises that want to adopt agents without losing control.