Key facts
- Capital: Co-founder John Marshall committed US$100m of founder capital in September 2025, made up of US$50m already invested and a further US$50m.
- Ownership: Airia is privately held and funded by John Marshall, who previously co-founded AirWatch, later acquired by VMware.
- Offices: Airia is based in Atlanta with teams in Singapore, London, Dubai, Melbourne, Sophia and Bangalore.
- Scale: Airia had more than 300 enterprise customers and 150 employees across five continents by September 2025.
- Analyst coverage: Gartner positioned Airia as a Visionary in its inaugural Magic Quadrant for AI Governance Platforms in June 2026, and ranked it first for the AI security use case in the companion Critical Capabilities report.
- Runtime security: In August 2026 Airia achieved conformance under Autonomous Action Runtime Management (AARM), a Cloud Security Alliance open specification for checking AI agent actions before they execute, and launched inline Security Runbooks that pause a risky tool call and return a resume, block or escalate verdict.
- Integrations: In July 2026 Airia connected to Anthropic's Claude Compliance API to give security teams visibility of Claude Enterprise use, with real-time policy enforcement for Claude Enterprise generally available from September 2026.
Focus areas
About
Airia is an Atlanta-based company founded in 2024 that sells a single platform for building, securing and governing AI agents across an enterprise. Security teams use it to discover AI use, classify risk, red-team models and enforce guardrails, while AI teams use it to build agents, route requests across models and control cost. More than 300 enterprise customers had adopted it by September 2025; its users include KOA, Northwestern, ArcelorMittal, BuzzFeed, Learning Care Group and Mars.
How they use AI
Airia discovers and inventories AI tools, models and agents in use, scores their security posture and risk class, and red-teams them. At runtime its MCP gateway, agent constraints and guardrails govern what agents can call and return, while model routing and cost controls pick models per task and dashboards produce compliance evidence. Since August 2026, inline Security Runbooks can intercept a risky agent tool call and run a multi-step review before it resumes, is blocked or is escalated.
Products and services
- AI security — AI discovery, inventory, security posture, risk classification and red teaming across models and agents
- MCP Gateway — Central control point for agents' Model Context Protocol tool connections, with agent constraints and guardrails
- AI governance — Governance dashboard and compliance reporting across AI use
- Agent Builder — Tools to build and orchestrate enterprise AI agents
- Model routing and cost optimisation — Routes requests across models and controls AI spend
Notable customers / partners
- KOA
- Northwestern
- ArcelorMittal
- BuzzFeed
- Learning Care Group
- Mars
Key people
- John Marshall — Co-founder and Chairman
- Kevin Kiley — Chief Executive Officer
- Andrew Clearwater — Chief Trust Officer
- Erich Stuntebeck — Chief Scientist