AIR Raises $50M to Vet Skills and Add-Ons for AI Agents

AIR, an artificial intelligence security startup, has emerged from stealth following the close of two sequential seed rounds totaling $50 million. The capital will fund a centralized governance platform designed to monitor and secure the rapidly expanding software supply chain surrounding enterprise AI agents. As organizations grant these autonomous systems broader access to internal databases and external networks, they are increasingly relying on third-party skills, plugins, model context protocol servers, and add-ons. This proliferation introduces significant exposure to supply chain compromises and content poisoning. Founded by Yair Saban and Niv Hoffman, both veterans of Israel’s Unit 8200 intelligence corps, AIR addresses this vulnerability through a three-tiered security architecture. The platform first identifies all active AI agents within an enterprise environment, including unapproved employee usage. It then deploys a runtime enforcement layer that intercepts and analyzes agent actions in real time. Crucially, AIR shifts the security paradigm from static scanning to continuous re-verification. The system cross-references every requested skill, plugin, and MCP server against a dynamically updated enterprise whitelist and AIR’s own vetted marketplace. By monitoring for package updates, developer account compromises, or malicious behavioral changes, the platform currently filters out approximately twenty-seven percent of externally sourced add-ons. The funding structure was executed rapidly, with Sequoia leading the initial ten-million-dollar tranche and Greenoaks headlining the subsequent forty-million-dollar round. The round featured participation from prominent angel investors and venture partners, reflecting institutional confidence in the category. AIR reports securing more than twenty enterprise clients to date, with roughly a quarter classified as large corporations. Demand is particularly acute in highly regulated sectors, notably financial services and pharmaceuticals, where compliance and data integrity remain paramount. The AI agent security landscape is becoming increasingly competitive, with several well-capitalized rivals including Noma Security, Zenity, Astrix Security, and Operant AI all pursuing similar discovery and governance capabilities. Despite this crowded field, AIR leadership maintains that its competitive advantage rests on the operational complexity of continuous ecosystem verification. Sequoia partners note that managing real-time inspections across an entire fleet of interacting agents represents an infrastructure challenge that extends beyond traditional security scanning. Operating with a team of approximately forty employees, AIR will deploy the newly raised capital to recruit specialized security researchers and accelerate go-to-market initiatives across the United States and Europe. Saban draws a direct parallel to the evolution of operating system security in the early two-thousands, when the industry mandated cryptographic signing for kernel-level drivers to prevent unauthorized code execution. He argues that enterprise AI tooling must adopt equivalent verification standards before autonomous agents can operate safely at scale. With the market shifting toward vendor-agnostic governance solutions, AIR aims to establish itself as the foundational infrastructure layer for secure AI agent deployment.