Air Security, which builds a security service for extensions and other tools installed on AI agents, emerges from stealth with $50M led by Sequoia and Greenoaks
Context & Ripple Effects
Air Security enters a fast-forming AI-security field in which Way Security is applying agentic workflows to IAM deployment, while Act Security is reducing cloud access exposure. Its focus on the tools an agent installs shifts attention from the agent itself to the software supply chain surrounding it.
The company’s finding that thousands of AI add-ons depend on untrusted sources gives its funding a defined problem to address. It also overlaps with Bloom Security’s monitoring of AI agents and extensions and Ent Security’s workspace-intent security layer, signaling that vendors are carving the market by control point rather than treating AI security as one category.
First-order effects
- Air Security can position vetting of agent extensions and tools as a distinct enterprise control, aimed at customers concerned about dependencies from untrusted sources.
- Sequoia- and Greenoaks-led funding gives Air Security resources to build and sell that control while the category’s other early vendors are also emerging from stealth.
Second-order effects
- Bloom Security must distinguish endpoint monitoring from Air Security’s tool- and extension-vetting approach where customers want coverage across the agent software supply chain.
- IAM and workspace-security vendors such as Way Security and Ent Security face pressure to show how identity and intent controls account for the tools agents invoke, not only the users or agents themselves.
Third-order effects
- If enterprises adopt separate controls for agents’ add-ons, the AI-agent security stack is likely to split among endpoint monitoring, identity and access, workspace behavior, and third-party tool assurance.
- The category points toward procurement processes that assess an agent’s permitted tools and their dependencies as part of deploying the agent, expanding the traditional software-supply-chain security boundary.
The trend: AI security is developing into a layered control market in which the agent’s tool chain becomes a security boundary alongside identity, endpoints, and cloud access.