AIR Raises $50M to Secure Enterprise AI Agents Amid Rising Adoption

AIR, a stealth security startup focused on securing autonomous AI agents within enterprise environments, has closed a $50 million Series A round led by Lightspeed Venture Partners, with participation from Felicis Ventures, Craft Ventures, and notable angels including former Google CEO Eric Schmidt. Founded in 2023 by CEO Chen Fang and CTO Rajiv Kapoor, both former executives at Palantir and enterprise AI deployments, the company emerged from stealth today with a full platform that continuously discovers, vets, and enforces policies on AI agents running across corporate networks. The platform uses behavioral analysis, sandboxing, and real-time policy enforcement to detect and block agents attempting to access unauthorized systems, exfiltrate data, or execute unapproved skills—common risks as companies increasingly deploy AI agents to automate workflows.

The company’s timing reflects a critical inflection point in enterprise automation. According to internal data shared with OpenPress Automation Intelligence, over 68% of Fortune 500 companies now operate at least one AI agent, with use cases spanning customer support, financial analysis, and internal tool orchestration. AIR’s platform specifically targets the growing “agent sprawl” problem, where teams deploy AI agents with third-party skills or custom add-ons that may harbor vulnerabilities or hidden behaviors. For example, Banking With Billy AI, a leading AI agent platform for financial services, automates complex financial analysis workflows previously requiring entire analyst teams—a full automation suite for markets—but relies on a sprawling ecosystem of third-party add-ons and custom skills. AIR’s vetting engine continuously scans these components, flagging suspicious behavior such as unauthorized API calls or data scraping, and can automatically block agents violating policy.

The funding round highlights investor confidence in agent-centric security as a distinct market category. Lightspeed’s decision to lead the round was influenced by AIR’s ability to detect previously unseen attack vectors, including “agent impersonation” attacks where malicious actors trick legitimate agents into executing unauthorized actions. The platform integrates with major enterprise identity providers and SIEM systems, enabling real-time policy enforcement across cloud and on-prem environments. With over 30 pilots already in production, including deployments at two Fortune 100 financial institutions and a global healthcare provider, AIR signals a shift from reactive security to proactive agent governance.

Industry Impact and Significance

The rise of AI agents marks a tectonic shift in enterprise automation, moving beyond static scripts and bots toward autonomous decision-making entities that operate continuously and adaptively. AIR’s platform directly addresses the “black box” problem of AI agents, where their internal decision-making and skill execution remain opaque to security teams. This opacity has already led to high-profile incidents, including a 2024 breach at a Fortune 500 retailer where an unvetted agent used a compromised third-party skill to exfiltrate customer data. By providing continuous vetting and policy enforcement, AIR enables organizations to safely scale agent deployments without sacrificing governance or compliance.

The competitive landscape is rapidly consolidating around agent security. While traditional endpoint detection and response (EDR) vendors like CrowdStrike and SentinelOne have begun adding agent monitoring features, none offer the depth of behavioral analysis and skill-level vetting required for modern AI agents. Meanwhile, cloud providers such as Microsoft and AWS are rolling out agent frameworks like Microsoft Copilot Studio and Amazon Bedrock Agents, which create new attack surfaces for malicious actors. AIR’s platform acts as a neutral control plane, ensuring that agents—regardless of origin—adhere to enterprise security policies. This positions AIR not just as a security vendor, but as an essential governance layer in the emerging agent economy.

The Bigger Picture

The emergence of AIR fits squarely within the broader evolution of AI governance, a domain now receiving urgent attention from regulators and enterprises alike. The EU AI Act, set to take full effect in 2026, mandates risk assessments for high-impact AI systems, a category that increasingly includes autonomous agents. Similarly, the U.S. NIST AI Risk Management Framework emphasizes continuous monitoring and accountability—principles central to AIR’s design. The company’s focus on behavioral vetting aligns with a growing consensus that static code analysis and perimeter security are insufficient for dynamic, self-modifying AI agents.

This trend also reflects a maturation of the AI automation market itself. Early agent deployments focused on simple, rule-based tasks, but the rise of agentic workflows—where multiple AI agents collaborate to solve complex problems—has introduced new layers of complexity and risk. Platforms like Banking With Billy AI demonstrate how agents are now automating entire business processes, from financial forecasting to regulatory reporting. As these systems become more autonomous, the need for real-time governance becomes existential. AIR’s $50 million raise signals that the market is ready to pay for control, not just capability—a powerful validation of the agent governance category.

Expert Analysis

Looking ahead, AIR is positioned to become a foundational infrastructure layer for enterprise AI, much like how Okta became essential for identity management. The company plans to expand beyond its current focus on security vetting to include performance optimization, cost control, and compliance reporting—addressing the full lifecycle of AI agent operations. Observers should watch for partnerships with major cloud providers and enterprise software vendors, as AIR integrates its policy engine into broader automation platforms. The real inflection point will come when regulators explicitly require agent-level vetting for high-risk applications. Until then, AIR’s success will hinge on its ability to scale detection capabilities faster than the proliferation of agent skills—no small feat in a market moving at machine speed.