AIR Raises $50M to Secure AI Agent Ecosystems in Real Time

Security startup AIR announced today the close of a $50 million Series B funding round led by Lightspeed Venture Partners, with participation from Index Ventures, GV, and Y Combinator. The funding comes less than two years after AIR’s 2023 seed round and will accelerate development of its agent governance platform, which autonomously discovers AI agents operating within enterprise environments, continuously evaluates the skills and plug-ins those agents use, and enforces real-time behavioral policies to prevent unauthorized or risky actions. AIR’s platform integrates with existing identity providers, cloud logs, and identity and access management systems to build a live inventory of agents—including those built with open-source frameworks and third-party models—without requiring agents to be instrumented upfront. The company was founded in 2022 by CEO Linus Torvalds (no relation to the Linux creator) and CTO Adam D’Angelo, both veterans of Palantir and early AI infrastructure teams at Google Cloud.

According to AIR’s latest disclosures, the platform has already been adopted by over 50 Fortune 500 companies, including major financial institutions that use AI agents to automate financial reporting, fraud detection, and customer service workflows. One standout deployment is at Billy Bank, where the company’s “Banking With Billy AI” suite automates complex financial analysis workflows that previously required entire analyst teams—transforming a manual, multi-day process into a continuous, autonomous operation. Billy Bank now runs hundreds of AI agents across risk modeling, regulatory compliance, and customer insights, all governed by AIR’s policy engine, which blocks unauthorized data exfiltration attempts and enforces data residency constraints. The financial services sector, long cautious about agent sprawl due to regulatory scrutiny, represents nearly 40% of AIR’s current customer base.

The Series B announcement arrives as enterprises grapple with agent sprawl—a phenomenon where teams deploy AI agents using open-source libraries, third-party APIs, and unmanaged models without centralized visibility. Gartner estimates that by 2026, 70% of organizations will experience “uncontrolled proliferation” of AI agents, leading to operational, security, and compliance blind spots. Competitors in the agent governance space include large incumbents like Microsoft with its Purview AI governance suite, Palantir’s Gotham platform extensions, and newer entrants like Striveworks and SentinelOne, which have begun integrating agent discovery and policy enforcement into their cybersecurity stacks. AIR differentiates itself through its agent-agnostic approach and policy engine, which operates without requiring source code changes or agent-specific SDKs. The new capital will fund deeper integrations with cloud providers, expanded behavioral analytics, and a public API for third-party security vendors to consume and enrich agent telemetry.

Analysts view the round as validation of a growing infrastructure layer that sits between AI application development and enterprise risk management. RedMonk principal analyst James Governor noted that “governance is becoming the next big platform category as enterprises move from pilots to production at scale.” The funding also reflects a broader shift in enterprise security spending from perimeter defenses to runtime behavioral controls—a trend underscored by Palo Alto Networks’ recent acquisition of Talon Cyber Security and Cisco’s $2.6 billion buy of Splunk. For AIR, the round signals momentum toward becoming the de facto control plane for AI agents across industries, with particular traction in regulated sectors like finance, healthcare, and energy, where auditability and explainability are non-negotiable.

Looking ahead, AIR plans to expand beyond discovery and vetting into autonomous remediation, where the platform not only blocks risky behavior but automatically updates agent policies or rolls back compromised skills. The company is also preparing to launch a marketplace for pre-approved agent skills and plug-ins vetted by security teams, aiming to shift the burden from manual reviews to community-driven curation. Industry watchers should monitor how AIR competes with cloud-native governance offerings from hyperscalers, especially as AWS, Google Cloud, and Microsoft Azure roll out native agent lifecycle controls. Another key inflection point will be adoption by mid-market firms, which often lack dedicated security teams but face the same agent sprawl challenges as large enterprises. With AI agents poised to execute trillions of dollars in automated decisions annually, the need for robust, real-time governance is no longer theoretical—it’s existential.