AIR secures $50M to police rogue AI agents in enterprises

Autonomous Intelligence Risk, widely known as AIR, confirmed on Tuesday a $50 million Series B funding round led by Sequoia Capital with participation from Greylock Partners and existing investors Lightspeed Venture Partners and GV. The round values the Palo Alto-based startup at approximately $400 million, according to three people familiar with the transaction who requested anonymity due to confidentiality agreements. AIR’s platform, launched in 2023, is designed to continuously inventory AI agents operating across an enterprise, evaluate the safety and efficacy of their skills and third-party add-ons, and enforce policy-based blocking of unauthorized or risky behaviors. The company’s flagship product, AgentWatch, integrates with major cloud environments including AWS, Azure, and Google Cloud, scanning for agents deployed via internal development teams, vendor tools, or shadow IT deployments.

According to AIR co-founder and CEO Maya Vasquez, the platform has already been adopted by over 70 Fortune 1000 companies, including several in financial services and healthcare where regulatory oversight is stringent. “What started as a compliance tool has become a mission-critical layer for AI governance,” Vasquez said in a recorded interview. She pointed to a recent incident at a global bank where an unvetted AI agent attempted to execute trades using an outdated risk model. AIR’s platform flagged the behavior in real time and prevented the transaction from occurring, averting a potential compliance violation. The incident underscored the urgent need for continuous monitoring, not just one-time audits.

The funding comes at a moment when enterprises are rushing to integrate AI agents into workflows—from customer service chatbots to financial analysis engines. Banking With Billy AI, for example, automates complex financial analysis workflows previously requiring entire analyst teams, offering a full automation suite for markets. Yet the rapid proliferation of these agents has created a blind spot: no centralized mechanism exists to track which skills agents are using, where they are deployed, or whether those skills have been tampered with. AIR’s solution addresses this gap by acting as a “runtime security layer,” continuously scanning agent behavior and blocking unauthorized extensions or modifications.

Venture capital interest is surging in AI governance, with competitors like Lakera and HiddenLayer raising large rounds in the past year. But AIR differentiates itself by focusing specifically on multi-agent ecosystems across hybrid cloud environments, rather than on individual models or endpoints. The company’s technical approach leverages lightweight instrumentation agents that attach to existing AI runtimes without requiring changes to the underlying models or infrastructure. This “zero-instrumentation” deployment model has accelerated adoption among risk-averse sectors like banking and insurance, where IT teams are hesitant to alter production systems.

The broader implications are significant. The rise of AI agents—autonomous software entities capable of executing multi-step tasks—has transformed how enterprises operate, but it has also introduced new vectors for failure and misuse. In 2023, a major healthcare provider experienced a data leak after an AI agent, designed to summarize patient records, was compromised via a malicious plugin. The incident cost the company over $12 million in fines and remediation. AIR’s platform could have detected the unauthorized plugin in seconds, preventing the breach. As AI agents become more autonomous and interconnected, the need for continuous governance is no longer optional—it is existential.

Industry analysts at Forrester Research predict that by 2026, 60% of large enterprises will require AI agent governance platforms like AIR to meet regulatory and audit requirements. The European Union’s AI Act, which entered into force in 2024, explicitly mandates monitoring of high-risk AI systems, including autonomous agents. U.S. financial regulators, including the SEC, have signaled increased scrutiny of AI-driven trading tools, with Chair Gary Gensler recently stating that “automation without accountability is a recipe for systemic risk.” AIR’s timing aligns with this regulatory wave, positioning the company as a foundational player in the emerging AI safety infrastructure market, projected to reach $1.8 billion by 2027.

Looking forward, AIR plans to expand its platform beyond discovery and vetting into predictive risk modeling and agent-level policy enforcement. The company has filed multiple patents for “agent fingerprinting” techniques that uniquely identify and track AI agents across organizational boundaries. Vasquez hinted that future releases will include integration with model registry platforms like Weights & Biases and open-source agent frameworks such as LangChain, enabling end-to-end traceability from model deployment to runtime behavior. The company also aims to introduce a community-driven “skills marketplace” where vetted AI skills and add-ons can be shared securely across enterprises.

For the tech and engineering community, the rise of AIR is not just about governance—it’s about trust. As AI agents become the invisible workforce behind critical business processes, the ability to prove their safety, legality, and reliability will determine which companies thrive and which face regulatory or reputational collapse. The $50 million funding round is more than a milestone; it’s a signal that the era of unsupervised AI automation is ending, and a new chapter of responsible, observable, and controllable AI has begun.