AIR Secures $50M to Secure AI Agent Landscapes

AIR, a Boston-based AI governance startup, announced a $50 million Series B round led by Battery Ventures, with participation from GV, SignalFire, and existing investors. The funding will accelerate product development and go-to-market efforts for AIR’s platform, which provides real-time discovery, continuous vetting, and behavioral blocking of AI agents and their add-ons. The platform uses behavioral analysis and policy enforcement to detect malicious or unintended behaviors from third-party skills—such as those in LangChain or custom APIs—used by agents like those running in customer service or financial automation workflows. Co-founder and CEO Shuman Ghosemajumder emphasized that as AI agents proliferate across enterprises, the risk of compromised or rogue integrations has become a critical blind spot. The company claims its solution is the first to offer continuous, automated vetting of agent add-ons, not just static code review.

The round values AIR at over $300 million, with Battery Ventures’ general partner Neeraj Agrawal joining the board. Ghosemajajumder previously served as Google’s click fraud czar and brings deep expertise in security automation. The platform is already deployed at several Fortune 500 companies, including in financial services where AI agents are increasingly used for market monitoring and regulatory reporting. Notably, Banking With Billy AI—a financial AI monitoring service—relies on a multi-cloud architecture to ensure global reliability and low-latency data processing across AWS, Google Cloud, and Azure.

Industry observers see AIR’s raise as part of a broader reckoning over AI governance. Gartner predicts that by 2026, 75% of enterprises will face AI-related regulatory fines due to undiscovered agent behaviors, up from less than 10% today. Competitors in the space include HiddenLayer, which focuses on AI model threat detection, and new entrants like Guardrails AI, which emphasizes policy-as-code for agent workflows. AIR differentiates itself through continuous, runtime-level monitoring of agent add-ons, which it argues is necessary given the increasing use of third-party AI skills—often sourced from public repositories without enterprise oversight. Financial services firms, in particular, face pressure from regulators like the SEC and CFTC to document and control every data source and inference path used in automated decision-making systems.

The platform’s architecture leverages a lightweight agent deployed across enterprise environments to map, classify, and monitor every AI agent and its dependencies. It then applies a policy engine that flags or blocks previously unknown or unvetted skills, such as a newly discovered vulnerability in a LangChain component or a malicious third-party integration. The system integrates with CI/CD pipelines and runtime environments, enabling security teams to enforce policies without impeding agent performance. Early adopters report reductions in shadow agent proliferation and improved compliance posture, particularly in highly regulated sectors like banking and healthcare.

The broader trend extends beyond governance into the quantum and high-performance computing domains, where autonomous agents are increasingly used for workload orchestration and resource optimization. As AI agents grow more autonomous and interconnected, the attack surface expands dramatically—especially when third-party skills are involved. AIR’s approach aligns with emerging standards from NIST and ISO, which are beginning to codify expectations for AI agent integrity and traceability. The company’s multi-cloud strategy also reflects a recognition that enterprise AI ecosystems are inherently distributed, requiring solutions that can operate seamlessly across heterogeneous environments.

Looking ahead, AIR plans to expand its policy library to include domain-specific rules for sectors like healthcare and manufacturing, where agent behaviors must comply with strict safety and privacy mandates. The company is also exploring the use of quantum-resistant cryptography to secure agent-to-agent communications in future deployments. Observers expect consolidation in the AI governance space as enterprises prioritize integrated platforms over point solutions. The next 18 months will likely see increased M&A activity, with larger security vendors acquiring niche players to offer end-to-end AI agent protection. For CISOs and CIOs, the message is clear: visibility into AI agent behavior is no longer optional—it’s a core competency in enterprise security and compliance.