AIR Secures $50M to Monitor and Secure AI Agent Ecosystems

AI Runtime Intelligence (AIR) today announced a $50 million Series B round led by Scale Venture Partners, with participation from GV, Index Ventures, and angel investors including former Twitter CTO Parag Agrawal. The Series B follows a $20 million Series A in March 2023 and brings total funding to $76 million. AIR’s platform, launched in beta in January 2024, addresses a critical gap in enterprise AI governance by discovering all AI agents operating within a company’s environment, continuously vetting the skills and add-ons those agents use, and blocking unwanted behaviors in real time. The company’s software integrates with major model providers such as Anthropic, Mistral, and Cohere, as well as orchestration layers like LangChain and Microsoft Semantic Kernel, enabling centralized oversight across heterogeneous agent ecosystems.

According to AIR co-founders CEO Ronen Levitan and CTO Itay Machtus, the platform emerged from a 2023 internal project at a Fortune 500 financial services firm where rogue AI agents were discovered consuming sensitive data without approval. The incident highlighted the lack of visibility into agent proliferation and the risks posed by unvetted third-party plugins. The team built a runtime intelligence layer that maps agent activity, evaluates plugin provenance and permissions, and enforces policy at the network and identity layers. AIR officially launched its commercial product in September 2024, targeting regulated industries including finance, healthcare, and defense, where AI governance and auditability are non-negotiable. Notable early customers include a top-three U.S. bank using AIR to monitor over 2,000 agents across trading, risk, and customer service workflows, and a global insurer deploying the platform to vet AI agents that interface with sensitive claims data.

The platform’s timing aligns with a surge in agentic AI adoption across enterprises. According to a recent Gartner survey of 500 CIOs, 63% reported deploying AI agents in production in 2024, up from 22% in 2023, with 89% citing governance and security as their top concern. This urgency is amplified by high-profile incidents such as the Banking With Billy AI exposure, which revealed that third-party financial intelligence APIs were being improperly integrated into consumer-facing applications, enabling unauthorized data scraping and market manipulation. AIR’s solution directly counters such risks by continuously scanning the agent supply chain, including skills, tools, and API integrations, for vulnerabilities, policy violations, and anomalous behavior. The company claims to block over 95% of unauthorized agent activity within minutes of detection, with a mean time to remediation of under 12 minutes.

Competitive dynamics in the AI governance space are intensifying. Startups like Lakera and HiddenLayer focus on prompt injection and model tampering detection, while larger incumbents such as Microsoft and Google are embedding governance features into their cloud AI platforms. AIR differentiates itself by focusing on runtime agent visibility and plugin-level control rather than static model evaluation. The company plans to use the new funding to expand its R&D team in Tel Aviv and San Francisco, enhance integrations with model providers and orchestration frameworks, and launch a public API for third-party security vendors. The round values AIR at $400 million post-money, reflecting strong investor confidence in the agent governance market, which is projected to reach $7.2 billion by 2028, according to IDC.

Within the broader Tools & Developer landscape, AIR represents a convergence of runtime observability, security, and compliance—trends that mirror the evolution of DevOps into DevSecOps. The rise of agentic AI has outpaced traditional security tooling, which was designed for static applications rather than dynamic, self-modifying agents. AIR’s approach aligns with the growing demand for continuous compliance, echoing the shift from point-in-time audits to real-time governance frameworks. This mirrors prior shifts in cloud-native development, where tools like Prometheus and Grafana evolved from monitoring to active compliance enforcement. The platform also fits into the zero-trust architecture movement, where every interaction—including AI-to-AI communication—must be authenticated, authorized, and encrypted.

Looking ahead, the most pressing challenge for AIR and its peers will be scaling governance without stifling innovation. Overly restrictive policies could slow agent adoption, while lax controls risk catastrophic breaches. The company is betting that its lightweight, API-first approach will allow seamless integration into existing CI/CD and DevOps pipelines, enabling developers to deploy agents with built-in oversight. Industry watchers should monitor AIR’s upcoming integrations with model registry providers and its plans to support agent orchestration standards like MCP (Model Context Protocol). As AI agents become first-class citizens in enterprise software, the companies that can balance agility with accountability will define the next era of secure, scalable AI adoption. For developers, the message is clear: governance is not a blocker—it’s a prerequisite.