AIR Security
AI agent firewall that vets skills, plugins, and MCP servers for malicious instructions and supply-chain risk before agents can act.
Visit Website ↗ + Add to CompareOverview
AIR Security builds a firewall purpose-built for AI agents rather than for the humans or servers around them. As enterprises plug large language model agents into tools, browser extensions, MCP servers, and third-party “skills,” AIR’s platform inspects those components for hidden malicious instructions, prompt-injection payloads, excessive permission requests, and software-supply-chain risk before they are allowed to execute inside an agentic workflow.
The company was founded in 2025 by Yair Saban (CEO) and Niv Hoffman (CTO), both veterans of Israel’s Unit 8200 offensive cybersecurity corps, joined by Ryan Knisley, former CISO of The Walt Disney Company and Costco Wholesale, as chief strategy officer. AIR is headquartered in New York with a substantial research and engineering presence in Israel. The company emerged from stealth in September 2026 with $50 million in funding led by Sequoia Capital and Greenoaks, alongside a roster of prominent security-industry angel investors.
What differentiates AIR from general AI-security posture tools is its narrow focus on the agent’s action layer: rather than monitoring outputs after the fact, it evaluates the components an agent is about to invoke — a plugin, an MCP server, a third-party skill — for malicious intent and blocks unsafe execution paths in real time.
Innovation Matrix Assessment
Went from founding to a $50M raise and a working product within roughly six months, addressing the MCP/agent-skill risk surface almost as fast as it has emerged.
Gives security teams a concrete control point — vetting agent skills and MCP servers before execution — for a gap most CNAPP and endpoint tools do not yet cover.
A $50M round led by Sequoia and Greenoaks with a credible ex-Disney/Costco CISO on the founding team is a strong signal, though customer names and renewal data are not yet public.
The AI-agent firewall is a genuinely new control layer rather than a repackaged existing category, but it remains a point solution addressing one slice of the agentic-AI risk surface.
No independent tests, named breach preventions, or third-party validation are publicly available yet; efficacy claims currently rest on vendor description alone.
Enterprise adoption of AI agents, plugins, and MCP servers is accelerating quickly, and the supply-chain risk this creates is unlikely to shrink over the next several years.
Why CISOs Should Care
It gives CISOs a way to vet third-party AI agent skills and MCP servers for malicious behavior before they're wired into production workflows, closing a blind spot most existing tools miss.
What Makes It Different
AIR inspects and gates the components an agent is about to execute, rather than monitoring agent output after the fact.
The Matrix Verdict
63/100 — INCREMENTAL INNOVATOR
AIR Security lands as an Incremental Innovator: a well-funded, well-pedigreed team addressing a real and fast-growing gap in agentic AI security, but still too new for independently verified efficacy evidence.
Editorial Note: Claims vs. Verified Findings
The $50M raise, investor names, and founder backgrounds are independently reported by multiple outlets (TechCrunch, SecurityWeek, Calcalist). Specific efficacy and detection-rate claims come from the company and have not yet been independently tested.
Sources
- SecurityWeek — https://www.securityweek.com/ai-agent-firewall-startup-air-security-emerges-from-stealth-with-50-million/
- TechCrunch — https://techcrunch.com/2026/09/01/air-raises-50m-to-help-companies-vet-the-skills-and-add-ons-ai-agents-use/
- Calcalistech — https://www.calcalistech.com/ctechnews/article/r13apdnugg
Alternatives to AIR Security
Adaptive Security
AI-driven platform that simulates deepfake, voice, and multichannel social-engineering attacks to train and test organizations against next-generation phishing.
Quilr
Early-stage agentic AI security startup building a 'Service-as-Software' platform to guard against human-related breaches and secure AI agent…
Zenity
Governance and security platform for AI agents and low-code/no-code development, securing agent identity, permissions and behavior across the…
Tenzai
An agentic AI penetration testing startup building autonomous 'AI hackers' to find and validate exploitable vulnerabilities at a…
Reco
Reco secures the "agentic ecosystem" — mapping what AI agents can access across SaaS and enterprise apps, detecting…
Charm Security
Agentic AI workforce that investigates and intervenes on scams and fraud in real time, reading manipulation and intent…