Arrakis Security, a New York and Israel-based provider of an Autonomous Workforce Governance Platform, has raised $8M in seed funding led by Hetz Ventures. The platform delivers three-stage behavioral security for autonomous AI agents across SaaS, cloud, and endpoints through observability, posture management, and real-time detection with kill-switches. The capital will expand R&D and product teams in Israel while strengthening detection and accelerating deployments.
Agent Security Funding Wave Builds
The timing comes amid rapid capital inflow into AI agent security. Noma Security raised $100M Series B in July 2025. WitnessAI secured $58M in January 2026. NeuralTrust closed $20M seed in June 2026, while Manifold Security took $8M seed in March 2026. Arrakis differentiates through its three-stage architecture focused on runtime behavioral governance rather than discovery or prompt guardrails alone.
Visibility Illusion Leaves Enterprises Exposed
Enterprises deploy AI agents that read and write production data yet lack real-time inventories. Only 21% of organizations maintain real-time agent inventories. Gartner predicts over 40% of agentic AI projects will be canceled by end of 2027 due to inadequate risk controls. Traditional identity-based tools fail because agents operate within permissions but execute unexpected high-risk patterns.
Three-Stage Architecture Targets Runtime
Arrakis built AI Observability for automatic discovery and behavioral baselines, AI Security Posture Management with pre-execution analysis and MCP Gateway, and AI Detection & Response featuring Zero-Trust Output and granular kill-switches. Competitors like Zenity focus on low-code copilots while WitnessAI emphasizes governance layers. Arrakis adds cross-agent contagion tracking and Financial DoS circuit breakers that others do not explicitly target.
"You cannot secure a machine that thinks and operates differently from humans using security tools that were designed to monitor humans interacting with machines," said Tal Baron, co-founder and CEO.
Hetz Ventures Signals Category Conviction
Hetz Ventures led the round with participation from ElevenLabs CEO Mati Staniszewski, Torq CEO Ofer Smadari, Pentera CEO Amitai Ratzon, and Palantir executives. The firm has backed multiple agent security plays including Willow ($7M seed, June 2026) and Oak ($60M). This investor mix validates the thesis at the cybersecurity-AI infrastructure intersection.
Agentic AI Cybersecurity Market Accelerates
The cybersecurity agentic AI market stands at $2.43B in 2026 and is projected to reach $9.63B by 2031. The broader agentic AI in cybersecurity market is forecast to grow from $2.86B in 2025 to $63.29B by 2035. Major platforms including Salesforce Agentforce and Microsoft Copilot Studio are shipping autonomous agents, creating urgency for behavioral controls before production data access outpaces governance.
Founders Combine Palantir and Torq Experience
The founding team includes Torq veterans Tal Baron and Omer Efrat plus eight-year Palantir veteran Ron Shani. The company employs roughly 20 people and maintains R&D operations in Israel. Arrakis Labs, staffed by former Microsoft security researchers and elite military tech unit veterans, publishes open threat research.
Next Steps Focus on Design Partners
With design partners already on n8n, Make.com, and Agentforce, Arrakis plans to expand R&D hiring in Israel and accelerate enterprise deployments following the stealth exit.