Palantir and ElevenLabs Executives Invest in New Israeli Startup Arrakis Security
Israeli startup Arrakis Security, founded by former Palantir and Torq employees, has raised $8 million in a seed round. The company provides a security layer to monitor and protect autonomous AI agents within corporate systems.
While the industry debates the security risks of autonomous code writers, a more passive problem is emerging under the radar of security managers: the AI agents running inside corporate systems. A new Israeli startup, founded by alumni of Palantir and Torq, was revealed today to address this issue with backing from prominent investors.
Securing agents in critical systems
"At Torq and Palantir, we saw that the real gap is not how to write code faster, but what happens when that same AI gets active execution permissions in critical systems," says Tal Bar-On, CEO and co-founder of Arrakis Security. "We are talking about agents that manage experiments, accelerate drug development, change settings on industrial production lines, or autonomously manage supply chains and financial systems."
According to Bar-On, existing security tools are currently blind in this area. Arrakis has created a security and management layer connected directly to the company's 'engine'. Unlike standard defense solutions, the company creates a security shell for agents operating in the cloud. The platform connects to dozens of runtime environments, such as Microsoft Foundry, Salesforce Agentforce, n8n, Make.com, and AWS Bedrock, analyzing fundamental logs including AI Reasoning.
Real-time behavioral monitoring
Access to deep logs allows the company to build a normative behavior model for each agent, identifying deviations from original tasks in real-time. "We are building and deploying dedicated AI agents whose role is to monitor and secure the AI agents operating in the organization," notes Bar-On.
In an early deployment at an organization with over 2,000 employees, the system mapped over 1,000 active AI agents in less than an hour—most of which were unknown to the security team. It identified dozens of policy violations and critical vulnerabilities, helping to resolve them during the POC stage.
Funding and team
Arrakis Security was founded in 2025 by Tal Bar-On, Omer Efrat, and Ron Shani. Today, the company announced an $8 million seed funding round led by the Israeli fund Hetz Ventures. Participants include Mati Staniszewski (CEO of ElevenLabs), Ofer Smadry (CEO of Torq), Amitai Ratzon (CEO of Pentera), and executives from Palantir.
Shani explains that he met Staniszewski while working in the "trenches" of Palantir on a massive project for BP. They dealt closely with reliability and safety challenges in AI while working on autonomous models for critical infrastructure. "When we set out to solve the burning challenge of securing autonomous agents, reconnecting with Mati was natural," he concludes.
