Pluto Security Secures $42 Million for AI Agentic Workflow Defense
Israeli startup Pluto Security raises $42 million to defend enterprise AI agentic workflows by analyzing agent intent and enforcing security policies using endpoint telemetry.
Pluto Security, an Israeli startup founded in October 2025, has successfully raised $42 million in a Series A funding round led by Jump Capital. The company aims to secure the increasingly complex agentic workflows powered by artificial intelligence. By leveraging endpoint telemetry and local analysis models, Pluto seeks to understand the intent behind AI agent actions, detect sophisticated threats like prompt injection, and identify risks within the AI supply chain.
Shahar Bahat, co-founder and CEO of Pluto Security, explained that the funding will enable the company to provide comprehensive workspace security across various tools and access configurations. This includes monitoring execution commands, mitigating prompt risks, securing third-party ecosystems, and validating newly generated artifacts. Bahat emphasized the need for deep context into running processes, user behavior, and file contents to achieve effective security in dynamic environments.
Pluto differentiates itself by running lightweight AI models directly on endpoints. This approach avoids the CPU and latency overhead often associated with cloud-based solutions, thereby reducing deployment friction. Instead of merely offering visibility, Pluto focuses on active enforcement, automatically implementing guardrails such as mandatory authentication or internal hosting for employee-built applications.
Bahat highlighted the inherent risks associated with the endpoint environment and the challenges of deploying heavy security tools. "The endpoint is a very risky area," she stated, noting that inefficient tools can consume excessive resources and create user friction. Pluto's lean architecture, developed by experts in endpoint security, is designed to overcome these hurdles and facilitate seamless integration.
Understanding the intent behind an AI agent's actions is crucial, according to Bahat. This requires a rich dataset derived from endpoint activity, including running processes, file contents, and general user behavior. Pluto's platform approach integrates these insights from tools usage, endpoint behavior, and supply chain analysis, building an infrastructure adaptable to cloud sandboxes and future technological shifts.
The company addresses a wide range of security concerns, including access and identity risks related to email and collaboration tools, prompt injection, agent misbehavior, and risks associated with prompts and sessions. Pluto also tackles tampering and malicious code within text-based skills, packages, and extensions, ensuring visibility and control over newly developed agents, workflows, and applications.
Pluto categorizes workspace security into four layers: the tools themselves, how users interact with them, the broader ecosystem of third-party integrations, and the final artifacts produced. The company provides vetted, secure extensions and packages from a curated catalog, ensuring employees install approved components. For user-created agents, Pluto enforces secure deployment locations like internal servers or sandboxes, implementing default controls for authentication, secrets management, and integrations.
"It would not be fair to bring organizations visibility without enforcement capability, because there's nothing to do there," Bahat concluded. Pluto aims to provide both comprehensive visibility into user activities and associated risks, coupled with the ability to enforce security policies, ensuring a more robust defense against emerging threats in the AI-driven workspace.