Secure AI Agents at Enterprise Scale

COMPANY OVERVIEW

Lasso Security is an Israeli startup headquartered in Tel Aviv, founded in 2023 to secure the rapidly evolving landscape of generative AI, large language models, and the autonomous agents built on top of them. The company was created around a simple observation: enterprises were adopting AI faster than their existing security controls could follow, and those controls were never designed for systems that reason, call tools, and act on their own.

In November 2023, Lasso emerged from stealth and announced $6 million in seed funding in a round led by Entrée Capital with participation from Samsung Next. Its platform is now used across public sector and commercial organizations including the US Department of Homeland Security, eToro, Fiverr, Kaltura, Nayax, Telit, Guesty, and Artlist.

CORE FOCUS

Lasso delivers end-to-end coverage for agentic AI: discovering every agent and AI application in the environment, assessing exposure and misconfiguration, adversarially testing models before they reach production, and enforcing policy on live traffic. The four disciplines run as one continuous loop rather than as separate point tools, so findings from red teaming feed posture management and posture gaps become runtime enforcement rules.

The differentiator is intent security — evaluating not just what an agent outputs but whether its behavior still falls within the scope it was authorized to operate in. Enforcement happens at the proxy and gateway layer with real-time decisioning, and the platform aligns to ISO 27001, SOC 2, GDPR, PCI DSS, NIST, and OWASP frameworks.

PRODUCTS & TOOLS

Discovery & AI-BOM – Inventories every agent and AI application across the enterprise, mapping the models, prompts, tools, and guardrails behind each one.

AI Security Posture Management – Analyzes exposure, misconfigurations, and policy gaps across the AI estate and prioritizes what to fix first.

Automated AI Red Teaming – Runs adversarial testing against models and agents using a library of over 3,000 attack patterns to surface vulnerabilities before deployment.

Runtime Enforcement – Applies policy at the proxy and gateway layer with real-time decisioning on live agent traffic.

AI Detection & Response – Detects and terminates threats against AI systems in production with a 98.6% accuracy rate.