Evoke provides an AI‑agent security platform that discovers, inventories, and governs every connected AI model, tool, and data source across an organization. By mapping agent skills, permissions, and attack paths, it enables real‑time detection and prevention of rogue actions, allowing security teams to see and control AI agents wherever they run—on endpoints, SaaS, or via SDKs. The solution offers per‑agent policies, threat modeling, and automated remediation of over‑permissioned agents.
Funding
Funding not disclosed
Founders
Product
Problem
Organizations increasingly rely on AI agents, models, and tools that connect to numerous data sources, creating a sprawling “shadow AI” environment where each agent can act as a single point of failure. Security teams often lack visibility into these agents, making it difficult to detect over‑permissioned or malicious behavior before a breach occurs.
Solution
Evoke provides a platform that automatically discovers and inventories every AI agent, model, tool, and data source within an enterprise. It maps agent capabilities, skills, and data connections to generate a comprehensive governance view and threat model that highlights potential blast radii. Real‑time detection and per‑agent policies enable the platform to block unauthorized or rogue actions as they happen. The solution integrates with agents wherever they run—on endpoints, production servers, SaaS services, or browser extensions—ensuring consistent control across the entire AI workforce.
Target Audience
Primary customers are enterprise security and risk teams responsible for protecting AI‑driven applications and data pipelines across large organizations.
Features
- Automated discovery and continuous inventory of all AI agents, models, tools, and data sources
- Mapping of agent skills, permissions, and data flows to visualize blast radius and attack paths
- Real‑time detection engine with per‑agent policies to block rogue or over‑permissed actions
- Threat modeling that identifies malicious skills and toxic data flows before exploitation
- Support for local endpoint agents, production agents, SDK/proxy APIs, SaaS agents, and browser extensions
- Centralized governance console for security teams to monitor, remediate, and enforce policies