Zenity AI Agent Security and Governance Platform

Zenity is an AI agent security company founded by Microsoft cloud security veterans Ben Kliger and Michael Bargury, operating as Zenity Ltd. with offices in Tel Aviv, San Francisco and Tokyo. It started in 2021 with a security and governance platform for low-code and no-code development and shifted to enterprise copilots and AI agents from late 2023. It addresses the risk that agents with broad access are manipulated by hidden prompts, receive over-shared data or take destructive actions on their own.

The platform is organised in three layers. Surface builds a live inventory of agents across SaaS, cloud and endpoints, evaluates configuration and permissions (AISPM) and validates which attack paths are exploitable. Enforce applies runtime boundaries, agentic identity controls and MCP security at the moment an agent acts. Protect provides AI detection and response that analyses the full execution path, including tool calls, memory and data use, mapped to OWASP and MITRE ATLAS. It covers Microsoft 365 Copilot, Copilot Studio, Salesforce Agentforce, ServiceNow, ChatGPT Enterprise, Amazon Bedrock and Google Vertex AI for Fortune 500 enterprises.

• AI observability: Builds a live inventory of agents across SaaS, custom and endpoint deployments and the data each touches.
• AI security posture management: Checks agent configuration and permissions against policy before go-live.
• AI exposure management: Validates and scores exploitable attack paths and ships ready-to-apply fixes.
• Runtime boundaries: Enforces policy at the moment an agent acts and blocks unsafe actions.
• Agentic identity and MCP security: Governs agent identities, access and MCP server connections.
• AI detection and response: Analyses tool calls, memory and data flows mapped to OWASP and MITRE ATLAS.

A Vietnamese bank rolling out Microsoft 365 Copilot can find agents built in Copilot Studio that over-share customer data and fix them before go-live. A conglomerate using Salesforce Agentforce for customer service can block zero-click prompt injection that tries to exfiltrate CRM records. An IT company can monitor coding agents on developer laptops and their MCP connections.

Security teams see which agents exist, who owns them and what data they can reach across SaaS, cloud and endpoints in one view. Guardrails stop over-permissioned agents before they reach production; a Fortune 200 consulting customer reports a 90% reduction in security violations and 95% of high-risk violations remediated automatically. Intent-based detection catches attacks that prompt filters miss, and a Fortune 50 financial services customer reports an 80% risk reduction across a tenant of more than 150,000 resources.

Enterprises are rolling out Microsoft 365 Copilot, Salesforce Agentforce and custom agents faster than security teams can track them, and much of this is built by non-developers. Analysts have started to treat AI agent governance and AI application security as their own categories. Banks, insurers and large groups in Vietnam and Southeast Asia that already run Microsoft 365 and Salesforce face the same need to inventory and control agents.

Zenity exited stealth in November 2021 with a 5 million USD seed round led by Vertex Ventures and UpWest, raised a 16.5 million USD Series A led by Intel Capital in 2023, took strategic investment from Microsoft's M12 in 2024 and raised a 38 million USD Series B co-led by Third Point Ventures and DTCP in October 2024. It was a 2025 Gartner Cool Vendor in Agentic AI TRiSM and a 2026 Market Shaper in Gartner's AI Application Security quadrant.