About Zenity
Zenity sells an AI agent security platform built to govern how agents behave, not just what they say or generate. Zenity governs how an AI agent is built, what it can access, and what it does. The Zenity platform spans discovery, policy, identity, and runtime defense. Zenity brings observability, enforcement, and protection under one roof.[1][4][5]
Zenity states that AI agents get created inside SaaS platforms, wired into custom frameworks, and spun up on individual laptops, often without security ever finding out. Zenity describes a malicious prompt hidden in a document, an email, or a webpage, which the agent executes with the access it has been given. Zenity describes an agent that hits a problem, improvises, finds a credential it was never given, and acts. Zenity states that every agent registers and connects to MCP servers on its own, with no shared point of control.[1][5]
Zenity's Surface layer builds a live inventory of agents, evaluates how they are configured, and tests what is exploitable. Zenity's Enforce layer acts at the moment a decision gets made, letting a safe action through and stopping an unsafe one. Zenity's Protect layer watches what happens next, catching anything that slips through. AI Observability is Zenity's discovery and visibility layer for AI agents. AI Observability builds a live inventory of every agent across an organization's environment and tracks the data those agents access. Data Lens surfaces sensitive files and pages that are both frequently accessed and broadly exposed across agents.[4][5]
AI Security Posture Management (AISPM) is Zenity's posture layer for AI agents. AISPM evaluates configuration, permissions, and integrations before an agent goes live, and keeps enforcing guardrails automatically as agents evolve. AISPM applies stricter enforcement in production and more flexible policy in development or sandbox environments. Zenity validates which of an agent's attack paths are exploitable, scores each one, and ships a fix ready to apply in Runtime Boundaries.[4][7]
Boundaries is Zenity's runtime enforcement engine. Boundaries evaluates every agent action in real time against rules of an organization and decides to let it through, block it, or shut the agent down. A rule written once in Boundaries is enforced on Copilot, ChatGPT Enterprise, Agentforce, Bedrock, coding agents, and homegrown builds. Boundaries includes a built-in AI assistant that turns a plain-language request into a working rule. Boundaries flags rules that duplicate or contradict an existing one before a user activates it. AI Detection and Response (AIDR) is Zenity's runtime security layer for AI agents. AIDR reads the full sequence, not just the individual moments, detecting threats mapped to the OWASP LLM and MITRE ATLAS frameworks and blocking unsafe actions in real time. AIDR detects direct and indirect prompt injection and blocks execution before the agent acts on them. AIDR flags permission changes and other mutating actions an agent takes outside its expected scope. AIDR breaks every agent interaction into individual steps, from user messages to tool calls to retrievals, and groups them by thread in a single activity table.[1][4][8][9]
Zenity gives security and platform teams a single place to surface, enforce, and protect all MCP usage, built on the same Boundaries engine and platform that govern agents. Live View shows each request, the tool invoked, and the identity behind it, in real time. Zenity allows, modifies, or blocks a tool call based on live context. Zenity adds servers from its inventory or from public registries, so both known and shadow servers surface in one place. Zenity pulls in each agent's assumed identities from Entra ID and Okta, with managed or unmanaged status and privilege level attached. Zenity uses identity to gate what an agent can do, such as blocking a deactivated Okta user or restricting privileged tools by role.[10]
Zenity covers security and governance across SaaS, home-grown agentic platforms (Cloud), and end-user devices (Endpoint). Zenity secures AI agents embedded in productivity tools like Salesforce Agentforce or built with Copilot Studio. Zenity secures home-grown AI agents on platforms like AWS Bedrock and Google Vertex AI. Zenity gives control over local agents with lightweight monitoring, detection, and response. Zenity names Fortune 500 and Fortune 50 enterprises among the organizations whose AI agent deployments it secures. Zenity describes a global financial services firm that blocks any agent message that references a named confidential project. Zenity describes a global asset manager that blocks tool calls to specific third-party domains across every agent with one rule. Zenity, Microsoft, and Slalom are partnering to drive secure AI agent adoption for financial services and beyond.[1][4][7][8][11]
Zenity launched a security and governance platform for low-code/no-code development. Zenity introduced support to secure and govern Microsoft Copilot for M365. Zenity exited stealth with a $5M seed round led by Vertex Ventures and UpWest. Zenity raised a Series A round of $16.5M led by Intel Capital. Zenity took in strategic investment from Microsoft's investment arm, M12. Zenity raised a $38M Series B round, co-led by Third Point Ventures and DTCP. Zenity raised $125 Million in Series C funding led by Norwest. Zenity Labs delivers original research, threat intelligence, and hands-on experimentation focused on the emerging risks of AI Agents. The Zenity Security Assessment Hub, built by Zenity Labs, offers 10 free, open-source tools to help security teams identify and understand immediate risks.[1][4][11][12][13]
Zenity on the record