Zenity Review

Industrial cybersecurity

TL;DR

Last updated: November 3, 2026. Testing track: Documentation Reviewed.

Zenity secures and governs AI agents built or used inside the enterprise, from SaaS agent builders like Microsoft Copilot Studio and Salesforce Agentforce to cloud agents and local coding agents. It combines agent inventory and posture management with runtime boundaries and detection. It is not a workforce prompt DLP tool for consumer AI sites.

Score: 7.3 / 10.

Initial review November 3, 2026.

How This Review Was Conducted

This review is based on Zenity’s public documentation, product pages, pricing pages and press coverage as of November 2026. We have not had lab access or a vendor demo.

Scores use the same published rubric and formula as every review on this site. Sources are listed at the end of the review.

Review methodology and evaluation process

Score breakdown

Dimension

Coverage breadth

Detection accuracy

Deployment friction

Policy & control depth

Framework alignment

Pricing transparency

Support & documentation

Weight

20%

20%

15%

15%

10%

10%

10%

Score

8

7

7

8

7

5

8

Notes

Wide agent coverage across SaaS, cloud, endpoint and MCP, but no workforce browser or consumer AI site coverage.

Documents prompt injection, tool misuse and data exposure detection with analyst praise, but no public benchmarks.

API connectors for SaaS agents are light, while local agent coverage needs an endpoint component and admin setup per platform.

Deterministic allow, modify or block boundaries, kill switch and automated remediation are granular.

OWASP and MITRE ATLAS work is documented; NIST AI RMF, ISO 42001 and EU AI Act mappings are not.

Marketplace listings exist but show no real price, so pricing is effectively quote only.

Public docs with integration guides, release notes and an active research team.

Coverage breadth

Weight
20%

Score

8

Notes

Wide agent coverage across SaaS, cloud, endpoint and MCP, but no workforce browser or consumer AI site coverage.

Detection accuracy

Weight
20%

Score

7

Notes

Documents prompt injection, tool misuse and data exposure detection with analyst praise, but no public benchmarks.

Deployment friction

Weight
15%

Score

7

Notes

API connectors for SaaS agents are light, while local agent coverage needs an endpoint component and admin setup per platform.

Policy & control depth

Weight
15%

Score

8

Notes

Deterministic allow, modify or block boundaries, kill switch and automated remediation are granular.

Framework alignment

Weight
10%

Score

7

Notes

OWASP and MITRE ATLAS work is documented; NIST AI RMF, ISO 42001 and EU AI Act mappings are not.

Pricing transparency

Weight
10%

Score

5

Notes

Marketplace listings exist but show no real price, so pricing is effectively quote only.

Support & documentation

Weight
10%

Score

8

Notes

Public docs with integration guides, release notes and an active research team.

What it does well

Covers many agent platforms and coding agents.

Broad agent platform coverage spanning Copilot Studio, Agentforce, ChatGPT Enterprise, Bedrock, Foundry and local coding agents (Zenity platform page, Series C release).

Runtime Boundaries can block agent actions.

Runtime Boundaries can allow, modify or block an agent action before it executes, not only alert (Zenity platform page).

Recognized by Gartner for agent governance.

Strong analyst traction: Gartner 'Company to Beat in AI Agent Governance' (April 2026) and Market Shaper in the Emerging Market Quadrant for AI Application Security (September 2026), per Zenity releases.

Public docs, integration guides and SIEM connectors.

Public documentation with per-integration guides and SIEM connectors (docs.zenity.io). Active research arm, Zenity Labs, publishing agent vulnerabilities such as AgentFlayer (Zenity site).

Where it falls short

Pricing unpublished; Marketplace requires private offer.

Pricing is not published and the AWS Marketplace listing requires a private offer.

No documented workforce browser coverage.

No workforce browser coverage for employees pasting data into consumer AI websites is documented.

No public detection benchmarks or false positives.

No public detection benchmarks or false positive rates found. Mappings to NIST AI RMF, ISO 42001 and EU AI Act are not documented on pages reviewed. SaaS agent coverage depends on platform APIs and admin roles (for example Power Platform Administrator for Copilot Studio).

Best fit

Large enterprises that are building or rolling out many AI agents across Microsoft Copilot Studio, Salesforce Agentforce, ChatGPT Enterprise, cloud agent frameworks and developer coding agents, and want one inventory, posture and runtime control layer for those agents, owned by the security team.

Poor fit

Organizations whose main problem is employees pasting sensitive data into ChatGPT or other AI websites, or small teams wanting published self-serve pricing. Workforce AI usage control tools such as Harmonic, Nightfall or Prompt Security fit that need better; teams focused only on custom LLM app guardrails may prefer Lakera.

Pricing transparency

Not published. AWS Marketplace listing (SaaS, 12-month contract) shows a placeholder $1.00 entry and directs buyers to private offers. Also listed on Azure Marketplace.

Alternatives

See how it compares in our Best AI Agent and MCP Security Platforms 2026 ranking. The poor-fit section above names alternatives for buyers it does not suit, and our methodology explains how every product is scored.

What We Would Test in the Lab

If Zenity grants lab access, we would run the following scenarios and upgrade this review to Lab Tested.

Agent and MCP inventory.

Discovery of a known set of agents, MCP servers and tool permissions.

Tool-call abuse.

Prompt-injection scenarios that try to trigger unsafe tool calls through MCP.

Policy enforcement.

Block, warn, redact and allow behaviors against a configured policy.

Audit logging.

What is logged, what is not, and retention behavior.

SSO integration.

Microsoft Entra ID and Okta where supported.

Deployment.

Time from access to enforced policy in a representative environment.

Latency.

Added latency on standard prompt sizes.

Feature coverage at a glance

Based on public documentation. “Not documented” means we could not find it in public material, not that the product lacks it.

CapabilityZenity
Inline prompt DLPPartial. Detects sensitive data exposure in agent interactions with inline prevention for Copilot Studio; not a workforce prompt DLP for AI websites
Shadow AI discoveryYes. Shadow AI and agent inventory across SaaS, cloud and endpoints, including local coding agents
Blocking unapproved AI toolsPartial. Can block or shut down agents and isolate GPTs; blocking of AI web apps not documented
File upload inspectionPartial. Scans GPT knowledge files and agent knowledge sources for risk; user file upload inspection not documented
Secure AI workspace (sanctioned chat)Not documented. No vendor-provided chat interface
Built-in AI vendor risk scoringNot documented. Posture findings are per agent, not vendor risk ratings
Microsoft 365 connected AI app review and controlPartial. Governs Copilot Studio and Microsoft Copilot agents; general OAuth app review and revoke not documented
Google Workspace connected AI app review and controlPartial. Gemini Enterprise integration listed in docs; Workspace OAuth app governance not documented
Governance scorecard, attestation and board reportingPartial. AISPM posture and exposure management; board-ready reports and attestation not documented
AI agent and MCP securityYes. Core focus: MCP Security, Agentic IAM, Runtime Boundaries, Guardian Agents
Prompt injection defenseYes. Detects direct and indirect prompt injection and RAG manipulation in agent runtime
AI red teamingPartial. Docs list an AI Red Teaming section; depth not reviewed
SIEM integrationYes. Docs list CrowdStrike Falcon Next-Gen SIEM and Cribl Stream; Splunk and Microsoft Sentinel adapters reported in March 2026 release notes

Recent changes

  • 2025-09: Expanded Microsoft Copilot Studio integration; Copilot Studio integration reported generally available November 2025.
  • 2025-12-02: Announced lifecycle security and governance for Amazon Bedrock AgentCore.
  • 2026-01-06: Available on AWS Marketplace.
  • 2026-04-17: Gartner report 'AI Vendor Race: Zenity Is the Company to Beat in AI Agent Governance'.
  • 2026-05-19: Selected for AWS Security Hub Extended.
  • 2026-08-03: $125M Series C led by Norwest.

Sources

Deployment and coverage

SaaS. API connectors to agent platforms (for example OpenAI Compliance API for ChatGPT Enterprise, Power Platform and Entra app registration for Copilot Studio), an inline webhook for Copilot Studio prevention, and a lightweight endpoint component for local agents. Available on AWS and Azure Marketplaces. Typical rollout time not published.

Agentic SaaS (Microsoft Copilot Studio, Salesforce Agentforce, ChatGPT Enterprise GPTs, Gemini Enterprise), cloud and custom agents (AWS Bedrock and AgentCore, Azure AI Foundry, Google Vertex AI), local and coding agents (GitHub Copilot, Cursor, Codex) via endpoint, and agent-to-MCP tool connections. Consumer AI web apps in the browser are not documented.

FAQ

Does Zenity secure Microsoft Copilot Studio agents?

Yes. Zenity documents a Copilot Studio integration, including inline threat detection and prevention via webhook, which requires a Power Platform Administrator role and an Entra app registration.

Yes. MCP Security is a named module, covering agent-to-MCP tool connections, alongside Agentic IAM and Runtime Boundaries.

Zenity does not publish pricing. It is sold through sales and via AWS and Azure Marketplace private offers.

No acquisition was found. Zenity announced a $125M Series C led by Norwest on August 3, 2026.