The Atlas take.
Zenity is worth evaluating when the first problem is understanding which agents exist, how they are configured and what they can reach across supported enterprise environments. Surface, Enforce and Protect should be treated as distinct control areas rather than one undifferentiated prevention claim. Atlas would begin with discovery accuracy and a small number of validated permission findings, then examine the remediation path actually available. The useful outcome is an owner who can reduce a specific exposure and prove the change. An attractive agent inventory alone does not demonstrate runtime interception of an unsafe action.
02 / DEFINE THE BOUNDARY
What you are actually evaluating.
Scope the evaluation to the selected Surface connectors. Enforce and Protect capabilities require their own integration and entitlement review.
Zenity’s platform divides work into Surface, Enforce, and Protect. This profile starts with Surface: establishing an inventory and understanding agent configuration, access, and exposure. The useful next step is connecting a specific posture finding to the exact runtime control and remediation workflow that can address it.
- Also covers
- Agent tool and execution controls · Agent detection, response and recovery · Agent security testing and validation
- Deployment surface
- SaaS agents · Custom/cloud agents · Endpoint/coding agents
- Integration point
- Platform connector
- Availability
- Confirm availability
Documented scope; confirm release status, supported integrations and entitlement for the selected offering.
03 / THE CONTROL PATH
Where it sits. How it works.
Atlas architecture interpretation: selected Surface connectors collect agent/configuration context, posture analysis identifies a supported exposure, and a separately authorized owner or control performs remediation.
- 01
Supported environment connectors
Connect the selected agent platforms and record readable configuration, identity and activity scope.
- 02
Agent inventory and posture
Inspect ownership, permissions and relationships behind a specific finding.
- 03
Reviewed remediation or runtime control
Verify the available action path and separately selected Enforce/Protect deployment when required.
Named SaaS and cloud environments
The pages identify Copilot Studio, ChatGPT Enterprise, Salesforce Agentforce, Azure AI Foundry, AWS Bedrock, and Google Vertex AI among the covered environments. Confirm connector scope and required privileges per platform.
Endpoint and custom-agent coverage
Zenity also describes local coding assistants, MCPs, and homegrown frameworks. Establish the collection and enforcement component for each surface; a SaaS connector and an endpoint deployment are different integration tasks.
Separate posture remediation from action prevention
AISPM documents policy evaluation and automated playbooks, while the broader platform assigns action-time decisions to Enforce. Ask for the precise mechanism, approval workflow, and rollback behavior of the selected remediation.
04 / WHAT EARNS ITS PLACE
Strengths, with the evidence attached.
Atlas analysis of documented capabilities; availability and behavior depend on the edition and environment.
Control areas are distinguished
The platform presents Surface, Enforce and Protect separately, enabling a more honest capability map.
Posture and observability have dedicated scope
The product pages describe agent context and configuration assessment that can be checked against known assets.
Inspect the documented capability notes
Evidence reviewed: Platform module descriptions and separate AI Observability and AISPM pages.
Inventory with ownership and connections
AI Observability describes agent configuration, permission, tool-access, and ownership context. Inventory spans SaaS, homegrown, and endpoint environments, with counts and trends for agents, threads, and MCPs.
Execution and data-access visibility
The observability page describes monitoring execution paths, tool invocations, memory updates, RAG queries, file access, and attachments. Data Lens highlights sensitive or overshared material based on observed exposure and use.
Configuration posture and playbooks
AISPM evaluates permissions, integrations, and configuration before deployment and as agents change. The page describes playbooks that can alert a team or apply a fix when new violations appear.
Exposure and runtime layers
The platform distinguishes exposure validation in Surface from action-time policy in Enforce and runtime detection in Protect. Those functions share context but require their own demonstrated integration and entitlement.
05 / KNOW THE COMPROMISES
The tradeoffs to take seriously.
Architectural implications and procurement questions from Atlas; these are not independently measured product failures.
Discovery is connector-dependent
Define the actual connected platforms and accessible data rather than assuming universal visibility from the platform name.
Additional coverage boundaries
- A discovered agent and a discovered tool connection do not prove every execution step is observable or preventable. Use a coverage matrix that distinguishes inventory, monitoring, remediation, and blocking.
- Configuration-derived risk may differ from effective permissions at runtime. Compare reported access to a real allow or deny result under the relevant delegated identity.
- Confirm connector permissions, scan freshness, custom-framework instrumentation, playbook safety, module packaging, and the handling of sensitive prompts, attachments, and retrieved data.
06 / FROM PURCHASE TO PRACTICE
Make it work in your environment.
A proposed implementation sequence. Confirm supported versions, permissions, and rollback with the current documentation.
- 01
Create a connector coverage matrix
Map actual agent builders and environments to supported connections, required permissions and refresh expectations.
- 02
Reconcile known agent ownership
Validate a sample of agents, identities and resources against platform-native records.
- 03
Close one exposure end to end
Assign an owner, approve permission reduction and verify effective access after the change.
07 / THE OPERATING COMMITMENT
The bill is only part of the cost.
Commercial shape
Quote the selected Surface connectors and posture scope explicitly. Price Enforce, Protect and any runtime integration separately and include connector administration and remediation work.
- Connected environments and covered agent scope
- Selected posture, observability and control modules
- Connector permissions, remediation and operational integration
08 / COMPARE THE JOB
Put these beside it.
Alternatives worth examining for the same problem. Check each offering’s scope before treating it as a substitute.
Noma AI Agent Security Platform
Compare discovery and AI-SPM coverage on the same environments, keeping runtime enforcement modules separately scoped.
Lasso SecurityLasso AI Security Posture Management
Compare agent graphing and developer remediation for the same supported application and permission sample.
09 / MAKE THE DECISION OBSERVABLE
Earn confidence through a real evaluation.
Suggested by Atlas. Agree acceptance criteria, use synthetic fixtures in an authorized environment, and capture both useful work and failed controls.
- TEST 01
Reconcile a known agent roster
Acceptance evidence: Retain discovered, missing and incorrectly associated records.
- TEST 02
Validate one excessive-permission finding
Acceptance evidence: Show the underlying platform permission and meaningful resource access.
- TEST 03
Remediate and rescan the test exposure
Acceptance evidence: Capture owner approval, effective target access and finding resolution.
START WITH ONE EXERCISE
A Copilot Studio test agent has access to an unrelated synthetic document library.
Evidence to look for: Compare the discovered agent and exposure finding with the source configuration and owner record.
Questions to bring to the vendor
- Which agents are discovered by each connector?
- How are effective permissions calculated?
- Which findings have an integrated enforcement path?
10 / SHOW THE WORK
Evidence, and what is still unknown.
What this research cannot establish
- Atlas has not independently measured discovery completeness or posture accuracy.
- Confirm exact connector scope, refresh behavior and integrated remediation permissions.
Dossier research reviewed October 18, 2026. Earlier profile references are retained for traceability; products and commercial terms change.
- Zenity: Surface, Enforce, and Protect ↗Product information
- Zenity: AI Observability data and environment scope ↗Product information
- Zenity: AISPM configuration and remediation ↗Product information
KEEP BUILDING YOUR JUDGMENT