Sarghaby Araghatta
Request a free assessment →
Home / Coverage

Know exactly where Sargha looks, and what it can find.

A sensor on machines that can run one, read-only connectors for everything else, and one inventory of 18 groups and 101 types of AI asset.

The Sargha coverage catalog: counts of items found, types seen, gaps and items from connectors, then each group of AI asset types with how many were seen and by which route, with gaps marked
Every group and type of AI asset Sargha looks for, whether it was seen, and by which route. Gaps are named, never hidden. Sample data from the Sargha sandbox.

Endpoints and servers

Windows, Windows Server, macOS and Linux, from one self-contained program per system. The Windows client and server sensors are generally available. Roadmap Wider validated Windows editions and server roles are planned.

Cloud accounts

AWS, Google Cloud and Azure through read-only connectors that add identity, access and storage posture. Oracle Cloud is read for AI services only, in beta. Roadmap

Phones and tablets

AI apps read from your device-management inventory (Intune, Jamf, Workspace ONE). It is an inventory import, not an on-device agent.

Containers and Kubernetes

A container image and a Kubernetes node agent that observes only, plus an agentless cluster read through a read-only account. Virtual desktops (Azure Virtual Desktop, Citrix, Horizon) are not covered today.

The AI runtime stack

Retrieval pipelines, vector stores, orchestrators, tool and memory layers and model files, including in-house components.

Always current

Every asset is checked, kept current and compared with the last look, so you see what changed and what disappeared.

Signal classes

Eleven kinds of signal, where shadow AI actually hides.

Normalized into one inventory.

Network traffic to AI servicesTool servers and what they can callAPI keys and secretsAI command-line tools, apps and SDKsIDE AI assistantsAgent skillsAI-written repositoriesAutonomous agentsBrowser extensionsAI apps on phonesCloud AI services
How it works

Step by step.

  1. Choose how to look

    Put the sensor where a machine can run one, and switch on connectors for everything else.

  2. Merge into one inventory

    Sensor and connector findings are matched and counted once, with gaps named.

  3. Check what changed

    Every asset is checked, kept current and compared with the last look.

Questions

Common questions.

Does the sensor read my files?
It reads configuration and package evidence on the host to find AI tools, masks secrets on the device, and never sends file contents, prompts or keystrokes. The optional browser extension checks submitted text in memory and does not store it.
Can I see AI on personal phones?
Only managed phones that appear in your device-management inventory. Personal devices you do not manage stay outside unless they show up in a log.
What about virtual desktops?
Azure Virtual Desktop, Citrix and Horizon are not covered today. Containers and Kubernetes nodes are, through a container image and a node agent that observes only.
What you need
  • Your device-management tool for the sensor.
  • Read-only accounts for each cloud or service you want connected.
For your technical team

The platform matrix, collection models, the full 101-type taxonomy and the signal classes are in the technical pack, shared with your team under NDA.

Request the demo and technical pack
Get started

See the AI on your fleet in about a week.

Run the sensor across your estate and receive an AI risk report and an AI Bill of Materials. The assessment is free and carries no commitment.

Request a free assessment → Book a 30-minute call