Sarghaby Araghatta
Request a free assessment →
Home / Non-sensor devices

See the AI that no sensor can reach.

72 read-only connectors in 16 families, plus 7 collectors for mobile devices, cover cloud, SaaS, identity, phones, gateways, network, storage and code.

Your cloud and SaaS
Cloud accountsAWS, Azure, Google Cloud, Oracle
SaaS and codeAI apps, repositories, pipelines
Identity and devicesDirectories, device management
Free add-onsChat, code, directories
Sargha cloud
Read-only connectorsReads metadata only.
One inventoryWith your sensor findings.
Your network and data
AI gatewaysProxies and DNS logs
Network, IoT and dataSwitches, buckets, vector databases
Logs you collectSent to Sargha
  • Metadata
  • Findings and alerts
  • Identity and access
  • Evidence and reports
  • Dashed: optional
  • Thicker: more data
  • Encrypted in transit
  • Read-only
  1. 1
    Cloud accountsAI services, roles and storage.Never sent Prompts, responses, files.
  2. 2
    SaaS and codeApps and pipelines in use.
  3. 3
    Identity and devicesOwners from your directory.
  4. 4
    AI gatewaysModels and routes in use.
  5. 5
    Network, IoT and dataWhere AI-capable gear sits.
  6. 6
    Free add-onsOff until you opt in.
  7. 7
    Pushed logsLogs you already collect.
  8. 8
    One inventoryA failed run deletes nothing.
Per tenant: your plan and your opt-in decide what is on.

Read-only by design

Each connector reads a system's own records with an account you create and can revoke at any time. It reads metadata, never prompts or file contents.

You choose what is on

A connector is on only when your plan includes it and you switch it on. Free add-ons stay off until you opt in.

A bad run deletes nothing

If a run fails, times out or stops early, nothing is marked removed. Demo data is refused for real tenants and never counted in coverage.

Pull or push

Pull from a system's API where one exists. Push logs you already collect where it does not. Four steps take you from choice to coverage.

Honest about the gaps

The catalog marks known gaps and the coverage score never reaches 100%. Personal devices, home networks and what a device does internally stay outside.

16 families

Every place a sensor cannot go, and the connector that covers it.

Each family reads one kind of system. Field-level detail is in the technical pack, shared with your team under NDA.

Cloud accounts9Finds the AI services, roles and findings inside your AWS, Azure, Google Cloud and Oracle accounts.
SaaS-embedded AI5Finds AI apps and AI note-takers that users authorized inside Microsoft 365, Google Workspace, Slack, Zoom and GitHub.
Identity providers3Lists the service principals, API tokens and AI sign-ins that agents use as their identity.
Mobile devices7Reads installed AI apps on managed phones and tablets from your MDM, plus sign-in, cloud-app and DNS evidence.
LLM gateways and AI proxies6Reads gateway routes, keys, budgets and guardrails, and accepts OpenTelemetry GenAI traces.
Web proxy, SWG and DNS logs6Turns proxy, secure web gateway and DNS logs into per-user AI usage, including for machines with no sensor.
Firewalls3Reads firewall logs and device lists so hosts behind the firewall are counted and their AI traffic is seen.
Network devices and NAC8Builds the device census from controllers, NAC, SNMP, DHCP leases and NetFlow, and flags AI-capable devices.
IoT and OT platforms4Reads the device inventory your IoT and OT platform already keeps and flags devices that could carry AI.
Cloud storage3Flags public or unprotected buckets that hold AI datasets, model weights or vector indexes.
NAS and storage appliances4Flags AI-named shares and volumes on NetApp, Pure Storage, Dell and Synology that are open or unencrypted.
Data platforms3Shows which models and AI endpoints your warehouses and lakehouses serve, and who may query them.
Vector databases8Lists vector stores and flags servers that accept unauthenticated requests or plain HTTP.
Source control and CI/CD5Finds AI libraries, provider credentials in pipelines and committed agent configs across GitHub, GitLab, Bitbucket, Azure DevOps and Jenkins.
Kubernetes (agentless)1Reads clusters with a read-only service account to find model servers, notebooks and GPU workloads.
EDR and XDR1Lists endpoints your EDR manages that have no sensor as known AI-blind, and turns AI-related Defender alerts into findings.
Generic import and webhooks3Imports CMDB or CSV asset lists and accepts webhooks, so any inventory can count toward coverage.

Maturity, stated plainly

  • BetaThe connectors were built against each vendor's documented interface and tested against recorded responses, not yet against a live customer tenant.
  • PreviewA few network and OT vendor connectors are in preview.
  • RoadmapEvent-driven collection Roadmap, response actions through connectors with dry-run, approval and rollback Roadmap, and vendor-certified EDR and ITSM integrations Roadmap are planned.
What you need
  • A read-only account in each system you want connected.
  • Your plan, which decides the default bundle. See Plans and connectors.
For your technical team

Every connector, field by field, with how data arrives and the read-only access each needs, is 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