Model Discovery

What is Model Discovery?

Model Discovery scans a connected cloud platform — Azure, AWS, or GCP — and surfaces the AI models currently deployed in that environment. Discovered models can then be imported directly into Asenion as AI system records, eliminating the need for manual entry.

This is the primary mechanism for organizations that deploy models via cloud ML platforms to build their AI system inventory without starting from scratch.

How it works

  1. An Org Admin sets up a cloud platform connection in Settings → Connections (Azure, AWS, or GCP). See Integrations.
  2. From Settings → Model Discovery, select the active connection and trigger a sync.
  3. Asenion queries the cloud platform and returns a list of discovered models with their endpoint, region, framework, version, and deployment status.
  4. Review the list and click Import on any model to create it as an AI system record in Asenion.

Discovered model statuses

Status Meaning
Deployed Model is currently serving predictions
Stopped Model endpoint exists but is not running
Pending Model is being deployed or stopped
Unknown Status could not be determined from the cloud API

What gets imported

When a model is imported, Asenion creates a new AI system record pre-populated with:

  • Model name (from the cloud platform)
  • Cloud provider and model ID
  • Endpoint URL and region
  • Framework and version (where available)

The record is created as a draft — the owner then completes the intake, assigns policies, and runs assessments.

Sync frequency

Syncs are triggered manually. Each sync fetches the current state of all models visible to the connected service account. The Last Synced timestamp is shown per connection.

Who can use Model Discovery?

Org Admins with org:manage_settings access can configure connections and trigger syncs. Any Org Admin can initiate an import from the discovered model list.