Client model integration (feature 089)
Client model integration (feature 089)
Section titled “Client model integration (feature 089)”Summary: The Ajutant Models page shows your Azure AI Foundry deployments enriched with central registry metadata (where available).
What feature 089 does
Section titled “What feature 089 does”Feature 089 (client-model-integration) enriches the Client Models page with metadata from a central platform registry.
When you open the Models page, you will still see your deployed models and their deployment details (as before). In addition, each deployed model may show registry metadata matched by model name after you sync from Azure.
Category-based model selection (important)
Section titled “Category-based model selection (important)”If your assistant has a recommended platform category, that category is advisory, not a restriction.
- Clients can choose any model for any assistant.
- The resolver uses the client’s chosen deployment (existing behavior).
- If the chosen model does not match the assistant’s recommended category, or lacks tested capabilities, Ajutant shows a warning, but the choice is still allowed.
What metadata you may see
Section titled “What metadata you may see”For models that exist in the central platform registry, the Models page can show enriched metadata including:
- Capabilities
- Platform category (economy/value/general/performance)
- Token cost history
- Lifecycle status
- Recommended API version
If the registry has no data for a model, the model still appears with deployment details only.
Deprecation and missing registry data
Section titled “Deprecation and missing registry data”Model not in the platform registry
Section titled “Model not in the platform registry”- The model shows deployment details.
- A note appears: “Not in platform registry”.
- No error and no blocking.
Deprecated models
Section titled “Deprecated models”If the registry marks a model as deprecated and provides a successor:
- A warning badge shows: “Deprecated, successor: [successor model]”.
How to use it (client admin)
Section titled “How to use it (client admin)”- Open the Models page.
- Review your deployed models and their deployment details (endpoint, quota, active status).
- Click Sync from Azure to refresh the deployed model list.
- After sync, Ajutant re-matches each deployed model by model name and displays registry metadata where available.
User scenarios and acceptance expectations
Section titled “User scenarios and acceptance expectations”Scenario 1: viewing enriched deployed models
Section titled “Scenario 1: viewing enriched deployed models”Expected behavior
- All deployed models in your Azure AI Foundry are shown with deployment details.
- For models found in the registry, enriched metadata is displayed.
- For models not found in the registry, you see deployment details only plus the note “Not in platform registry”.
Scenario 2: managing Azure deployments from the Models page
Section titled “Scenario 2: managing Azure deployments from the Models page”Feature 089 preserves existing deployment management on the same page, including:
- Viewing deployments grouped by capability type (chat, embedding, image)
- Setting defaults
- Editing pricing overrides
- Toggling active/inactive
- Syncing from Azure
After Sync from Azure, registry metadata is re-matched for all models.
Notes for reviewers
Section titled “Notes for reviewers”- This feature is listed as status: partial in the feature catalogue entry.
- The spec overview provided includes the behaviors above, but does not include UI screenshots, exact field names, or the full set of deployment management controls beyond the items mentioned.
If you need additional UI-level details (exact labels, where badges appear, or the full set of controls), they must be supplied from the remaining documentation sources for feature 089.