Platform AI configuration
Platform Administration › AI Configuration shows which AI service the installation uses for Nora, workplace analyses, document generation and the company search when signing up. Only platform administrators see it. The configuration comes from the server or from this page.
1. Configured by the server
If the server defines the AI (endpoint, key and deployment in its configuration) and no endpoint or key has been saved on this page, the page shows the notice «Configured by the server» and the configuration in use, with no fields to edit and no save button: provider, endpoint, API key, main model, small model with its endpoint and key, models for Nora's tool rounds and answers, embeddings deployment, and Speech region and key. Keys show only «Configured» or «Not configured», never the value. To change anything, change the server configuration and restart the service.
2. AI features on
With a complete server configuration, AI features are on without doing anything else: the company search when signing up appears and Nora answers. If the server is missing the endpoint, the key or the deployment, or turns the features off, the page says «AI features off».
3. Configuring the AI from this page
If the server does not define the AI, or this page already has a saved endpoint or key, the page is a form: endpoint, API key, deployment name, small model, models for Nora's rounds, Nora's voice (Speech region and key), model settings, caching, retries and request limits. «Enable AI Features» turns them on or off for the whole platform. What is saved here takes priority over the server configuration. The key is never shown again: leave the field blank to keep it.
4. Testing the connection
In the form, «Test Connection» sends a short request to the deployment with the endpoint and key entered (or the saved key if the field is blank) and shows the result and the response time. If it fails, the message says whether the problem is the key, the deployment or the endpoint.