Specify the settings for Tungsten AI provider

See Tungsten AI.

Procedure
  1. Navigate to System > System settings > System > Tungsten AI.

    The Tungsten AI dialog box is displayed.

  2. On the Region list, select the region to be used when using the Tungsten AI Provider, such as EU. (Default: US)

    The supported regions include APAC, EU, US, APAC NVT, EU NVT, and US NVT.

    The Region URL is automatically displayed based on the selected region and is read-only.

    For region-based URLs that need whitelisting, refer to the Tungsten TotalAgility Administrator's Guide.

  3. To specify a fallback AI provider after you have exhausted your token allocation for the Tungsten AI provider, on the Secondary AI provider list, select an AI provider such as OpenAI or AzureOpenAI.

    When you execute a prompt using the Tungsten AI provider and have no tokens remaining, the secondary provider is used for the specified region.

  4. Select Use async to execute prompts using asynchronous API calls while interacting with Generative AI or Knowledge Base. (Default: Clear)
    Asynchronous API calls enable you to continue the conversations without waiting for long responses.
  5. In Model Selection, select the model to be used for different AI features in TotalAgility when the Tungsten provider is used:
    • Copilot for development

      • Model for text

      • Model for vision

    • Copilot for extraction

      • Model for extraction

      • Model for classification

    • Knowledge Base search

      • Model

    If the selected model is not supported or if no model is specified, the default model is used.

  6. Select Save.