Configure multimodal captioning for AI Search

  • Release version: Australia
  • Updated May 26, 2026
  • 1 minute to read

    • Use the AI Search Admin console to select the visual language model (VLM) provider and model for multimodal captioning.

    Before you begin

    • The Platform Multimodal Service plugin (com.glide.platform_mm_service) must be installed on your instance. If the plugin is not installed, multimodal captioning options will not appear in the AI Search Admin console. For more information, see Activate the Platform Multimodal Service plugin.

    • Role required: ais_admin

    About this task

    When multimodal captioning is activated, attachments retrieved from supported sources are automatically analyzed and given descriptive captions, making visual content, such as images, discoverable through text-based search. You can specify the VLM provider and model to use when generating captions.

    Procedure

    1. Navigate to All > AI Search Admin > AI Search Admin Home.
    2. Select System Properties.
    3. In the Multimodal captioning section, select a Provider from the drop-down list.
      A provider is the VLM you want to use for captioning.
      Note:
      If your preferred provider is not listed, use the AI Control Tower (AICT) to configure approved third-party LLMs. For more information, see Configure third-party LLMs using AI Control Tower. NowLLM does not support this feature.
      The corresponding Model options appear.
    4. Select a Model from the drop-down list.
      A model is a specific version of AI offered by the provider.
    5. Select Save.

    Result

    Changes to the provider and model take effect immediately for multimodal captioning.

    What to do next

    AI Search administrators have additional configuration options available, as follows: