Activate the Platform Multimodal Service plugin
Enable automatic generation of searchable descriptive captions for images, tables, charts, and other visual elements found in indexed attachments.
Before you begin
Role required: admin
About this task
AI Search offers optional automatic multimodal generation of searchable descriptive captions for images, tables, charts, and other visual elements found in attachments that you index for search. These generated captions improve search recall, allowing you to find attachments using search keywords that match generated terms.
As an example, an attachment might include an image that yields the generated caption man on bicycle. Searches for man or bicycle can match the caption terms. Such searches return the attachment as a search result even if its text doesn't otherwise contain either of those terms.
The multimodal captioning feature has been validated and tested for English-language content. For content in other languages, it may function but has not been evaluated for caption quality or accuracy. Multimodal captioning for non-English content is not currently supported.
Procedure
What to do next
- AI Search administrators can activate multimodal captioning for individual AI Search indexed sources. For details on this procedure, see Activate multimodal captioning for attachments from an indexed source.
- Connector administrators can activate multimodal captioning in the crawl settings for an external content connector. To learn about configuring crawl settings for external content connectors, see Configuring crawl settings for external content connectors.
- AI Search administrators can select the VLM (visual learning model) provider and model used by the Platform Multimodal Service. For details on this process, see Configure multimodal captioning for AI Search.