Add an Ensemble Retriever to a Prompt Template

Use ensemble retrievers in prompt templates to ground your LLM prompts with relevant knowledge. An ensemble retriever is a collection of individual retrievers. When you run an ensemble retriever, it executes the individual retrievers, combines their search results into a single list, reranks the list according to relevance to the search request, and returns just the most relevant information to the prompt or agent. Ensemble retrievers conduct searches from multiple sources in parallel.

Required Editions

Available in: Lightning Experience

Available in: Enterprise, Performance, and Unlimited Editions with the Einstein for Platform, or Einstein or Agentforce for Sales or Service add-on, or Agentforce Foundations

User Permissions Needed
To create and manage prompt templates in Prompt Builder:	Prompt Template Manager permission set Manage Prompt Templates Execute Prompt Templates OR Customize Application permission set
To create and manage retrievers:	Data Cloud permission set
To run prompts with Einstein Search:	Data Cloud User permission set

Note These instructions assume that the ensemble retriever has been created and configured in AI Models. To learn more, see Create an Ensemble Retriever.

1. In Prompt Builder, create or edit a prompt template.
2. In the Prompt Template Workspace, click where you want to insert the ensemble retriever.
3. Type @ or click Insert Resource, click Retrievers, and click Configure Retrievers +.
4. Select the ensemble retriever you want to add to the prompt template.
5. Optionally, change retriever settings in the Template Settings side panel.
   1. Descriptors: Change the Label and Description for this retriever in your prompt template.
   2. Search Text: Focus your search by specifying additional criteria. Select a resource or specify search text.
   3. Output Fields: If available for the selected retriever, select one or more returned fields of data to ground your prompt.
   4. Number of Results: If available for the selected retriever, specify the maximum number of retrieved results.
6. Click Apply and Insert.
7. Save your work.
8. Test your changes. Preview the prompt template and review the LLM-generated response.