Enabling adaptive RAG mode for Ask your repository search

You can configure Content Assistant to send both relevant document chunks and complete document text to the LLM for vector queries.

Before you begin

Make sure that you have one of the following:
  • For Content Platform Engine 5.7.0 deployments, you must have Content Platform Engine 5.7.0 IF5 or later.
  • For deployments earlier than Content Platform Engine 5.7.0, you must have the latest version of Content Assistant.

About this task

When a natural language query is submitted to IBM® Content Assistant, the system typically retrieves the most relevant document chunks from the vector database and sends them to the LLM. You can configure the system to send both the most relevant document chunks and the complete document text to the LLM, enabling Content Assistant to deliver comprehensive answers with document citations based on both relevant chunks and complete document text. You can adjust this configuration by editing the GenAI configuration object and specifying the option as an advanced configuration option.

Procedure

  1. In the Administration Console for Content Platform Engine (ACCE), access the Gen AI Configuration object for the object store where you want to enable complete document content for Content Assistant.
    Tip: Starting with Content Platform Engine 26.0, you can open the Gen AI Configuration object for the selected object store directly from Administrative > Content Assistant > Gen AI Configuration.
    1. Go to the object store where you want to enable complete document content for Content Assistant.
    2. In the left pane, click Search.
    3. Click New Object Store Search.
    4. In Simple View, specify the Class as Gen AI Configuration, and click Run.
      The search results display a single Gen AI Configuration object.
    5. From the search results, click the object ID to access the properties for the Gen AI Configuration object.
  2. In the Properties tab, enable complete document content in the Gen AI Advanced Options field.

    Specify the VectorQueryUseDocContent option as a JSON value. Set the value to true to enable complete document content, or false to use only document chunks.

    For example, to enable complete document content for vector queries, enter {"VectorQueryUseDocContent":true}.
    Note: API callers can override this default by specifying their own value in API calls.
  3. Save your changes.