› Workflow › Business processes › Design a basic process › Add an activity › Activity types › AI Knowledge Base activities › Knowledge discovery agent activity › Configure Knowledge discovery agent activity › Configure the Knowledge discovery agent activity using the Dynamic configuration

Configure the Knowledge discovery agent activity using the Dynamic configuration

You can dynamically configure the Knowledge discovery agent activity so you can use a single process with different configurations.
Procedure
  1. Add an activity to the map. See Add an activity.

    By default, the Configuration tab is open in the properties panel of the activity.

  2. Enter a Name for the activity.
  3. On the Type list, select AI Knowledge Base > Knowledge discovery agent.
  4. Select the Configuration mode as Dynamic.
  5. When you perform a search, the system by default returns the entire search results (not just chunks), including response text. To limit execution to retrieving relevant indexed documents and references without calling a Large Language Model (LLM) to generate a response, select Search only. (Default: Clear).

    This option helps search and retrieve documents efficiently while minimizing token usage and the cost associated with LLMs.

  6. On the Dynamic configuration (JSON) list, select a process or server variable (String or Data object).

    • The String variable must have a valid JSON configuration. If the JSON is not valid, an error appears at runtime.

    • For a data object variable, you can use the "KBDynamicSearch" data model available in the System category. This data object holds the dynamic configuration JSON for the Knowledge discovery agent. The Index, Filters, Query type, Content Fields, Vector fields, and other configuration fields are available as the fields in the "KBDynamicSearch" data model.

    At runtime, you can enter the values in these fields.

  7. On the Response list, select a Data object variable that holds the output. You can create a new data object variable or map the "KBSearchResult" variable available under System category.

    If "Search only" is enabled, you can map the out-of-the-box Data object variable, "KBSearchOnly" available under System category. KBSearchOnly Datamodel contains the DocumentChunks list field with score, rerankerScore, DocumentId, DocumentName, ReferenceId, ChunkText, Coordinates of type Object. If you create your own data model instead of using KBSearchOnly, the system will only populate fields that match names/types from the standard model.

  8. Save the process.

    When the activity is executed:

    • If "Search only" is not enabled, the query is processed, and the search results, along with the generated response, are populated accordingly.

      • The system logs the number of tokens used during the search process.

      • If a new data model is mapped, chunk text and scores are set to default values (0 or null), and only matching fields are displayed.

    • If "Search only" is enabled, and the activity runs using the variable mapped to KBSearchOnly, the search is processed, and relevant document chunks (specific sections of the document) based on the query are retrieved.

      • The re-ranker score is between 0-4, and the score is unbounded.

      • No LLM call is made, and the response text is not generated. The token usage is displayed as 0.

More info
.