The Knowledge Search by a few Categories action allows your AI assistant to perform focused semantic searches across multiple specified Knowledge Base categories. Unlike the standard Knowledge Search, this action retrieves a set number of documents per category, allowing for broader coverage across relevant topics.

​
🔍 What It Does

This action queries selected categories in your Knowledge Base and retrieves the top-matching documents from each, based on a similarity threshold. It’s useful when your assistant needs a balanced perspective across multiple content segments.
✨ Use this when you want to fetch the most relevant documents from several categories, rather than the overall best matches.

​
🖼️ Action Interface

Knowledge Search by Categories UI

​
⚙️ Configuration Options

Type: string The search term or question. Supports dynamic values using parameters (@yourParam).Example: @userIntent, return policy
Type: string Comma-separated list of Knowledge Base categories to search across.Example: Returns, Shipping, Billing
Type: number Sets how many documents to return per category. Set to 0 to include all relevant documents in each category.Example: 3
Type: float Sets the similarity filter (range: 0.0 to 1.0).
  • 0.0 = highly relevant only
  • 1.0 = return all documents
Example: 0.25
Type: boolean When enabled, combines the retrieved document content with any existing output before passing it to the next action.

​
🧵 Tips

  • Use when you want more structured knowledge retrieval across defined segments.
  • Great for assistants that need to consult multiple product/service areas.
  • Works well as a first step in a multi-step reasoning chain.
Let us know if you want to explore how to chain this with summarization or API calls!