Skip to main content
The Search Knowledge node runs a semantic search over a Knowledge Base and returns the most relevant passages. Use it to pull facts, policies, or product docs into a workflow before the next step decides what to do.
Search Knowledge node configuration form

Search Knowledge configuration — pick the knowledge base, write the query, optionally cap results and toggle reranking.

​
When to use

  • Grounded answers — Pull relevant policy or product passages before Ask AI drafts a reply.
  • Lookups — Find the right SKU, FAQ entry, or playbook step based on a free-form query.
  • Citations — Surface source passages so an actor or a downstream node can quote them with context.

​
How it works

  1. Pick which knowledge base to search.
  2. Write the query — usually a template that includes the user’s message or a derived variable.
  3. The node returns ranked matches (text + source metadata) into the output variable.
  4. Downstream nodes read {{<output_variable>.results}} and pass it to Ask AI, App Action, or wherever it’s needed.
For multi-agent conversations, you usually want to attach the knowledge base as a Capability on the actor instead — the LLM decides when to search and which query to use.

​
Next steps

Knowledge Base

Index documents that workflows and actors can search.

Ask AI

Pair Search Knowledge with Ask AI for grounded responses.