What is a Document Search node?

A Document Search node lets your AI agent answer questions using the documents you connect to the node.

You can use files such as PDF, DOCX, TXT, and Markdown, and the AI searches their content to find relevant information before generating an answer.

Document Search is useful when you want answers grounded in your own source material instead of unrelated general knowledge.

When to use it

Use Document Search when your information is:

• In documents or text files
• Unstructured rather than organized into spreadsheet rows and columns
• Spread across policies, FAQs, guides, manuals, procedures, or other written material
• Meant to answer questions such as:
  • “What does this document say about X?”
  • “What is our policy on Y?”
  • “What are the instructions for this process?”

How it works

1. Upload your documents.
2. Select up to 10 files in the node.
3. The user asks a question.
4. Document Search finds the most relevant information in those files.
5. The AI responds using the retrieved document content.

The connected documents act as the source of truth for the node.

If the information is not available in those documents, the AI should not invent an answer from unrelated general knowledge.

How Document Search finds information

Document Search uses hybrid search, combining exact text matching with meaning-based retrieval.

This helps it find relevant information even when the user does not use the exact wording found in the document.

Exact wording

Document Search can match specific:

• Words
• Phrases
• Names
• Numbers
• Dates
• Product codes
• Technical terms

For example, if a policy says:

Late checkout is available until 1:00 PM for an additional fee.

A user can ask:

What time is late checkout?

The node can find the relevant passage based on the matching terms.

Meaning-based search

Document Search can also find information based on meaning.

For example, a document may say:

Guests may bring one domestic animal weighing no more than 25 pounds.

A user could ask:

Are small pets allowed?

Even though the wording is different, the node can still retrieve the relevant information.

This combination helps users ask questions naturally while keeping answers grounded in the documents you selected.

What you configure

Inside the node:

• Documents to Search
  Select the files the AI should use.

• AI Response Style (prompt)
  Control how the AI responds, including tone, format, and instructions.

Example:

“Answer clearly and concisely using only the provided documents.”

Document Search vs. Data Search

Use Document Search for unstructured written knowledge such as:

• Policies
• FAQs
• Manuals
• Guides
• Procedures
• Product documentation
• Service information
• Training material

Use Data Search when the information is structured in CSV or Excel files and you need row lookups, filtering, calculations, or analysis across structured data.

Important to know

• Document Search works best for text-based, unstructured knowledge.
• You can select up to 10 documents in one Document Search node.
• Answers stay grounded in the documents connected to the node.
• Hybrid search helps match both exact wording and similar meaning.
• It is not intended for spreadsheet calculations or structured-data analysis.
• For structured CSV or Excel data, use Data Search instead.

Simple rule

Have documents with written information? → Use Document Search

Have structured data in rows and columns? → Use Data Search