Kodey.ai Platform Knowledge Base Guide

Overview

The Kodey.ai platform provides a robust knowledge base functionality that allows agents to access and utilize various document types when responding to user queries. This guide outlines the supported file formats and best practices for leveraging the knowledge base effectively.

Supported File Formats

Kodey.ai's knowledge base currently supports the following file formats:

  • Text files (.txt)
  • Word documents (.doc, .docx)
  • PDF documents (.pdf)
  • CSV files (.csv)

Unsupported File Formats

The following file formats are not currently supported by the Kodey.ai knowledge base:

  • Video files (.mp4, .mov, etc.)
  • PowerPoint presentations (.ppt, .pptx)
  • Excel spreadsheets (.xls, .xlsx)
  • Image files (.jpg, .png, etc.)

Best Practices for Accessing Knowledge

Using the AgentKnowledgebase Tool

When building agents that need to access the knowledge base, always use the AgentKnowledgebase tool to:

  1. Retrieve relevant information from uploaded documents
  2. Match the speaking tone and style present in the documents

Example implementation:

RULE: You MUST call the AgentKnowledgeTool tool in order to answer. "Never output ' Final Answer:' until after a tool has been used

Benefits of Proper Knowledge Base Usage

  • Consistent Brand Voice: Ensure your agent maintains the appropriate tone and style across all communications
  • Accurate Information: Access up-to-date information from your knowledge documents
  • Improved User Experience: Provide more relevant and contextual responses

File Preparation Tips

For optimal knowledge extraction, consider these tips when preparing files for upload:

  1. Text Files: Use clear headings and structured formatting
  2. Word Documents: Utilize proper heading styles and document organization
  3. PDFs: Ensure text is selectable and not embedded in images
  4. CSVs: Include clear column headers and consistent data formatting

Future Enhancements

The Kodey.ai team is continuously working to expand supported file types and improve knowledge extraction capabilities. Stay tuned for future updates regarding additional file format support.