Pre-built Agents
RAG Powered Agents
Chat Agent
Data Sources
Overview

Chat Agent Data Sources

AI-Horizon's Chat Agent features a Retrieval-Augmented Generation (RAG) pipeline that can load, index, and query various format data sources dynamically. By providing dedicated loaders, the Chat Agent converts raw unstructured data into semantic chunks stored in your chosen vector database.

Supported Data Formats

You can initialize and feed your Chat Agent using several data types:

  • PDF Documents: Index and search text, financial reports, or user manuals.
  • Word DOCX Files: Scan corporate spec files or documentation folders.
  • Flat Text Files: Index simple text dumps or product descriptions.
  • Single Webpages: Scrape and digest a specific web page URL.
  • Recursive Websites: Scrape entire domains up to a specified depth for corporate knowledge bases.
  • YouTube Videos: Extract audio tracks and transcribe streams into interactive text models.

Unified Execution Interface

Each data type corresponds to a core initialization technique in the ChatBot class:

  • ChatBot.pdf_chat()
  • ChatBot.docx_chat()
  • ChatBot.txt_chat()
  • ChatBot.webpage_chat()
  • ChatBot.website_chat()
  • ChatBot.youtube_chat()
UsagePDF