Skip to main content

Introduction

CrawlChat is a platform where you can connect documentation from different sources and get an AI agent trained on that documentation. You can embed it on your website, connect it as a Discord or Slack bot where people can get answers to their queries, or use it as an MCP server so your community can integrate your documentation into AI apps such as Cursor, Claude Code, and more.

Process​

The CrawlChat workflow consists of four major steps at a high level.

  1. Add your documentation
  2. Customize the agent
  3. Connect the channels
  4. Analyze

Let's briefly walk through those four steps.

1. Add your documentation​

The first step is to add your existing documentation to the CrawlChat agent you create after signing up. CrawlChat covers the major source types for building the knowledge base that will answer queries later. Here are a few:

  • Scraping websites
  • Upload files
  • GitHub Issues
  • Notion pages

2. Customize the agent​

Once you have added your documentation, customize the agent so that it works as expected. You can configure options such as:

  • System prompt
  • Branding
  • AI models
  • Sample questions
  • Minimum scores

3. Connect the channels​

You can connect to different channels that cover the requirements of most tech companies. A few of them are:

  • Embed to web
  • Discord bot
  • Slack app
  • MCP server

4. Analyze​

The best part about CrawlChat is the analysis it provides from the queries it answers and the types of questions people ask across channels. More about it here.