How It Works
When a user asks a question, Crow searches your knowledge base to find relevant information. This process is called Retrieval-Augmented Generation (RAG):- User asks a question — “How do I reset my password?”
- Crow searches — Finds relevant content from your uploaded documents, website, or Notion
- Agent responds — Uses the retrieved information to provide an accurate answer
Knowledge Sources
Crow supports three types of knowledge sources:Files
Upload documents like PDFs, Word docs, spreadsheets, and more.
Website
Automatically crawl and index your website content.
Notion
Connect your Notion workspace for dynamic content access.
Choosing the Right Source
| Source | Best For | Updates |
|---|---|---|
| Files | PDFs, manuals, static documentation | Re-upload when content changes |
| Website | Public-facing docs, help centers, blog posts | Re-crawl to update |
| Notion | Internal docs, wikis, frequently updated content | Always current (live connection) |
Best Practices
Keep Content Focused
Upload content that’s relevant to user questions. Support docs, FAQs, and product guides work great. Avoid uploading unrelated internal documents.Update Regularly
When your product changes, update your knowledge base. Outdated information leads to incorrect answers.Test After Adding Content
After adding new content, test your agent in the Sandbox to ensure it can find and use the information correctly.Combine Sources
You can use multiple sources together. For example:- Files for product manuals and guides
- Website for your public documentation
- Notion for internal policies and procedures