Knowledge Base
The Knowledge Base stores information that your bot uses to answer questions. Using RAG (Retrieval-Augmented Generation), your bot can provide accurate, contextual responses based on your content.
How It Works
- You add content — Files, websites, or data sources
- We process it — Content is chunked and embedded
- Bot retrieves context — Relevant chunks are found for each query
- AI generates response — The model uses retrieved context to answer
Key Benefits
- Accurate answers: Responses grounded in your actual content
- Reduced hallucination: AI stays within known information
- Easy updates: Add or remove content anytime
- Source attribution: Know where answers come from
Quick Start
Adding Your First Source
- Go to your bot’s Knowledge tab
- Choose a source type:
- Files: Upload PDF, Word, Excel, or text files
- Websites: Enter a website URL to crawl
- Data Sources: Connect external data sources
- Add content and wait for processing
Testing
After processing completes:
- Go to Bot Studio to test your bot
- Ask a question about your content
- Verify the bot uses knowledge base information
Content Guidelines
What to Add
- FAQ pages
- Product documentation
- Help articles
- Policy documents
- Company information
What to Avoid
- Duplicate content (confuses retrieval)
- Outdated information (leads to wrong answers)
- Irrelevant content (dilutes search results)
- Sensitive data (security risk)
Next Steps
- Supported Sources — Detailed source type information
- Training Your Bot — Optimize knowledge retrieval
- RAG & Semantic Search — Technical deep dive
Last updated on