Chatbot - Building Your Knowledge Base
Overview
Uploading documents to your Chatbot project is the foundation of RAG (Retrieval-Augmented Generation). Your chatbot uses these documents to provide accurate, context-aware responses based on your specific knowledge base.
Prerequisites
Before uploading documents:
1. Design Your Bot First: Ensure you have clicked "Design Your Chatbot" in Step 1.
2. Check Your Limits: Verify your remaining item slots (shown as "Items Left: X / Y") on the project page.
Upload Methods
1. Document Upload (All Tiers)
Supported File Types: PDF, DOCX, TXT, CSV.
- Student/Personal: Upload 1 file at a time.
- Smart/Professional: Upload up to 10 files simultaneously.
Step-by-Step:
1. Select "Document Upload" from the dropdown.
2. Drag and drop your files or click browse.
3. Once the upload finishes, the system will process the text. This usually takes 1-3 minutes.
2. URL Scraping (All Tiers)
Use this to index a single webpage (like an article or a pricing page).
1. Select "URL Scraping" from the dropdown.
2. Enter the full URL (including https://).
3. The system will visit the page and extract the text content.
3. Website Crawling (Smart & Professional)
Use this to index an entire website or a large documentation section.
1. Select "Website Crawling".
2. Enter the base domain (e.g.,
https://example.com).
3. Set the
Crawl Depth:
- Depth 1: Just the homepage and direct links.
- Depth 2+: Deeply linked pages.
4. Set
Max URLs: Limit the number of pages to avoid exceeding your quota.
4. Database Connection (Smart & Professional)
Connect your chatbot directly to structured data.
- Requirements: Database host, name, user, and password.
- Note: Ensure your database is accessible to our servers. We recommend using a read-only user for security.
Managing Your Knowledge Base
Monitoring Status
Your uploaded items will appear in the table below the upload form:
- ⏳ Processing: We are currently extracting and indexing the text.
- ✅ Completed: The data is ready to be added to the AI's memory.
- ❌ Failed: The file might be corrupted or the URL inaccessible.
Deleting & Updating
- To remove an item, click the trash icon. This will free up a slot in your quota.
- If you update a document, delete the old version and upload the new one.
- Important: Whenever you add or delete items, you must Rebuild the Vector Store in Step 2 for the changes to take effect in the chatbot's memory.
Best Practices for High Accuracy
✅ Clean Your Data: Remove headers, footers, or repetitive navigation text from documents when possible.
✅ Small, Focused Files: Rather than one giant 500-page PDF, try uploading multiple smaller PDFs organized by topic.
✅ Use "Only Include" Filters: When crawling, use the include filter (e.g., /docs/) to avoid indexing irrelevant pages like "Terms of Service" or "Login".
Next Steps
Once your items show as Completed:
