Knowledge Base

Train your avatar with custom knowledge to answer questions about your specific business, product, or domain.

What is Knowledge Base?

The Knowledge Base is a RAG (Retrieval-Augmented Generation) system that lets you provide custom information to your avatar. When users ask questions, the avatar searches your knowledge base and uses relevant information to answer accurately.

Adding Knowledge

Option 1: File Upload

Upload documents directly:

1. Go to Dashboard > Avatars > [Your Avatar] > Knowledge Base
2. Click Upload Files
3. Select your documents (PDF, TXT, MD, DOCX)
4. Wait for processing (usually under 30 seconds)

Supported Formats:

• PDF (up to 50 pages)
• Text files (.txt)
• Markdown (.md)
• Word documents (.docx)
• Max file size: 10MB per file

Option 2: Website Crawling

Import content from your website:

1. Go to Knowledge Base > Add Source > Website
2. Enter your website URL
3. Configure crawl depth (how many links to follow)
4. Click Start Crawl

Settings:

Option 3: Custom Q&A

Add specific question-answer pairs:

1. Go to Knowledge Base > Add Source > Q&A
2. Enter the question (how users might ask it)
3. Enter the ideal answer
4. Click Save

Example:

• Question: "What are your business hours?"
• Answer: "We're open Monday-Friday, 9am-6pm EST. We're closed on weekends and major holidays."

API Integration

Add knowledge programmatically:

Upload a Document

POST /avatars/{avatarId}/knowledge
Content-Type: multipart/form-data

{
  "file": <file>,
  "type": "document"
}

Add Text Content

POST /avatars/{avatarId}/knowledge
Content-Type: application/json
 
{
  "type": "text",
  "content": "Your product FAQ here...",
  "metadata": {
    "source": "FAQ Page",
    "category": "product"
  }
}

Add Q&A

POST /avatars/{avatarId}/knowledge
Content-Type: application/json
 
{
  "type": "qa",
  "question": "What are your business hours?",
  "answer": "Monday-Friday, 9am-6pm EST."
}

How It Works

When a user asks a question:

1. Query Analysis – We analyze the user's question
2. Semantic Search – Find relevant chunks from your knowledge base
3. Context Injection – Add relevant info to the AI prompt
4. Response Generation – AI answers using your knowledge

This ensures accurate, on-topic responses based on your actual information.

Best Practices

Keep Content Focused

• Remove marketing fluff – just the facts
• Break large documents into logical sections
• Use clear headings and structure

Update Regularly

• Remove outdated information
• Add new products/features as they launch
• Review and refine Q&A based on real user questions

Test Thoroughly

• Ask common questions and verify answers
• Test edge cases and ambiguous queries
• Check that the avatar admits when it doesn't know

Limits by Plan

Manage Knowledge Base → (opens in a new tab)