ChatMaxima Docs

Knowledge Source

Train your AI with your business knowledge

Knowledge Source Overview

Knowledge Source allows you to upload and manage content that powers your AI chatbots, enabling them to answer questions accurately using your business information.

Content Types

Documents

Upload various document formats:

  • PDF - Product manuals, guides, policies
  • Word (.docx) - Documentation, SOPs
  • Text (.txt) - Plain text content
  • Markdown (.md) - Formatted content

Web Pages

Import content from websites:

  • Single page import
  • Sitemap crawling
  • Automatic updates
  • Content extraction

FAQs

Create structured Q&A pairs:

  • Question and answer format
  • Category organization
  • Priority ranking
  • Quick retrieval

Custom Data

Add structured information:

  • Product catalogs
  • Pricing tables
  • Service details
  • Company information

Managing Knowledge

Upload Content

  1. Go to Knowledge Source
  2. Click Add Content
  3. Select content type
  4. Upload or enter content
  5. Process and index

Organization

Collections

Group related content:

  • Product Knowledge
  • Support FAQs
  • Company Policies
  • Training Materials

Tags

Label content for filtering:

  • By topic
  • By product
  • By department
  • By language

Content Status

StatusDescription
ProcessingBeing indexed
ActiveAvailable for AI
InactiveDisabled but retained
ErrorProcessing failed

AI Training

How It Works

  1. Content is uploaded and processed
  2. Text is extracted and chunked
  3. Embeddings are generated
  4. Knowledge is indexed for retrieval
  5. AI uses relevant chunks to answer

Retrieval Settings

Configure how AI finds information:

  • Similarity Threshold - Match accuracy
  • Max Chunks - Content pieces per query
  • Source Priority - Preferred sources

Using in Chatbots

Connect to Bot

  1. Open bot in ChatMaxima Studio
  2. Add AI Response node
  3. Select Knowledge Source
  4. Choose collections to use
  5. Configure response settings

Response Configuration

  • Cite Sources - Show where info came from
  • Fallback Message - When no match found
  • Confidence Threshold - Minimum certainty
  • Response Style - Tone and format

Best Practices

Content Quality

  • Keep information up-to-date
  • Use clear, concise language
  • Structure content logically
  • Remove outdated information

Organization

  • Create logical collections
  • Use consistent tagging
  • Regular content audits
  • Version important documents

Performance

  • Monitor AI accuracy
  • Review unanswered questions
  • Add missing information
  • Refine based on feedback