Knowledge bases are the foundation of your KnowFlow AI assistants. They store and organize all the information that your AI will use to generate responses.

​
Understanding Knowledge Bases

A knowledge base in KnowFlow is a collection of information from various sources that your AI assistant will use to answer questions. By creating comprehensive knowledge bases, you ensure your AI provides accurate, contextual responses based on your organization’s specific content.

​
Key Concepts

Knowledge Base

A container for all related information on a particular topic or product

Knowledge Source

Individual sources of information (documents, websites, GitHub repos)

Chunking

Breaking down content into smaller, digestible pieces for better retrieval

Embedding

Converting text into numerical vectors for semantic search

​
Creating a Knowledge Base

1

Navigate to Knowledge Bases

From your organization dashboard, click on Knowledge Bases in the left sidebar.
2

Create New Knowledge Base

Click the Create Knowledge Base button in the upper right corner.
3

Configure Settings

Enter the following information:
  • Name: A descriptive name for your knowledge base
  • Description: (Optional) A brief description of its purpose
  • Access Level: Choose between private (organization only) or public (shareable)
4

Submit

Click Create to finish setting up your knowledge base.

​
Adding Knowledge Sources

KnowFlow supports multiple types of knowledge sources to build comprehensive AI assistants.

​
Document Upload

Upload files directly from your computer:
1

Select Knowledge Base

Open the knowledge base you want to add documents to
2

Add Documents

Click Add Sources → Upload Document
3

Select Files

Choose files from your computer to upload. Supported formats include:
  • PDF (.pdf)
  • Microsoft Word (.docx, .doc)
  • Text files (.txt)
  • Markdown (.md)
  • CSV (.csv)
  • Excel (.xlsx)
4

Configure Processing

Optionally adjust settings for:
  • Chunking method (by paragraph, page, or custom size)
  • Text extraction preferences
  • Metadata extraction
5

Upload

Click Upload to process and index your documents

​
Web Chat Integration

Import content from websites:
1

Select Knowledge Base

Open the knowledge base you want to add website content to
2

Add Website

Click Add Sources → Website URL
3

Enter URL Details

Provide the following information:
  • Website URL (e.g., https://example.com)
  • Source name (for identification)
  • Crawl configuration:
    • Depth (how many links deep to follow)
    • Maximum pages
    • Include/exclude patterns
4

Start Crawling

Click Add to begin the crawling process

​
GitHub Repository Connection

Connect a GitHub repository to your knowledge base:
1

Select Knowledge Base

Open the target knowledge base
2

Add GitHub Source

Click Add Sources → GitHub
3

Enter Repository Details

Provide:
  • Repository URL or owner/repo format
  • Branch name (defaults to main/master)
  • Source name for identification
4

Configure Options

Set:
  • File types to include/exclude
  • Directory patterns to include/exclude
  • Whether to include README files
5

Connect

Click Connect to import and index the repository

​
Managing Knowledge Bases

​
Viewing Knowledge Sources

From your knowledge base dashboard, you can see all added sources with the following information:
  • Source name
  • Type (document, website, GitHub)
  • Last updated time
  • Status (processing, completed, error)
  • Number of chunks/vectors

​
Updating Sources

To update an existing source:
  1. Select the knowledge source from your knowledge base
  2. Click the Update button
  3. For documents, upload a new version
  4. For websites or GitHub, trigger a re-crawl/re-sync

​
Removing Sources

To remove a knowledge source:
  1. Select the knowledge source from your knowledge base
  2. Click the Delete button
  3. Confirm the deletion
Deleting a source will immediately remove all its content from your knowledge base and affect any AI assistants using this knowledge.

​
Advanced Features

​
Custom Metadata

Add custom metadata to your knowledge sources for more targeted retrieval:
  1. Select a knowledge source
  2. Click Edit Metadata
  3. Add key-value pairs for relevant information
  4. Save changes
This metadata can be used for filtering in your AI assistant’s responses.

​
Content Filtering

Set up content filters to exclude certain information from your knowledge base:
  1. Go to your knowledge base settings
  2. Navigate to the Content Filters tab
  3. Set up rules based on:
    • Keywords or phrases to exclude
    • Regular expressions for pattern matching
    • Content categories to omit

​
Testing Knowledge Retrieval

Test how effectively your knowledge base retrieves information:
  1. Go to your knowledge base
  2. Click the Test tab
  3. Enter a sample question
  4. View retrieved chunks and their relevance scores
  5. Refine your knowledge base based on results

​
Best Practices

  • Organize logically: Create separate knowledge bases for different products, topics, or departments
  • Update regularly: Keep your knowledge base current by updating sources when information changes
  • Monitor quality: Use the testing feature to ensure accurate information retrieval
  • Balance comprehensiveness with relevance: Include enough information to be helpful without overloading the system
  • Use descriptive naming: Name knowledge sources clearly for easy management

​
Troubleshooting