Skip to main content

​
What are Knowledge Bases?

Knowledge bases are organized collections of information that give your AI assistant the expertise it needs to answer questions accurately and helpfully. Think of them as the “brain” of your assistant - the more comprehensive and well-organized your knowledge base, the smarter your assistant becomes.

​
Knowledge Base Interface

​
Getting Started

When you first access the Knowledge Base tab in your assistant configuration, you’ll see: No Knowledge Base Connected This shows that no knowledge base is currently connected to your assistant.

​
Creating a Knowledge Base

Click the “Create Knowledge Base” button to get started: Create Knowledge Base This will create a new knowledge base specifically for your assistant and automatically connect it.

​
Selecting an Existing Knowledge Base

If you already have knowledge bases, you can select from existing ones: Select Knowledge Base Choose the knowledge base that best fits your assistant’s purpose.

​
Knowledge Base Management

Once connected, you’ll see the knowledge base management interface: Knowledge Base Connected This interface provides three main tools:
  • Web Crawler: Add content from websites
  • File Uploader: Upload documents (PDF, Word, etc.)
  • File Importer: Import files from various sources

​
Advanced Configuration

After your knowledge base has data and is synced, additional options become available in the Configuration tab: Strict Mode and Keywords These options include:
  • Trigger Keywords: Words that automatically trigger knowledge base searches
  • Strict Mode: Control whether the assistant uses only knowledge base information or both knowledge base and general knowledge

​
How Knowledge Bases Work

​
Information Storage

Your knowledge base stores information in multiple formats:
  • Documents: PDFs, Word documents, text files
  • Web Content: Crawled websites and online resources
  • Custom Entries: Manually created knowledge snippets
  • Structured Data: Organized information for easy retrieval
When a user asks a question, your assistant:
  1. Analyzes the query to understand what information is needed
  2. Searches the knowledge base for relevant content
  3. Retrieves the best matches based on relevance and accuracy
  4. Uses the information to generate a helpful response
The quality of your assistant’s responses directly depends on the quality and organization of your knowledge base.

​
Types of Knowledge Content

​
Document Uploads

Upload existing documentation and content: 
PDF files (.pdf)
PowerPoint files (.pptx) - automatically converted to PDF
Word documents (.docx)
Text files (.txt)
Markdown files (.md)
Images (JPEG, PNG, GIF, WebP)

​
Web Content Crawling

Add online resources to your knowledge base:
  • Website URLs: Crawl entire websites or specific pages
  • Blog Posts: Include articles and blog content
  • Documentation Sites: Add technical documentation
  • Knowledge Articles: Include help center content

​
Cloud Storage Integrations

Connect to your existing cloud storage:
  • Google Drive: Import files directly from Google Drive
  • OneDrive: Access Microsoft OneDrive files
  • SharePoint: Import SharePoint documents
  • Other cloud storage: Through integration services
Only crawl websites you own or have permission to use. Respect robots.txt and rate limits.

​
Custom Knowledge Entries

Create specific knowledge snippets:
  • FAQ Entries: Common questions and answers
  • Procedures: Step-by-step instructions
  • Definitions: Important terms and concepts
  • Examples: Real-world use cases and scenarios

​
Organizing Your Knowledge Base

​
Categories and Tags

Organize your content for better retrieval:
  • Categories: Group related content (e.g., “Product Features”, “Troubleshooting”)
  • Tags: Add specific labels for easy searching
  • Hierarchy: Create logical content structures

​
Content Prioritization

Not all content is equally important:
  • Primary Sources: Most authoritative and up-to-date information
  • Secondary Sources: Supporting documentation and examples
  • Reference Material: Background information and context

​
Best Practices for Knowledge Bases

​
1. Quality Over Quantity

It’s better to have 100 high-quality, accurate documents than 1000 outdated or incorrect ones.
  • Ensure information is accurate and current
  • Remove outdated or incorrect content
  • Regular reviews and updates
  • Fact-check important information

​
2. Clear Organization

  • Use descriptive file names
  • Organize content logically
  • Add appropriate tags and categories
  • Create clear content hierarchies

​
3. Comprehensive Coverage

  • Cover all common use cases
  • Include troubleshooting information
  • Provide examples and context
  • Address edge cases and limitations

​
4. Regular Maintenance

  • Schedule regular content reviews
  • Update information as needed
  • Remove obsolete content
  • Monitor usage and effectiveness

​
Knowledge Base Analytics

Track how your knowledge base is performing:
  • Most Used Content: See what information is accessed most
  • Search Patterns: Understand what users are looking for
  • Gaps: Identify missing information
  • Effectiveness: Measure response quality and user satisfaction

​
Security and Privacy

​
Data Protection

  • All content is encrypted in transit and at rest
  • Access controls and permissions
  • Audit logs for content access
  • Compliance with data protection regulations

​
Access Control

Important: Knowledge base content is accessible to anyone who has access to your assistant or chat embed. This includes:
  • Users of your embedded chat widget
  • Anyone with access to publicly shared assistants
  • Team members you’ve invited to use your assistant

​
Privacy Considerations

  • Public Accessibility: Upload only content you’re comfortable sharing publicly
  • Sensitive Information: Never upload confidential or private data to knowledge bases
  • Content Review: Regularly review what information is available through your assistants
  • Access Management: Be mindful of who has access to assistants with knowledge bases
For private data analysis, use MCP tools in your workspace instead of knowledge bases.

​
Content Ownership

  • You retain full ownership of your content
  • Easy export and backup options
  • Complete deletion capabilities
  • No training on your proprietary data

​
Next Steps

Assistant Configuration

Learn how to connect knowledge bases to your assistants.

Creating Assistants

Create AI assistants that can use your knowledge base.

Data Retrieval Methods

Understand the differences between knowledge base and MCP data access.

Workspace

Use the workspace to interact with your knowledge base data.
⌘I