Skip to main content

Knowledge Base

The knowledge base allows you to upload documentation and reference material that agents can use to answer questions. This powers RAG (Retrieval-Augmented Generation) so agents have access to your organization's specific information.

How It Works​

When you add a knowledge source:

  1. Content is fetched - URLs are crawled, files are parsed
  2. Text is chunked - Content is split into searchable segments (~500-1000 tokens)
  3. Chunks are embedded - Each chunk gets a vector embedding for semantic search
  4. Agents can search - When users ask questions, agents find relevant chunks

Knowledge Scopes​

Knowledge sources can be scoped at different levels:

ScopeAvailable ToUse Case
SystemAll workspacesRaven Docs help documentation
WorkspaceAll spaces in workspaceCompany policies, shared docs
SpaceOnly that spaceTeam-specific documentation

Inheritance​

  • Agents in a space can access: Space + Workspace + System knowledge
  • More specific scopes take priority in search results

Adding Knowledge Sources​

Workspace-Level Knowledge​

For documentation available across all spaces:

  1. Go to Settings → General → Knowledge
  2. Click Add Knowledge Source
  3. Enter a name and URL
  4. Click Add Source

The source will show as "Processing" while being indexed, then "Ready" when available.

Space-Level Knowledge​

For team or project-specific documentation:

  1. Open the space settings (gear icon)
  2. Go to the Knowledge tab
  3. Click Add Knowledge Source
  4. Enter a name and URL
  5. Click Add Source

Source Types​

TypeDescriptionStatus
URLWeb page that gets fetched and indexedAvailable
FileUploaded document (PDF, DOCX, MD, TXT)Coming soon
PageInternal Raven Docs pageComing soon

Source Status​

StatusMeaning
PendingQueued for processing
ProcessingCurrently being chunked and embedded
ReadyAvailable for agent searches
ErrorProcessing failed (check error message)

Managing Sources​

Refresh a Source​

To re-fetch and re-index a URL source:

  1. Find the source in the knowledge list
  2. Click the menu (three dots)
  3. Select Refresh

This is useful when the source content has changed.

Delete a Source​

  1. Find the source in the knowledge list
  2. Click the menu (three dots)
  3. Select Delete

All indexed chunks will be removed.

Best Practices​

  1. Use specific, focused sources - Smaller, targeted documents work better than large general ones
  2. Keep URLs up to date - Refresh sources when content changes
  3. Scope appropriately - Put shared docs at workspace level, team docs at space level
  4. Check processing status - Ensure sources are "Ready" before expecting agents to use them
  5. Review error messages - If processing fails, check if the URL is accessible

How Agents Use Knowledge​

When a user asks a question:

  1. The query is converted to a vector embedding
  2. Similar chunks are found using semantic search
  3. Top matching chunks are included in the agent's context
  4. The agent uses this context to answer the question

This means agents can answer questions about your specific documentation without needing to be trained on it.

Example Use Cases​

Company Documentation​

Add your company wiki, policies, and procedures at the workspace level so all teams can ask questions about them.

API Documentation​

Add your API docs so developers can ask the agent "How do I authenticate?" or "What's the endpoint for creating users?"

Project Documentation​

Add project-specific docs at the space level so the team can ask "What's our deployment process?" or "How do we handle errors?"