Knowledge Base

A Knowledge Base (KB) allows your agents to access custom documents, structured data, and domain-specific content, enabling them to generate accurate and context-aware responses.

By linking a Knowledge Base to your agent, you can:

  • Enhance document-based Q&A capabilities

  • Provide custom context for domain-specific tasks

  • Enable retrieval-augmented generation (RAG) workflows

Supported Content Types

A Knowledge Base can include:

  • Documents: TXT, PDF, DOCX, MD, HTML

  • Structured Data: JSON, CSV

  • Presentations: PPTX

Creating a Knowledge Base

  1. Navigate to “Knowledge” from the sidebar.

  2. Click “Create Knowledge Base”.

  3. Enter a Name for your Knowledge Base (description is not required).

  4. Upload your documents using any of the supported methods:

File Upload Options:

  • Upload New Document – Select local files or provide a URL to fetch the document

  • Upload New Folder – Import an entire folder for batch processing

  • Import from Cloud – Connect cloud storage such as Google Drive or Dropbox

After processing, your KB will appear in the Knowledge panel, and uploaded files will be listed with:

  • File Name

  • Created Date

  • Size

Managing Knowledge Bases

From the Knowledge page, you can:

  • Browse Folders & Files – View your organized KB content

  • Upload Additional Files – Add new documents to an existing KB anytime

  • Delete Files or KBs – Remove outdated content to maintain relevance

Linking Knowledge Bases to Agents

You can link KBs to any agent type during creation or editing:

  1. Open the agent in Prompt mode.

  2. Select the Knowledge Base to attach.

  3. Test in Preview to verify that the agent correctly references your KB content.

Example Use Cases:

  • Support Agent – Answer customer FAQs based on uploaded manuals

  • Internal Wiki Assistant – Provide employees with company policy info

  • Research Agent – Summarize insights from uploaded reports or scientific papers

Best Practices

  • Organize KBs by topic or domain to improve accuracy.

  • Use multiple smaller KBs instead of one large KB for easier updates.

  • Regularly update documents to keep agent responses up to date.

  • Verify outputs in Preview to ensure knowledge retrieval is functioning correctly.

PreviousModel APIsNextPublishing & Settings

Last updated

Was this helpful?