Knowledge Bases

A knowledge base is a collection of documents (uploaded files or fetched web pages) that gives an agent factual context during calls (RAG). When attached to an agent, the agent answers from your content instead of guessing.

Where: left nav → Knowledge Bases (/knowledge-bases).

Create a knowledge base

1. Click Create Knowledge Base.
2. Enter a Name (e.g. “Product Documentation”) and an optional Description.
3. Click Create. You land on its detail page to add content.

Add content

On the detail page, add documents two ways:

• Upload Document — a file picker accepting .pdf, .txt, .docx, .md.
• Add URL — enter a web page URL; the platform fetches the page and adds its content as a document.

Each document shows a Status: Ready (usable), Processing / Pending (indexing — large files process asynchronously), Failed, or Static.

Small documents become “Static” context

A document under ~500 bytes is too small to chunk for search, so its content is injected directly into the agent’s prompt (its Chunks count shows “—”). This is normal and works well for short facts.

The detail page also shows Documents, Total Chunks and Total Size stats, plus Storage and Documents quota gauges with an Upgrade plan link.

Attach to an agent

Linking happens on the agent side, not here:

1. Open the agent → Knowledge tab.
2. Click Add Knowledge Base and select this base.
3. Click Save Changes on the agent.

Tips

• Supported upload types are only .pdf, .txt, .docx, .md.
• Re-uploading the same filename replaces the existing document (reprocessed with the new version) rather than adding a duplicate.
• Quotas are enforced: an oversized file or too many documents is rejected — the toast offers Upgrade plan.
• Deleting a document or a whole knowledge base is permanent.