Skip to Content
Knowledge BaseOverview

Knowledge Base

Knowledge 101

The Knowledge Base is the content foundation for grounded answers and citations. It stores: files you upload, website content you scrape, RSS items, YouTube transcripts, text you paste, and synced cloud storage (Drive / OneDrive / Dropbox).

What It Is

A central content store where you add documents and data from multiple sources. Content is processed, segmented, and vectorized so that when you ask a question, the system finds the most relevant documents and includes them in the AI context.

Why It Matters

You get:

  • Better answers — AI can reference internal content
  • Citations — Traceable sources in responses
  • Automation inputs — Workflows can retrieve or use KB content

Data Source Types

  • File Upload – PDFs, Word, text files, images (up to 50 MB per file)
  • Website Scraper – Add URLs; content is scraped with configurable depth and page limits
  • RSS Feeds – Subscribe to feeds; articles sync automatically
  • YouTube – Add video URLs; transcripts are extracted and added
  • Text Upload – Paste or type plain text directly
  • Cloud Storage – Google Drive, OneDrive, Dropbox – connect and sync folders

Collections

Data source items can be organized into collections with the same permission model as Assistants (owner, company EDIT/VIEW/RESTRICTED, user/group when restricted). All Company is the label for unassigned items (visible to all company users). Use the collection selector to switch between All Company and collections you can view; create, edit, share, and delete collections; and Move to collection from the row context menu. New documents you add are created in the currently selected collection, or in All Company if none is selected.

See Collections for details.

Main Interface

  • Collection selector – All Company or a collection you can view; controls which items the grid shows and where new items are created
  • Add button – Opens a menu of data source types
  • Delete button – Remove selected items (enabled when items are selected)
  • Grid – Tree view of data source objects in the selected collection (or All Company) with columns: Icon, Title, Link, Source, Status, Created By, Last Updated

Understanding Status (Processing / Processed / Failed)

  • Processing — Decoder is extracting text and indexing it
  • Processed — Ready to use (searchable; can power answers + citations)
  • Failed — Error occurred; the content won’t be used until fixed

Important: Only Processed items are used by Assistants and Automations.

Next

Last updated on
ConversationsCollections