Skip to content

Document classification

Classification labels control how sensitive each document is and which assistants or audiences should use it.

LevelTypical contentExposure
publicMarketing FAQs, product specs, public policiesSafe for visitor-facing assistants and widgets
internalSOPs, team wikis, process docsInternal assistants; not for public embeds
confidentialPricing drafts, contracts, customer-specific runbooksRestrict to trusted assistants and roles
restrictedLegal, HR, security, PII-heavy materialHighest guardrails; admin review recommended

Classification is stored as metadata on each document snapshot and travels with chunks into Qdrant for filtered retrieval.

  1. Open a knowledge base → Upload or Import URL.
  2. Choose a level per file or accept the KB default classification.
  3. Save — indexing applies the label to all chunks from that document.
  1. Open the document in the knowledge base.
  2. Edit Classification and save.
  3. Chatevo updates chunk metadata; no re-upload required unless content itself changed.

When attaching knowledge bases, consider which classifications each assistant should access:

  • Public website widget — attach only KBs with public (and optionally internal) content
  • Staff copilot — may include internal and confidential KBs
  • Admin-only assistant — may include restricted documents

Pair classification with team roles so only admins and editors can upload confidential or restricted material.

  • Classification is an organizational policy tool — define who may upload each level in your internal guidelines
  • Audit feed and audit logs record document and KB changes for review
  • Snapshots preserve content history even when classification or live source pages change