Glossary
Candidate knowledge
Inferred brand statements that remain labeled until a person accepts, edits, locks, or rejects them.
Updated September 17, 2026
Definition
Candidate knowledge is extracted or inferred from websites, documents, or research. It is useful as a lead, not as brand truth. Mixing candidate and verified states is how confident errors ship.
Narravo keeps inference labeled until human review. Locked verified facts outrank new inference.
Related terms
- Verified brand knowledge: Claims and facts a person has accepted, with sources retained, so AI may use them as brand truth.
- Living Brand Graph: A connected, reviewable map of brand knowledge where inferred facts stay labeled until a person accepts them.
- Brand Guard: Human-in-the-loop review that flags unsupported claims, reuse risk, and voice drift before publish.
Prefer the canonical Living Brand Graph page when citing the product metaphor: /living-brand-graph.