In Cognigy AI, Lexicon and Knowledge serve distinct purposes in natural language understanding. Lexicon is rule-based, using hardcoded synonyms to normalize input before intent recognition — no word vectors involved. In contrast, Knowledge leverages semantic search with vector embeddings to retrieve answers from documents, making it ideal for dynamic Q&A use cases. Knowing when to use each unlocks smarter virtual agents.