Given an input prompt/query from user - generate embedding - search for neighbor embeddings (chunks) in vector index - extract text of those chunks and build context for the user query Failure scenario: - No relevant chunks -> refuse to answer
Given an input prompt/query from user
Failure scenario: