loading
loading
Combining keyword search and vector search, then merging the results, so you catch both exact terms and semantic meaning.
Pure vector search misses exact strings — error codes, product names, SKUs — because 'E_4012' has no meaningful embedding. Pure keyword search misses paraphrases. Hybrid runs both (typically BM25 plus cosine similarity) and fuses the ranked lists, usually with Reciprocal Rank Fusion, into one result set. It's the pragmatic default for production RAG because real queries are a mix of 'find this literal token' and 'find this idea.' Example: a support bot that nails both 'reset 2FA' (semantic) and the exact ticket ID a user pastes in.
Plainly
Think of Hybrid Search as a simple recipe for doing the work better. Combining keyword search and vector search, then merging the results, so you catch both exact terms and semantic meaning.
In practice
Use it when you need a repeatable method instead of guessing from vibes. In practice, define the owner, input, output, and failure mode before you rely on it.