Concept / Term · New (AI-era)
Embedding search database
A database designed to store, index and search embeddings, commonly used for semantic search, recommendations and RAG retrieval.
Field note
A vector database stores numerical representations—embeddings—and searches for items whose vectors are close in a high-dimensional space. Nearness acts as a measure of semantic similarity.
Vector search grew from information retrieval and machine learning. Generative-AI applications made the term mainstream because retrieval-augmented generation depends on finding semantically relevant material.
Documents are split into passages, converted into embeddings and retrieved by similarity to a user’s question before an AI model writes its answer.
It can retrieve by meaning rather than exact keywords, but quality still depends on the embedding model, chunking, filters and evaluation.
In a sentence“The vector search found the returns-policy paragraph even though the question used different words.”