What Are Embeddings and Why Should I Care?
The vector thing behind semantic search, recommendations, duplicate detection, and most RAG diagrams.
An embedding is a list of numbers that represents the meaning of some input: text, image, audio, or code. Similar things end up with similar vectors.
That sounds abstract until you use it.
What embeddings do
Embeddings let software answer questions like:
- Which document is closest to this query?
- Are these two support tickets about the same issue?
- Which products are semantically similar?
- Which chunks should we give the LLM before it answers?
Why RAG needs them
Retrieval-augmented generation usually works like this: split documents into chunks, turn chunks into embeddings, store them, embed the user's query, find similar chunks, put those chunks into the prompt, and ask the model to answer.
The LLM is not magically searching your documents. Your retrieval layer is feeding it context.
Where vectors live
They can live in vector databases like Pinecone, Weaviate, Qdrant, and Chroma, or inside Postgres with pgvector.
The important part is not the database brand. It is whether retrieval returns the right context.