What is vector search?

Vector search is a way of finding information by meaning instead of exact wording. It turns text, images, or audio into lists of numbers called embeddings, then finds the items whose numbers sit closest together. Search for "cheap laptop" and it can surface a result that says "affordable notebook" — because the two ideas land near each other in that number space, even with no shared words.

How does vector search work?

An embedding model reads a piece of content and outputs a vector: a long list of numbers that captures its meaning. Similar things get similar vectors, so the whole collection becomes points scattered across a high-dimensional space. To answer a query, the system embeds your query the same way, then finds which stored vectors sit nearest to it and ranks them from most to least similar.

Nearness is measured with a math formula. Cosine similarity compares the angle between two vectors, while Euclidean distance compares the straight-line gap between them. Either way, the closest points come back as the best matches.

How is it different from keyword search?

Keyword search matches the literal words you type. Vector search matches the concept behind them, so it copes with synonyms, paraphrases, typos, and vague intent that keyword search tends to miss.

Where is vector search used?

It shows up anywhere "find me things like this" matters more than an exact string match:

• Semantic search across documents, help centers, and product catalogs.
• Retrieval-augmented generation (RAG), pulling relevant context to feed a language model.
• Recommendations, surfacing items similar to what someone already liked.
• Image and audio matching, comparing content by how it looks or sounds rather than by its filename.

In practice many systems blend the two, using keyword search for precision and vector search for meaning.