Embeddings
Embeddings are numerical representations of text that capture its meaning, letting AI systems find passages that are semantically related to a query even when they share no exact keywords.
Embeddings turn text into coordinates that encode meaning. Each passage is converted into a long list of numbers (a vector) positioned so that pieces of text with similar meaning sit close together — regardless of whether they use the same words. This is what lets an answer engine match a conversational question to a relevant passage on your site.
Embeddings are why keyword stuffing is obsolete and why writing for meaning matters. Retrieval no longer looks for an exact phrase; it looks for the passage whose embedding is nearest to the question's embedding. So a paragraph that clearly and completely expresses one idea — the extractability discipline — embeds cleanly and gets matched, while a vague or padded one lands in fuzzy, low-relevance space and gets skipped.
Example. A page that says "how to lower your monthly energy bill" can be retrieved for the query "ways to spend less on electricity" because their embeddings are close in meaning, even with almost no shared words. That semantic match is embeddings at work.
Relevant pillar
Related terms
- Vector SearchVector search is a retrieval method that finds passages by meaning rather than keywords, comparing the numeric embedding of a query against the embeddings of indexed content to surface the closest matches.
- RAG (Retrieval-Augmented Generation)RAG is the technique behind most AI answer engines, where the model first retrieves relevant documents from the live web or an index and then generates an answer grounded in what it found.
- ChunkingChunking is how a retrieval system splits your page into smaller passages before indexing it, so AI engines retrieve and cite chunks of a page rather than the whole document.
- RerankingReranking is a second pass in retrieval where an initial set of candidate passages is reordered by a more precise relevance model, deciding which few actually make it into the AI's answer.