Definition. The technical process by which an AI system fetches relevant content from a source to embed it in a generated answer, typically via embeddings, vector search, reranking and filtering. It requires content that can be chunked cleanly, identified uniquely and is semantically annotated.

Why it matters. If your own website is to be a source in an AI answer, it has to be retrievable in the first place. Content that exists only as a visual composition (sliders, images without text) drops out of retrieval.

Related. Retrieval-ready content, AI Search, Structured Content, Semantic HTML