Retrieval
Retrieval is the step where an AI system fetches candidate products or sources before it writes an answer. The model does not answer from memory alone; it pulls in current pages or catalog entries and grounds its response in them.
How it works
Given a question, the system searches an index or a product feed, takes the most relevant candidates, and writes an answer grounded in them, usually with citations. This is often called retrieval-augmented generation, or RAG. For shopping, the candidates are products, and query fan-out means retrieval runs many times per question.
A commerce example
If a shopper asks for a waterproof daypack and your pack is not retrieved for that search, it cannot be compared or recommended, no matter how good it is. Retrieval is the gate everything else sits behind.
What this means for your store
Being retrievable comes down to being indexed, readable as plain HTML, and complete enough to match. Stay indexed in Google and Bing, keep your product data structured, and make sure the facts are in the page an AI crawler receives.