AI Overview
An AI Overview is a generated summary shown above search results, produced by retrieving pages from the index and having a language model synthesise them, with links to the sources used. The retrieval step is ordinary search, so an unindexed page cannot be cited.
The pattern underneath is retrieval-augmented generation: run the query against the index, take the retrieved documents as context, and generate an answer grounded in them. This matters because it means the AI layer inherits every constraint of the search layer. If a page is not crawled, it is not indexed. If it is not indexed, it cannot be retrieved. If it cannot be retrieved, it cannot be summarised or cited.
For complex questions the query may be decomposed into several related searches whose results are combined — Google has described this fan-out behaviour in the context of AI Mode. One answer can therefore draw on many pages, each supplying a fragment.
Two claims to be sceptical of. There is no AI-specific structured data type, and
adding invented properties achieves nothing. And Google has stated that it does
not use llms.txt — publishing one is harmless but is not an AI visibility
tactic.
What does appear to matter is unremarkable: be indexable, answer the question directly in a self-contained passage, be specific enough to quote, and name your sources.