A growing share of buyers now ask an assistant before they ask a search engine, and an assistant does not return ten blue links. It returns an answer, and it names a handful of sources. That is a different game with a different winning move: you are no longer trying to rank a page, you are trying to be the thing that gets quoted.
Answers are lifted, not reconstructed
An assistant will far more reliably quote a self contained paragraph that directly answers a question than it will assemble one out of a hero headline, a card grid and an accordion. Most marketing sites are built the second way, because that is what converts a human reader. The fix is not to abandon the design, it is to make sure that somewhere on the page there is a clean, complete, quotable answer to the question the page is about.
Structure is the difference between being read and being understood
Schema and entity work stopped being a technical nicety the moment machines started summarizing pages for people. Marking up what a page is, who published it, what business it belongs to and how that business relates to its locations and services gives an assistant something unambiguous to reason about. Without it, the assistant is guessing from prose, and it guesses in favour of whoever made it easy.
The llms.txt convention and the AI crawler standards are the newer, cheaper half of this. They cost almost nothing to implement and they tell the crawlers that matter what is on the site and where.
Reputation still carries
Assistants lean heavily on sources that other sources agree with. Reviews, citations, mentions and coverage do the same job here that they always did in organic search, which means the unglamorous work of being consistently and verifiably who you say you are is still the foundation. This is not a separate discipline bolted onto search optimization. It is search optimization, aimed at a reader that does not click.
Measure it, because you can
The useful part is that citation is observable. You can ask the major assistants the questions your buyers ask and record which businesses get named, then watch that set change as the work lands. We track that for clients rather than treating AI visibility as something you hope for.
