Found by construction
“Does Google find my site?” has the same answer as “does ChatGPT find my site?”: yes, by construction. A product page says it is a Product — name, price, image, description — in the fields a search engine expects. One file renders the page and its structured data, so the two cannot disagree.
Structured data on every page
Each page carries its JSON-LD: the organisation, the website, the page and the entity it is about, with references resolved. Nobody writes it separately — it is the same record the page is rendered from, made fit for the outside.
Addresses from the data
A page exists because an entity claims a slug. Its canonical address, its language links (hreflang) and its sitemap line all derive from that one fact. On a site with more languages, every entity carries its address per language; a page without a translation simply does not exist in that language — no empty page, no invented one. Move a page by changing its slug, and a redirect catches the old address.
What a crawler finds
- A sitemap that cannot drift.
/sitemap.xmlis computed from the same register that makes pages exist — per language, alternates included. A sitemap entry for a page that is gone is impossible. - The first file a crawler reads.
robots.txtpoints at the sitemap of every language, the content signals and the machine views. - A feed when you want one. Name the types or lists that feed it, and
/feed.xmlis announced and kept current.
Publish an entity and every one of these follows on the next request. Nothing to export, nothing to regenerate.
What you control
The views write themselves; the words are yours. A title and a description in plain language, a sentence each. An image with a caption. An address that says what the page is. The order of the curated sections in llms.txt, which decides what an assistant reads first. Ask your agent how a page looks to a machine, and it reads the machine view back to you.