Publishing an ontology: graphs, feeds and llms.txt

Publish the ontology as schema.org @graphs on pages, a croutons.json feed of atomic Q/A blocks, an llms.txt index, and the same @ids across every owned surface.

Publishing

A mapped ontology that stays in a spreadsheet helps nobody. Publish it:

1. schema.org @graph on every page, referencing the core @ids.
2. A croutons feed - atomic question/answer blocks with stable ids and sources, so AI systems can cite specific fragments instead of paraphrasing the homepage.
3. llms.txt - a plain index of the site's key pages and entity ids, written for language models and their crawlers.
4. Same @ids everywhere - sub-brands, profile links, directory listings. The graph's job is identity; consistency is the mechanism.

Measure the result by citation, not by validation: when you ask an AI system about the company, does it name the entity, the category and the geography correctly - and does it point at a source you control?

That's the published ontology doing its job.