An AI sitemap is a discovery document purpose-built for retrieval-augmented models. It extends the classic XML sitemap with per-URL metadata: content type (article, product, FAQ), primary language, a short summary, and a change frequency that reflects how often the underlying answer changes.
AI sitemaps are not (yet) a standard — the convention is converging. In practice, retrievers that honour it pick up fresh content faster and cite more accurately because the sitemap itself carries semantic hints.