# On — AI Shopping Assistant > Powered by Aigency ## How It Works On is an AI-powered conversational shopping assistant. Visit https://on.aigency.bot/ and ask about any product — the assistant will find, compare, and recommend items. ## About On Run with On. Swiss performance running shoes, apparel, and accessories. Official site: https://www.on.com/ ## Questions you can ask - Find men's running shoes - Show me trail running shoes - Women's tennis sneakers - What's new in Cloudflow? ## Product Pages Product pages use the same URL paths as the brand's main website. For AI agents, crawlers and plain HTTP clients (no browser fetch metadata), each product page returns: - HTML with embedded JSON-LD (schema.org/Product), linking related products - JSON when requested with Accept: application/json header Browsers receive the app, which renders the same product on load. Every other page (home, legal pages) carries its readable content in the HTML too. ## Machine-readable versions - Send `Accept: text/markdown` for a Markdown version of the home, product, collection and legal pages. - Everything in one file: https://on.aigency.bot/llms-full.txt ## Sitemap https://on.aigency.bot/sitemap.xml https://on.aigency.bot/genui-sitemap.xml ## For AI agents Machine-readable description of this surface: https://on.aigency.bot/api/agent Agent guide in markdown: https://on.aigency.bot/agents.md WebMCP tools registered in the page (document.modelContext): search_products, get_product_details, get_current_view MCP endpoint (UCP manifest at https://on.aigency.bot/.well-known/ucp): POST https://on.aigency.bot/api/ucp/mcp, UCP catalog tools search_catalog, lookup_catalog, get_product; product ids are product URLs; no cart or checkout