Reverse Engineering How Muse Shops
How Meta's Muse agent picks the products it shows: the catalog search, the Unicorn system behind it, hidden seller ratings, and the evidence behind every number.
Blog
Short, dated posts on what appears in AI shopping tests: which brands and products show up, which destination appears, and what merchants can inspect. Longer research lives in the Genesis archive.
How Meta's Muse agent picks the products it shows: the catalog search, the Unicorn system behind it, hidden seller ratings, and the evidence behind every number.
A dated note on 248 observed brand mentions, owned-store links, and the limits of treating returned URLs as evidence of clicks or sales.
A field note on Shopify catalog matches, official product cards, and the conditions under which a usable product-page path appears.
A bounded case for measuring ChatGPT and Google’s Gemini and AI Mode first, then deciding which other surfaces belong in the next test.
How a shopper’s sentence becomes product destinations, and how a routing outcome can be measured without implying purchase completion.
We apply the research to merchant questions about product visibility, returned destinations, and access restrictions.