Shopify's Global Catalogue turns inconsistent merchant listings into product records that search and shopping assistants can compare. Models examine images and text to assign categories and extract attributes such as color, material, and brand. A separate matching process identifies listings for the same product, removes questionable matches, and combines complementary information into a common record. This addresses a concrete obstacle to product discovery: two merchants can describe the same item differently, while a useful attribute may appear only in a photograph.
Terms in this piece · Glossary
multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
