Wayfair uses OpenAI to fix millions of product tags

Wayfair has gone all-in on OpenAI, embedding its models into core systems to fix product catalog chaos and speed up supplier support. Instead of treating AI like a fun side project, the home goods giant built it directly into the workflows that handle roughly 30 million products. The results are staggering: 2.5 million product tags corrected and 41,000 supplier support tickets automated every month. Previously, Wayfair relied on manual reports or expensive custom AI models that didn’t scale across 47,000 different tags. Now, a single ‘definition agent’ ingets web and internal data to understand what each tag means, then classifies products across hundreds of categories. This new system is expanding coverage 70 times faster than a year ago. Wayfair has also deployed 1,200 ChatGPT Enterprise seats across the company. Why it matters: This is a blueprint for how massive retailers can finally tame unwieldy product data at scale, directly reducing customer returns and building trust—proving generative AI isn’t just for chatbots but for fixing the messy operational backends that define the customer experience.