Wayfair is running GPT-5.5 via the OpenAI API across a 40 million product catalog. The system handles enrichment tasks: organizing product information, sharpening differentiation between listings, and improving how attributes are understood at scale. Fiona Tan leads the team executing this.
The scale is the story here. Doing this manually across 40 million SKUs is not a staffing problem you solve by hiring. It is a structural problem, and this is how they are solving it. The specifics of how classification, attribution, and differentiation are being handled across that volume are worth reading for anyone building similar pipelines.
The next phase points Codex at what they call their gnarliest problems, meaning code-level automation is coming into the workflow. That transition from content enrichment to engineering assistance is where this gets more interesting and more replicable for other organizations watching how Wayfair is moving.
[WATCH ON YOUTUBE →]