{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/9cbedaea290d4044b54d5e8ff0713728\" frameborder=\"0\" width=\"1662\" height=\"1246\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1246,"width":1662,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1246,"thumbnail_width":1662,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/9cbedaea290d4044b54d5e8ff0713728-68b8f842479097a1.gif","duration":203.601,"title":"Automating Fair, Cost-Effective Furniture Refunds","description":"This Loom explains how Wayfair designs a fair, low-cost refund decision system for customers. The presenter argues the key question is not how much to refund, but what is fair while minimizing Wayfair costs, noting that bulky furniture often makes return shipping cost more than the item itself, so customers may keep items with partial refunds. The architecture uses a multi-layer approach with an exception-team model and a strict JSON React loop on the client, plus tool support via 16-processor MCP2 with no network or deployed service, and it escalates to a human agent only in hard cases or when amounts exceed a cap. It also logs every decision to an auditor trail, using an abuse model with both transparent and learned scores and a financial model to choose between partial refund, store credit, or replacement."}