{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/140b345f95c64353bcdcc556c6bb6bdd\" frameborder=\"0\" width=\"1262\" height=\"946\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":946,"width":1262,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":946,"thumbnail_width":1262,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/140b345f95c64353bcdcc556c6bb6bdd-8ed2baaa3b07ac1a.gif","duration":211.124,"title":"Automating Dispute Resolutions with Agents","description":"This Loom shows how Matie built an agent-based system to win Stripe chargeback disputes quickly by automatically assembling and verifying evidence. She describes a map that groups orders into clusters where agents can erase AP and device evidence, and notes merchants and Visa cannot tell the difference from the platform alone. For a selected dispute, multiple agents investigate history, package logistics and identity, and forensics, then a critic weights the results to route the dispute and produce the full evidence package for fast submission to Stripe. She demonstrates that the dispute can be won based on correct claims, with a human remaining in the loop to verify before submitting."}