{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/fcb5d9e7cc354a8393d964f2e2645092\" frameborder=\"0\" width=\"1280\" height=\"960\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":960,"width":1280,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":960,"thumbnail_width":1280,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/fcb5d9e7cc354a8393d964f2e2645092-d57aeff55293102b.gif","duration":300.157,"title":"Building a Multi Brand AI Assistant","description":"This Loom explains the AI brand assistant architecture and how it handles multi brand conversations without context leakage. The author describes an approach where each brand chat remains independent, and they avoid sending full chat history to the AI on every request by using a rolling summary stored in the system. They outline the routing setup under a single connection, including dedicated routers and services like the brand service for database interactions. The demo shows creating a new fitness brand and using the rollback summary to update brand context, such as suggesting and selecting a brand name."}