video1.0<iframe src="https://www.loom.com/embed/085ad5c6e2c54efd9bca62e15c41b664" frameborder="0" width="1108" height="831" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>8311108Loomhttps://www.loom.com8311108https://cdn.loom.com/sessions/thumbnails/085ad5c6e2c54efd9bca62e15c41b664-945d9effa9baf607.gif437.117This Loom demonstrates a multitenant AI customer support platform called VenusRAM and how it works end to end. The speaker built it by forking Chatwoot, then added AI orchestration with a knowledge base retriever, conversation memory via a sliding window, and a knowledge graph using facts enriched from a self-hosted Graffiti framework. Audio inputs are transcribed with Whisper (English now, Uzbek planned), and the system runs on GCP with Docker and includes Langfuse tracing for LM cost visibility. The Loom also shows escalation to a human agent based on an escalation flag and a working Uzbek translation that the open source Chatwoot version lacked.