video1.0<iframe src="https://www.loom.com/embed/f2e5ae7018c147e2ad001c6f7bb972c2" frameborder="0" width="1658" height="1243" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>12431658Loomhttps://www.loom.com12431658https://cdn.loom.com/sessions/thumbnails/f2e5ae7018c147e2ad001c6f7bb972c2-076ad78bb855831a.gif182.976This Loom explains how LogLists uses transcript summarization and taxonomy clustering to extract product insights without reading sensitive chat history. It uses the WildChats dataset with about 800k rows of human to AI conversations, summarizes sample transcripts with GLM 5.3 Flash, and clusters them in real time using a decision model. Viewers can zoom into clusters to see subsets and proportions of query types, such as one cluster where history accounts for about 10% of total queries. The agent then answers product questions, generating data grounded fictional user stories and turning them into a PRD or prompt to share with teams, with logging secured via a gVisor sandbox on Vultr and Netbird for remote control.