This presentation explores the opportunity for operators to improve voice search using AI that understands natural, conversational queries like ‘Give me something about space’. It will explain how AI can combine video, images and text to present a structured results list that reflects individual preferences in an image-based search interface. This solution is already live at an operator, and suggests content from across traditional TV, on-demand and streaming services. We hear how the product was defined and learn about the backoffice guardrails that are in place, including secure routing between AI, recommendation engine and UI. This discussion includes usage results and considers the full impact a new generation of voice-based search can have for TV.