In a video network context, video sessions can be aggregated and delivered to users, and smart replay can be provided responsive to user commands. Various and different commands are received from users at different times. Video sessions (e.g., program streams) are identified and/or created based on the commands. Video sessions are multicasted to the users. Joining multiple users to a video session reduces a need to create multiple identical or similar individual sessions, thereby increasing session capacity. Delivering a video session to users likely to be seeking a particular moment of interest in a video, based on an increased tolerance of received user commands, increases the user experience by delivering the moment of interest directly, without requiring multiple commands from users attempting to navigate towards the moment of interest.