Give it a list of topics and it queues one AI video render per topic. The renders run durably in the background; publishing each one still waits for your approval.
The steps
YouTube — render a batch of videos runs each_topic → queue_render → done.
Who can run it
YouTube — render a batch of videos reaches an instance's own resources, so it runs on an operator's Leumas Studio; open it here to read and adapt it.