You text the bot your video topic — the scenario gathers examples, composes a description, draws the image with a neural net and sends the ready thumbnail.