Screen tutorial with a narrator

The brief Sensei follows

Make a screen tutorial that teaches [What the tutorial teaches, in one sentence] in [Your product name]. Call the tutorial maker with action start: topic, product_name from [Your product name], start_url = [The link the tutorial starts on] (you drive the browser through the product yourself), steps = [What the viewer does, one line per step] one line each in order, result_url = [The finished result the viewer will make] if given, voice_url = [A 10 to 20 second clip of the narrator's voice] if given, and login_email / login_password only if [A demo login for your product, if the flow needs one] gives a demo login for their own product. Only if the user uploaded [Or upload one screenshot per step instead] instead, pass those as screens with one step line each. App screens are never regenerated or sent to a video model. Tell the user it takes 10 to 20 minutes, keep calling status until it is ready, then deliver the video and the SRT, and name any step the product could not do (it is skipped, never faked).