I don’t think it would be good UX with voice either. An agent could dynamically recreate something resembling the DoorDash UI which probably would work, but would lose out on the familiarity the user has with the actual doordash