Systems, methods, and computer program products to perform an operation comprising receiving, by a first device, metadata describing a first action of a plurality of actions performed on a second device, identifying a target object of the first action based on the metadata describing the first action, and outputting, by the first device: a sequence of images depicting performance of the first action on the second device, a textual instruction specifying how to perform the first action on the first device proximate to the target object, and an audio instruction specifying how to perform the first action on the first device.