In some aspects, disclosed herein is a device that stores one or more pre-recorded or dynamically generated voice commands that are capable of activating a virtual assistant via a voice-based user interface and causing the virtual assistant to perform a specific task. The device may be activated manually, e.g., by pushing a button. In some aspects, the device relieves a user of a virtual assistant of the need to speak a specific command in order to cause a virtual assistant to perform a specific task.