A computer system may receive, from an acoustic transducer and/or a portable electronic device, information specifying one or more desired operations that are to be performed. This information may include natural language describing the one or more desired operations. In response, the computer system may transform the natural language into the one or more desired operations, and may identify a subset of the electronic devices based on the one or more desired operations and a predetermined supervised learning model that relates desired operations and the electronic devices. Next, the computer system may determine a set of commands based on the one or more desired operations and the subset of the electronic devices, and the computer system may provide the set of commands to the identified subset of the electronic devices, which may perform the one or more desired operations.