Abstract: Techniques for presenting personalized content to a user are described. A smart home system may determine activity data (including present activity data and device usage history) specific to interactions involving a first device, and generate activity embedding data representing the activity data. The smart home system may also receive a first list content for presentment, and generate a second list of content using the first list and the activity embedding data. The smart home system may send the second list to the first device. The first device may determine a third list of content based on the second list and an identify of a user presently interacting with the first device, such that the third list of content is specific to interactions involving the first device and the user. The first device may then present at least one instance of content represented in the third list.

Abstract: Devices and techniques are generally described for local device embeddings for automation. In various examples, first data representing first state change data for network-connected computing devices configured in communication with a first network may be determined. The first data may be input into a first machine learning model. In some examples, the first machine learning model may generate first embedding data representing a combination of the first data and second data. In some examples, the second data may represent historical state change data for the network-connected devices. In some examples, the first embedding data may be stored in memory. A first action may be performed by a first network-connected device based at least in part on the first embedding data.

Abstract: Systems and methods for target-device resolution are disclosed. A user may speak a user utterance requesting an action to be performed with respect to an accessory device, such as a smart-home device. The user utterance may include an identifier for the accessory device, but that identifier may not correspond to a naming indicator of an accessory device and/or may correspond to multiple naming indicators. In these examples, one or more models are utilized to identify which accessory device is most likely to correspond to the accessory device targeted by the user utterance for operation.

Abstract: This disclosure describes, in part, techniques for utilizing global models to generate local models for electronic devices in an environment, and techniques for utilizing the global models and/or the local models to provide notifications that are based on anomalies detected within the environment. For instance, a remote system may receive an identifier associated with an electronic device and identify a global model using the identifier. The remote system may then receive data indicating state changes of the electronic device and use the data and the global model to generate a local model associated with the electronic device. Using the global model and/or local model, the remote system can identify anomalies associated with the electronic device and, in response to identifying an anomaly, notify the user. The remote system can further cause the electronic device to change states after receiving a request from the user.

Abstract: Systems and methods for device configuration usage optimization include identifying device configurations that are likely to result in device usage cessation within a predetermined period of time. When device usage cessation is determined to be likely, alternative device configurations are determined that are likely to increase the chances of device usage. One or more recommendations for the alternative device configurations may be sent to user devices, and when accepted, devices may be reconfigured, device functionalities may be enabled, and/or additional devices may be acquired.

Abstract: Systems and methods for floor plan generation for device visualization and use are disclosed. For example, image data may be utilized to segment a building from non-building elements in an image, and the outline of the building may be utilized to generate a floor plan of the building. A user interface may be generated to present the floor plan and allow for placement of representations of walls, doors, windows, and electronic devices within the floor plan. Placement of these elements may be performed in response to input from a user and/or may be performed automatically utilizing naming indicators, device-affinity data, historical usage data, signal-strength data, etc. Once placed, the device representations may be utilized to operate the devices and display related information, such as alerts.

Abstract: Systems and methods for device state reversion are disclosed. For example, a requested and/or scheduled device state change may occur, and prior to the device state change, devices may be queried for their device states. This prior device state data may be saved. A user may provide an undo request and the prior device state data may be utilized along with current device state data to select a device to revert device state on, as well as the device state to revert to. In more complex situations and/or when prior state data is unavailable, machine learning techniques may be utilized to select the target device and device state.

Abstract: An Applicant Assistant System may receive, from a user device, a user input indicating a query or task associated with information displayed in a user interface. The device may determine one or more components associated with the user interface and access component metadata of the determined one or more components, the component metadata for each of the one or more components indicating one or more of a fact, a data link, or an action. A prompt comprising at least some of the component metadata, at least some context information, and an indication of one or more available response elements may be provided to a large language model (LLM). The LLM may response with one or more response elements that may be processed by the System, such as to cause updates to the user interface on the user device.

Abstract: An inspection application may display, on a user device, a user interface including at least a portion of a vehicle inspection report including a plurality of inspection categories. The inspection application may configure the user device to obtain inspection information associated with a vehicle, the inspection information comprising photographs, videos, audio, and/or text. The inspection application and/or a network-accessible inspection assistant system may generate a prompt including at least a portion of the inspection information and information indicating potential vehicle defects. The prompt may be transmitted to a large language model that returns a response indicating any potential vehicle defects identified in the inspection information. The vehicle inspection report may then be updated to indicate the potential vehicle defects identified by the language model.

Abstract: Systems and methods for intelligent device grouping are disclosed. An environment, such as a home, may have a number of voice-enabled devices and accessory devices that may be controlled by the voice-enabled devices. One or more models, such as linguistics model(s) and/or device affinity models may be utilized to determine which accessory devices are candidates for inclusion in a device group, and a recommendation for grouping the devices may be provided. Device-group naming recommendations may also be generated and may be sent to users.

Abstract: Systems and methods for personalized device routines include determining devices associated with user account data and generating device usage data indicating aspects of smart device usage over a period of time. The usage data may then be utilized to identify candidate smart devices, time indicators, and trigger event types to associate with candidate routines. One or more of the candidate routines may be recommended to a user based at least in part on the usage data and the trigger event types at issue to personalize the device routine recommendations.

Abstract: Systems and methods for target-device resolution are disclosed. A user may speak a user utterance requesting an action to be performed with respect to an accessory device, such as a smart-home device. The user utterance may include an identifier for the accessory device, but that identifier may not correspond to a naming indicator of an accessory device and/or may correspond to multiple naming indicators. In these examples, one or more models are utilized to identify which accessory device is most likely to correspond to the accessory device targeted by the user utterance for operation.

Abstract: Systems and methods for state prediction of devices are disclosed. A group of reference devices may be identified and a subset of the reference devices may be identified and/or determined based at least in part on a degree of similarity between reference usage-patterns associated with the reference devices and a usage pattern of a target device. The current state of the subset of the reference devices may be determined and may be utilized to determine a probability that the target device should be in a given state. The state prediction information may be utilized for one or more actions, such as sending recommendations, target inference operations, and/or device configuration, for example.

Abstract: This disclosure describes, in part, techniques for utilizing global models to generate local models for electronic devices in an environment, and techniques for utilizing the global models and/or the local models to provide notifications that are based on anomalies detected within the environment. For instance, a remote system may receive an identifier associated with an electronic device and identify a global model using the identifier. The remote system may then receive data indicating state changes of the electronic device and use the data and the global model to generate a local model associated with the electronic device. Using the global model and/or local model, the remote system can identify anomalies associated with the electronic device and, in response to identifying an anomaly, notify the user. The remote system can further cause the electronic device to change states after receiving a request from the user.

Abstract: Systems and methods allow for floor plan generation for device visualization and use. For example, image data may be utilized to segment a building from non-building elements in an image, and the outline of the building may be utilized to generate a floor plan of the building. A user interface may be generated to present the floor plan and allow for placement of representations of walls, doors, windows, and electronic devices within the floor plan. Placement of these elements may be performed in response to input from a user and/or may be performed automatically utilizing naming indicators, device-affinity data, historical usage data, signal-strength data, etc. Once placed, the device representations may be utilized to operate the devices and display related information, such as alerts.

Abstract: Systems and methods for device functionality control utilizing activity prediction are disclosed. For example, an activity model is utilized to identify activities associated with sensor data collected from connected devices within an environment. Target device usage data is then associated with the activities to determine probability values that the target device will be operated when the identified activities are detected. Commands may be generated and sent to the target device to perform a function when the identified activities are detected.

Abstract: Systems and methods for intelligent device grouping are disclosed. An environment, such as a home, may have a number of voice-enabled devices and accessory devices that may be controlled by the voice-enabled devices. One or more models, such as linguistics model(s) and/or device affinity models may be utilized to determine which accessory devices are candidates for inclusion in a device group, and a recommendation for grouping the devices may be provided. Device-group naming recommendations may also be generated and may be sent to users.

Abstract: Techniques for automatically combining devices into a single group of devices, and splitting devices into multiple groups of devices are described. A machine learning model may process device profile data, associated with devices registered to two different users, and determine the devices should be combined into a single group of devices. Such enables a user to control each of the devices, of the two different users, but providing user inputs to a single device of the group. The machine learning model may also process device profile data, associated with devices registered to a single user, and determine the devices should be split into two or more groups of devices. Such may decrease the likelihood that a system may inadvertently control a device not intended by a user.

Abstract: Described are systems, methods, and apparatus that gathers environment condition data from different sensors at various locations within an area, aggregates the environment condition data to produce aggregated environment condition scores for the area and provides the aggregated environment condition scores to different locations within the area. While sensor data from a single sensor/device, such as a camera may provide low quality environment information, by collecting and aggregating information from multiple sensors and/or locations in the area, highly accurate aggregated environment condition scores for environment conditions may be realized. The aggregated environment condition scores may be provided to various locations within the area as representative of the environment condition at that point in time within the area, regardless of whether those locations have sensors.

Abstract: Systems and methods for target-device resolution are disclosed. A user may speak a user utterance requesting an action to be performed with respect to an accessory device, such as a smart-home device. The user utterance may include an identifier for the accessory device, but that identifier may not correspond to a naming indicator of an accessory device and/or may correspond to multiple naming indicators. In these examples, one or more models are utilized to identify which accessory device is most likely to correspond to the accessory device targeted by the user utterance for operation.