Abstract

This disclosure describes techniques that enable a computing device (e.g., a smartphone or other mobile computing device) to present a single smart button or predictive UI element that predicts and automatically completes a task on the user's behalf. This prediction may be based on contextual signals the computing device logs over time, such as accessibility events, notifications, location, time of day, and a recognized activity of the user (e.g., walking, driving, or sitting still). Using these signals and a large language model (LLM), the computing device may learn to distinguish discrete, repeatable tasks appropriate for single-tap completion (e.g., opening a garage door, unlocking a vehicle charging port) from open-ended activity, such as browsing a social media feed, and may associate a recognized location with a semantic label (e.g., “home” or “work”) based on the user's patterns at that location. Once the computing device determines that the user's present context matches a predicted task, the computing device may surface an interface element, such as on a lock screen, dynamic home screen widget, or home screen, representing that task. Selecting this interface element may prompt the computing device (acting as an autonomous AI agent) to complete the task automatically by interacting with the relevant application in the background, allowing the user to complete a routine task through a single interaction rather than through a complex, multi-step process. Aspects of the disclosure may include predictive interfaces, contextual signal logging, activity recognition, semantic location labeling, lock screen and home screen interface elements, automated background task completion, intent prediction, and proactive assistants.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Download

Share

COinS