Towards Design Guidelines to Support Young Adults' Digital Wellbeing with Large Language Models

The widespread adoption of smartphones has consolidated access to a broad range of services into a single device. This convenience, however, also enables purposeless use patterns that can compromise digital wellbeing, particularly among young adults. Existing digital wellbeing applications often exhibit limited long-term efficacy, as they prioritize restriction rather than guiding users through a gradual, sustainable behavioral change process. To address this limitation, this paper investigates the potential of Large Language Models (LLMs) to enhance young adults' digital wellbeing by exploiting their capacity for personalized, context-aware content generation. Drawing on prior literature and a case study simulating user-application interactions across four representative personas, we propose an initial set of design guidelines aimed at supporting users through a progressive and adaptive process. Future work will evaluate these guidelines through an in-the-wild study of a mobile application designed in accordance with the proposed LLM-driven approach.