Vybly: A User Centric Mood Based Creative Assistant with Adaptive Suggestion Interface

In recent years, increasing levels of stress, anxiety, and emotional imbalance have created a growing demand for intelligent systems capable of supporting emotional well being and adaptive user interaction. Existing wellness applications mainly focus on productivity and entertainment features while providing limited emotionally adaptive interaction. This paper presents Vybly, a mood based creative assistant designed to generate personalized activity recommendations according to the emotional condition of the user. The proposed system aims to improve emotional engagement, creativity, and mental wellness through structured recommendation mechanisms. The application is developed using Flutter for cross platform mobile deployment and Firebase Authentication for secure user management and session handling. The proposed system integrates a lightweight CNN based facial emotion recognition model deployed using TensorFlow Lite for real-time on device inference. The model classifies user emotions into categories such as Happy, Sad, Neutral, Stressed, and Bored. Based on detected emotions, the system generates personalized wellness recommendations and adaptive UI responses. The optimized TFLite model size is approximately 363 KB, making it suitable for mobile deployment. Based on mood input, available time, and user interaction patterns, the recommendation engine generates wellness oriented suggestions such as journaling, mindfulness exercises, relaxation activities, music engagement, and creative tasks. The proposed framework integrates emotion interaction, recommendation processing, and adaptive mobile application design within a unified platform. The proposed work demonstrates the practical implementation of an emotion aware wellness assistant aimed at improving personalized digital interaction and emotional well being.