Mental disorders affect 1 in 7 people globally. Negative emotions are linked to mental disorders, while positive emotions build resilience and improve quality of life. Emotion regulation (ER) manages negative emotions (e.g., fear, sadness, and anger) and reinforces positive emotions (e.g., happiness) for improved mental well-being. However, there is limited knowledge on how to implement personalized, real-time, AI-driven ER interventions using persuasive technology tailored to an individual’s evolving emotional state while motivating behaviour change. To address this gap, we developed Recilify, an AI-driven and emotion-adaptive persuasive mobile health (mHealth) app that helps users to regulate emotions and improve mental well-being. The app was systematically developed following an evidence-based and user-centered process. First, we conduct a comprehensive literature review to identify persuasive strategies (PS), as well as ER strategies and how they can be implemented in a system. Second, we conduct an empirical study involving 660 participants to identify PS that support ER. Third, we develop a novel AI model to predict users’ emotional states in real time from journal or diary entries. Fourth, we develop an interactive prototype of the app and evaluate it with 20 target users. Our evaluation findings show that Recilify is significantly persuasive, motivating, and usable. Fifth, we develop the app which integrates the AI model, PS, and ER strategies, and then privately deploy it on both Google Play and Apple App Stores for real-world testing. Finally, we build on our empirical insights to provide practical recommendations for designing persuasive mHealth apps to improve mental well-being.