The number of smartwatches is constantly growing and emerging in many different fields; fitness is one of the areas. To measure heart rate and other fitness-related metrics, smartwatches use Artificial Intelligence (AI) to generate real-time representations of the different metrics and provide activity notifications to encourage exercise. Feedback is essential for any learning process, and today, there is limited research on how to enable user interaction with smartwatches. This study is a design science research study focusing on how activity notifications can be designed to enable user interaction through a smartwatch with the help of persuasive design. The study resulted in seven guidelines (Minimize steps, Routine, Social deviance, Timing, Encouragement, Cognitive load, and Trigger) that were tested as a prototype through a Wizard of Oz test with smartwatch users. This study concludes the importance of not creating unhealthy behaviors and the importance of notifications that show at a suitable time, making it convenient for the user to interact. The results from this study will be valuable for developers of AI systems and smartwatches to increase a healthy population through more desirable and more tuned smartwatches.