This thesis explores the landscape of digital health and wellbeing applications and reviews the theoretical foundations of machine learning, with a focus on convolutional neural networks and large language models. It compares state-of-the-art foundation models, GPT-4 and LLaMA-3, in terms of architecture, training strategies, and practical trade-offs. Cybersecurity challenges are examined in relation to health coaching apps that rely on Bluetooth-enabled wearables, with attention to standards, authentication mechanisms, and compliance with privacy frameworks such as GDPR and HIPAA. Building on these analyses, the thesis presents the design of an AI-powered digital health coach. A central feature of the prototype is a lesson and quiz generation tool, which offers a platform for creating personalized educational content to support user engagement and experience.