This blog post series introduces Jetpacker, an open-source Android demo app showcasing how to build GenAI features. The series will guide developers through choices regarding execution location, complexity, and integration of AI models. Jetpacker demonstrates on-device intelligence for features like itinerary summarization, expense tracking from receipts, and voice note transcription. These on-device features offer privacy benefits and work offline, leveraging Gemini Nano and ML Kit. Cloud and hybrid inference are utilized for features requiring greater world knowledge or complex tasks, such as a museum assistant or review drafting. Hybrid inference dynamically selects between on-device and cloud models to optimize performance and cost. Android system integration allows app capabilities to be accessed by the operating system via the AppFunctions API. Agentic workflows, which enable models to act with autonomy, are also explored, with a booking assistant example running in the cloud. The series offers detailed implementation steps and architectural insights for creating intelligent Android applications.