We introduce an open-ended project-based assignment for introductory programming courses. It positions LLMs as collaborative programmers, while ensuring student agency by requiring feasible hardware validation for the computer programs. The goal is to promote student learning by asking students to complete joint software-hardware tasks that AI agents cannot complete due to the hardware validation requirement. Specifically, the assignment asks students to pick a personal project (from an instructor provided list) and build it with an AI partner. The key step is validating the final code with a real hardware setup. This moves beyond pure simulation and compels students to ground their code, and their understanding, in real-world interactions. Students keep a reflection journal and submit their AI chats, showing their thought process. Using low-cost hardware ensures students are active participants. They must understand, adapt, and verify the AI's output, not just copy it. This fosters a deeper, more authentic learning experience and shows students how to use AI tools. The assignment is highly adaptable, with many personalized project options to promote engagement, agency, and creativity. The list of open-ended projects, instructional material, and steps for integrating hardware components into each project will be available as an open-access resource.