Unlike traditional design, which often starts with user research and then considers technical feasibility, AI Product Design begins by exploring the potential of data to solve a problem. This ensures early validation of the AI's recommendations (even with rough prototypes like Excel-based mocks), where the accuracy and impact of the AI's output become critical design criteria. If the AI isn't ready, the process demands designing fallback solutions and planning for continuous learning and iteration to enhance the AI's capabilities over time. This approach ensures that the eventual user experience is both effective and genuinely improved by AI, rather than complicated by it.