Software development is undergoing a dramatic transformation with the rise of generative AI techniques such as vibe coding. However, in the field of embedded systems, many teams still struggle to contain the rising costs of ensuring software quality when adopting AI-driven workflows. The root cause lies in two critical gaps: the misconception that “simply adopting generative AI will deliver results,” and the persistent challenge of securing skilled engineering resources. In this session, a veteran engineer with years of hands-on project experience will share three essential points for maintaining quality in AI-assisted development. We will explore how to properly balance “tool adoption” (such as AI) with “human resources,” and present practical methods to leverage the benefits of vibe coding in embedded development. Specific examples will include techniques for automating target-devices testing and verifying system behavior from execution logs.