Purpose This study analyzes how technical development and organizational adaptation are coordinated during Artificial Intelligence (AI) integration in manufacturing quality control. Design/methodology/approach A nine-month case study followed an AI-based anomaly detection integration at a heavy-duty vehicle manufacturer. Data were collected through interviews, workshops, observations, and sensor data from 1,658 transmission axle tests. Findings The AI model achieved 12–37% better anomaly detection than traditional methods. However, the core finding is that AI cannot be a local add-on. Successful integration requires system-level coordination, which was made possible by a dual Plan-Do-Check-Act (PDCA) framework as parallel iterative learning cycles. This structure synchronized technical development with organizational adaptation. It secured physics-based validation, user-centered interface design, and sustained cross-functional collaboration. Research limitations/implications This study analyzes the development and initial deployment phase during which cross-functional coordination shaped both technical design and organizational readiness. While AI integration literature typically addresses technical development and organizational factors separately, this study demonstrates that they must be managed together during development through explicit coordination mechanisms. Future research should examine how systemic coordination requirements vary across different AI applications and organizational contexts, and should investigate long-term operational sustainability. Practical implications Successful AI integration is dependent on systemic coordination, facilitated by a dual-cycle framework for parallel technical and organizational development. Companies should prioritize coordination mechanisms that bridge the gap between data science and manufacturing engineering. Originality/value This study identifies “systemic coordination” as a distinctive characteristic where technical and organizational changes cannot be sequenced but require parallel management during development. The dual PDCA cycles framework provides a mechanism for managing this coordination. The study also demonstrates design principles for human-AI collaboration that preserve human agency while leveraging AI capabilities.
Leberruyer et al. (Mon,) studied this question.