Computed tomography (CT) scanning can now produce detailed three-dimensional views of assembled parts, including internal components that cannot be inspected visually. Despite these advances, distinguishing between acceptable and defective assemblies remains slow and error-prone, even with modern imaging software. This project responds to that challenge by exploring how artificial intelligence can support inspection by comparing CT scan data with a computer-aided design digital twin. The work focuses on developing and training an artificial intelligence model to identify missing components, burrs, and piece-part defects, and flag areas requiring human review. Project teams use CT scanners to generate real scan data, build their own assemblies, and introduce controlled defects to test recognition accuracy. The project seeks to enhance inspection efficiency and reliability by matching scanned assemblies with idealized CAD models, while ensuring that final decisions remain in human control.