CNC machining of 3D files has an automatization issue, which is still time-consuming with significant use of manual work and CAD experience. This paper addresses the issue of minimizing human intervention in the production of 3D models for use in CNC milling operations. The models are traditionally created manually in a CAD environment, which is time-consuming and needs special skills. As a reaction, the proposed project includes the design and development of an automated CNC milling machine with an artificial intelligence feature, which can recognize geometric figures and create IPT files directly based on 2D images. The offered system incorporates a homemade CNC milling machine with a 3-axis GRBL board and NEMA-23 motors, as well as a segmentation model based on YOLOv8, trained on a set of 338 labeled photos. The precision of this model was 95.21% in identifying and isolating simple geometric objects, including rectangles, circles, and triangles. The resulting segmented outputs are fed into the dimension parameters, which produce automation of 3D models in Autodesk Inventor. Real-world images were used to confirm the experimental results, which showed that the system could handle quality fluctuations in the input images, saving a tremendous amount of time in the modeling process without compromising accuracy in dimensions. The results hold promise for a promising way ahead, although some challenges remain, including sensitivity to image noise and shape overlap. The given strategy helps to develop intelligent manufacturing, as it provides scalable use in industrial automation and instruction in engineering education regarding Industry 4.0.