The thesis explores incorporating emerging computing technologies in architectural assembly and design processes, with a focus on developing an AI system for recognizing self-similar components and generating step-by-step assembly instructions. The proposed methodology includes the use of object recognition algorithms, computer vision, digital twins, and mixed reality. The thesis consists of a series of investigations, such as an initial prototypical study using Lego blocks, synthetic dataset generation, secondary case studies with 3D-printed nodes, real-time video streaming and processing, and assembly instruction generation. The AI-Assembly system can respond to real-time user changes, error detection, assembly status tracking, and structure analysis to support human creative agency during the design iteration process. The thesis aims to facilitate Human-AI co-creation in assembling complex structures, enhancing efficiency and flexibility in architectural processes. Furthermore, this research contributes to the growing field of architecture by demonstrating the potential of emerging technologies and integrating physical modeling processes.