The main goal of this study is to look into how deep learning methods can be used in building design. N model takes building outlines as input and instantly makes floor plans. Before examining how useful ArchiGAN and the pix2pix model could be for building design, we will briefly explain their different methods, problems, and successes in the field. We build a stable training process using TensorFlow and Keras on a dataset with more than 12,000 samples. In addition to a generator with improved custom loss functions, we offer an Adam optimizer and a PatchGAN discriminator based on U-Nets. We wanted to create creative floor plans that fit the given size by making small changes and trying new things repeatedly. This study aims to bring attention to architectural design that uses AI and machine learning by using conditional generative adversarial networks (C-GANs). This will give architects new tools for making floor plans quickly and creatively.