Abstract

Abstract—Recent advances in artificial intelligence (AI) have significantly transformed website design, prototyping, and deployment. Traditional web design workflows rely heavily on manual coding and repetitive client-designer interactions, making the process resource-intensive and error-prone. Over the past decade, researchers have attempted to automate various stages of this workflow using computer vision, machine learning, and generative modelling. This literature review synthesizes current developments in AI-driven automation for website design, emphasizing the evolution from layout clustering and visual feature extraction to end-to-end code generation. The review categorizes existing research into heuristic, computer vision-based, deep learning-driven, and unsupervised clustering approaches, and outlines their methodologies, datasets, and evaluation metrics. It pays special attention to recent approaches that combine unsupervised layout analysis with automated code-generation pipelines, enabling the discovery of design prototypes from real-world websites and their transformation into functional web templates. By linking these complementary research directions, this review establishes a conceptual bridge between structural pattern discovery and generative website synthesis. It concludes by identifying key challenges such as dataset scarcity, model semantics, and balancing automation with creative flexibility, and discusses emerging trends including multimodal AI frameworks and large language models that are poised to further advance web design automation further.