Introduction

Protein degraders are attracting attention as a novel modality, and drug discovery based on this concept has been actively pursued by start-ups, pharmaceutical companies, and academic institutions. Many compounds have already advanced to the clinical-trial stage, and the first PROTAC was finally approved by the FDA in May 2026. However, the traditional medicinal chemistry approaches and empirical rules commonly used in conventional small-molecule drug discovery are not always effective, making it challenging to improve activity while maintaining acceptable physicochemical and DMPK profiles. To address this issue, there are attempts to use AI drug discovery technologies, which have been advancing rapidly in recent years. Deep learning-based protein-structure prediction models, as well as generative AI technologies that allow AI to design small molecules or proteins suitable for target proteins, have begun to be utilized in practical drug discovery research. The recent announcements that AI-designed molecules have advanced to Phase II clinical trials have drawn broad attention in this field. This article provides a concise overview of protein degraders, describes bottlenecks in their research and development, and highlights representative applications of AI-driven drug discovery technologies aimed at addressing these challenges.