Organic reaction development guided by large language models: the next stage of AI-driven chemistry

Abstract Digitalization-driven transformative organic synthesis has become an established research style in organic synthesis over the past decade, including reaction optimization, molecular design, and materials informatics (MI). Despite methodological advances, fundamental challenges remain in molecular representation, data curation, and hypothesis generation. This perspective review reexamines these issues from the standpoint of experimental organic synthesis and discusses the emerging role of large language models (LLMs). By integrating contextual reasoning, experimental records, and molecular networks, LLMs are positioned not as predictive engines but as cognitive layers that shorten the cycle time of hypothesis–experiment–revision loops.