This paper discusses the evolution of Electronic Design Automation (EDA) from localized assistance to autonomous agentic execution, highlighting Large Language Models as a unified intelligent interface for front-end design.
It reviews representative advances in circuit and testbench generation from shared specifications, as well as design quality improvement in high-level synthesis workflows. The authors also outline key challenges and limitations of integrating LLMs into EDA and propose future opportunities for leveraging agentic AI technologies.