Abstract

Integrating complex sets of Electronic Design Automation (EDA) tools for seamless information exchange has become a critical challenge for designers and engineers. Recent advancements in Large Language Models (LLMs) have driven significant progress in Natural Language Processing (NLP), enabling improved interaction with EDA tools. In this thesis, we propose SmartEDA, an agent and executor framework for EDA workflows, powered by LLMs and AutoMage. SmartEDA streamlines the design flow from RTL to GDSII by efficiently managing task decomposition, script generation, and task execution.