Large language models (LLMs), like ChatGPT, have been studied and used in a variety of domains, including information retrieval, creative writing, reasoning, code creation, and translation. LLMs exhibit a profound understanding of natural human language. In this study, we introduce an autonomous Geographic Information System (GIS) empowered by LLMs as its core reasoning engine. This system leverages automatic geographic data acquisition, analysis, and visualization to address spatial challenges, harnessing LLMs’ versatile capabilities in natural language processing, reasoning, and programming. Through three case studies, the authors have constructed a system that utilizes the GPT-4o API in a Python context, showcasing the possibilities of autonomous GIS without human interaction. This technology significantly reduces manual processing time by consistently delivering accurate results across all case studies, encompassing aggregated data, charts, and maps. This system demonstrates considerable promise in integrating LLM and GIS to streamline spatial analysis, enhancing speed and accessibility for a broader audience. Despite being in its early developmental phase and lacking essential modules like logging and code testing, the system showcases potential for advancing spatial analytical capabilities.