Abstract: With the rapid development of remote sensing-related science and technology, remote sensing information products are characterized by massive volume and great diversity. Despite possessing basic data services and information processing capabilities, the existing platforms still face limitations in automated knowledge extraction and specialized question-answering services. Consequently, they fail to intelligently parse user needs and dynamically generate adaptive analytical workflows. To address these limitations, this study proposed and preliminarily developed an intelligent analysis framework for remote sensing information products based on large language models (LLMs), with land cover mapping serving as an experimental scenario for validation. Using natural language parsing and prompt engineering techniques, the framework can achieve autonomous spatiotemporal mapping, core variable construction, and indicator generation. Moreover, it enables efficient data management and computation using the Open Data Cube (ODC). Experimental results show that this intelligent system outperformed the solely GPT-based approach in comprehensive capability for remote sensing data analysis, significantly enhancing the accuracy and specialization of knowledge services. It has the potential to advance intelligent knowledge services for remote sensing information products.