Prinsepia utilis Royle is a characteristic medicinal and edible plant resource in Southwest China. Its seeds are rich in a variety of bioactive components and have broad application prospects. Different extraction processes significantly affect the acquisition of effective ingredients from plants, restricting their precise development and industrial application. In this study, seeds of Prinsepia utilis collected from a high-altitude region (2700 m) in Lijiang, Yunnan Province were used as raw materials to prepare water extract (PuWE) and ethanol extract (PuEE). Combined with chemometric analysis and untargeted metabolomics, we systematically compared the metabolic profiling characteristics of the two extracts. Artificial intelligence (AI) was used to explore potential application scenarios of differential metabolites. The results showed that the constituents of Prinsepia utilis seeds were mainly lipids (35.97%) and protein (14.33%), followed by flavonoids (2.29%), phenolics (0.37%), vitamin E (0.03%), etc. A total of 2260 metabolites were identified from the two extraction methods, of which 241 were differential metabolites. PuWE was mainly enriched in hydrophilic compounds including organic acids, carbohydrates and amino acids, while PuEE was rich in lipophilic constituents such as lipids, phenolics and alkaloids. The two extracts exhibited complementary patterns in polarity, metabolic pathways and biological functions. AI prediction indicated that PuWE and PuEE had distinctly different application potentials in multiple industrial fields including daily chemical products, green pesticides, pharmaceutical intermediates and health care products. This study preliminarily clarified the directional enrichment rules and utilization values of metabolites obtained via aqueous and ethanol extraction, providing experimental evidence for the development and comprehensive exploitation of Prinsepia utilis resources.