Abstract: Agricultural non-point source pollution not only causes deterioration of the surface water environment but also seriously restricts the green and sustainable development of agriculture. In recent years, relevant monitoring plans and management systems with increasing emphasis on non-point source pollution control have gradually been established in China. However, there are still some problems with the low effectiveness of non-point source pollution control. We summarized the status of monitoring targets, spatiotemporal resolution, source identification methods, and management measures for non-point pollution monitoring at different scales. This pointed out the problems of discontinuous monitoring of non-point source pollution, insufficient spatiotemporal refinement and estimation accuracy of non-point source pollution load, lack of dynamic identification of pollution sources, and insufficient matching between pollution sources and management measures, which directly or indirectly lead to an imperfect monitoring system, insufficient intelligence, low level of refinement, and low efficiency of pollution control for non-point source pollution. With the development of informatization and digitization, the Internet of Things platform has become a key lever for intelligent supervision by integrating sky and ground monitoring equipment and algorithms and providing technical support for finely monitoring non-point source pollution. The construction of an intelligent observation system for Internet of Things for agricultural non-point source pollutant to improve the observation accuracy of key parameters, using the “non-point source model + fingerprint tracing” technology system for identifying and tracking multiple sources of non-point source pollutants sources area, and considering the effectiveness evaluation and scheme optimization of multi-source and multi governance models to solve the problem of precise and differentiated governance was proposed. Finally, the systematic, integrated, and intelligent level of non-point source control will be improved, and connect the key breakthrough path from “monitoring” to “governance”. The application and demonstration of the Internet of Things intelligent monitoring platform were carried out in the Taihu Basin, which has both plain and hilly complex terrain, a dense population, and active agricultural activities. This typical case has integrated monitoring instruments, a non-point source pollution simulation model, pollution source identification, and other parts, providing an important reference for the intelligent supervision of agricultural non-point source pollution.