Cell identification and sorting have been hot topics recently. However, most conventional approaches can only predict the category of a single target, and lack the ability to perform multitarget tasks to provide coordinate information of the targets. This limits the development of high-throughput cell screening technologies. Fortunately, artificial intelligence (AI) systems based on deep-learning algorithms provide the possibility to extract hidden features of cells from original image information. Here, we demonstrate an AI-assisted multitarget processing system for cell identification and sorting. With this system, each target cell can be swiftly and accurately identified in a mixture by extracting cell morphological features, whereafter accurate cell sorting is achieved through noninvasive manipulation by optical tweezers. The AI-assisted model shows promise in guiding the precise manipulation and intelligent detection of high-flux cells, thereby realizing semi-automatic cell research.