Objective To develop an evaluation system for assessing research innovation competency among undergraduate nursing students and to evaluate its preliminary application effectiveness. Methods An AI-enabled evaluation system for research innovation competency among undergraduate nursing students was developed based on the Kirkpatrick Model through literature review, semi-structured interviews, and two rounds of expert consultation. Using a convenience sampling method, 30 undergraduate nursing students were recruited to participate in an 8-week AI-assisted research innovation training program. The effectiveness of the training was evaluated before and after the intervention using the newly developed evaluation system. Results The final evaluation framework consisted of four first-level indicators, 13 second-level indicators, and 42 third-level indicators. The total score for research innovation competency among undergraduate nursing students increased significantly from the pre-training level (P<0.05). Conclusion The AI-enabled evaluation system for research innovation competency among undergraduate nursing students demonstrates a well-structured framework and comprehensive content. Its preliminary application indicates good feasibility and practicality, providing a useful reference for evaluating the effectiveness of AI-assisted research education in nursing schools.