Abstract: Aiming at the problems of relying on manual attendants to provide service, low service efficiency and poor user experience in railway customer service center, this paper designed railway intelligent customer service system. The system has functions such as natural language interactive understanding and intelligent knowledge search, which can provide anthropomorphic human-machine communication based on dialogue for passengers, effectively improve the artificial replacement rate of IVR in railway customer service center, and can quickly and accurately provide self-service consultation and inquiry service for passengers and cargo owners. Practice has proved that the designed system can improve the manual replacement rate by 20% compared with the traditional IVR mode, which can effectively relieve the pressure of manual service, implement the steady improvement of the quality and efficiency of railway customer service, and promote the high-quality development of railway customer service center.