The number and distribution of electricity customer service targets are large and extensive, and the embedding of intelligent customer service mode effectively improves the service satisfaction of electricity customers. However, with the continuous growth of actual electricity consumption scale and the evolution of diversified and personalized development trends in customer electricity demand, the methods and implementation effects of knowledge recommendation for power customer service are facing challenges of low accuracy and insufficient depth of knowledge recommendation. In response to this issue, the article proposes a knowledge recommendation method for power customer service that introduces inference chain retrieval enhancement technology. Based on Alibaba’s open-source language model, a basic knowledge graph of power customer service is constructed by combining the Q&A records of power customer service. The potential related relationships in the constructed knowledge graph are aggregated to generate virtual connection relationships. Integrating high-order semantic information to generate inference chains to capture the complex network features and knowledge connectivity of the knowledge graph structure, and using a traceable topological mapping mechanism to equivalently reduce the distribution bias of knowledge nodes. After encoding the neighboring features of adjacent levels, the learning process of the recommendation model is adjusted through interval experience replay and delayed reward mechanism to improve the effectiveness of the recommendation results. Compared with existing knowledge recommendation methods, the knowledge recommendation process introduced in the article with the inference chain retrieval enhancement mechanism takes into account the acquisition of high-order information and the elimination of the negative effects of interference information in the recursive propagation process, which can more accurately intervene in the inference process and improve the practicality of recommending knowledge for power professional customer service.