Integrating RAG with Face Recognition for Personalized Guest Services for Hospitality Industries

The hospitality industry has increasingly adopted advanced technologies to enhance guest experiences, yet many guests remain dissatisfied with on-site services despite effective pre-stay recommendation systems. Our proposed AIReceiptionist system integrates Face Recognition (FR) and Retrieval-Augmented Generation (RAG) modules to provide personalized and efficient guest services. The FR module identifies guests as they enter the public lobby, and the RAG module retrieves detailed guest information from the relevant database, generating prompts to notify staff. We present the system implementation of our AIReceiptionist by leveraging existing deep learning tools, including MTCNN for face detection and InceptionResnetV1 for face recognition, combined with OpenAIEmbeddings for data embedding and GPT-4o for language model responses. Using synthetic data tailored to hospitality operations, we validated AIReceiptionist's performance and compared it with the existing GPT-4o model. The chatbot responses show that our AIReceiptionist enhanced by FR-RAG module significantly outperformed the generic GPT-4o in delivering personalized guest services, demonstrating its feasibility in enhancing guest satisfaction in hospitality environments.