The paper introduces an intelligent product search system that combines the CLIP model with large language models (LLMs) to efficiently handle and enhance multimodal query processing based on both text and images. The research proposes four main approaches: (1) applying fuzzy clustering techniques to optimize the search space over large-scale datasets, thereby accelerating retrieval speed; (2) integrating a fine-tuned LLM capable of understanding user query context and automatically routing it to appropriate query-processing modules; (3) mapping both images and text into a shared embedding space using the CLIP model to improve retrieval and matching accuracy; (4) building a recommender system based on fuzzy logic and fuzzy inference system. Experimental results demonstrate that the proposed methods significantly improve response time, query precision, and adaptability to complex real-world queries.