This research focuses on the multimodal question and answer system by enhancing the images and forms in different files. The system has two parts: processing system and retrieval system. Users can upload files that they want to use in the future, and the processing system would analyze and store the information in files. Using text similarity calculation makes sure the relevance of answers to the question in retrieval system. The system mainly uses the TF-IDF method, Paddle OCR, and its built-in feature pp-structure, which is useful on forms recognition. In this paper, it mainly elaborates on the design ideas of the proposed system’s processing flow. In the latter part of the paper, it will briefly introduce the testing of the system and the test results, using different types of files to test the processing system and different questions to test the retrieval system. The system is working well after testing and can be further optimized subsequently. The project cooperated with AIA Life Insurance Co. LTD. (“AIA” for short). Test data and development environment of the system was also from the inside enterprise. Sixteen typical documents were used to test the system, and the result shows this system has a high accuracy in a particular scope on professional information searching.