AI+ Tool and Solution Catalogue

SensePedia is an advanced question-and-answer system that leverages documents and knowledge bases to deliver rapid query responses, document analysis, and reference cross-checking. It harnesses the power of Large Language Model (LLM) and Retrieval-Augmented Generation (RAG) technologies to engage in comprehensive conversations across various subjects. SensePedia supports multiple languages and inputs, including Cantonese and both traditional and simplified Chinese. It offers functionalities such as continuous conversations, prompt customization, web search, and multimodal capabilities. The system is adept at handling document-based chats, allowing users to upload documents, receive abstracts, and ask questions to obtain detailed answers. It also facilitates the creation of custom knowledge bases and provides structured data intelligence modules for embedding and retrieving knowledge efficiently. SensePedia ensures user-friendly interactions with role assignments, tone tuning, and sentiment management, all while maintaining data privacy and eliminating hallucinations in responses.

SensePedia serves multiple use cases across various industries by leveraging its advanced capabilities in natural language processing and document management. In one scenario, it functions as an AI assistant in the insurance sector, where it utilizes a knowledge center developed from customer files to provide precise answers and references, significantly reducing the need for professional labor. In another use case, the application is employed by a telecommunications company to enhance their call center operations. By fine-tuning the large language model, the system improves the understanding of modalities, particularly in languages like Cantonese, thereby increasing the direct usability rate of the generative AI in responding to customer queries. Additionally, the application aids in document preparation for governmental tasks, where it automates the drafting of documents while ensuring compliance with current standards and maintaining confidentiality. This is accomplished by constructing a comprehensive knowledge database trained on relevant documents, equipped with features for compliance checking and document drafting, thus boosting productivity and efficiency in handling heavy workloads of text processing.