Group 1: Core Insights - Kunshan Rural Commercial Bank is advancing its digital transformation by deploying multiple domestic open-source large models to create an intelligent office matrix, providing strong technical support for its initiatives [1] - As of May 2025, the bank has delivered over 10,000 efficient services to more than 500 individuals through intelligent office scenarios [1] Group 2: Technology Planning and Implementation - The bank has established a technology planning framework that emphasizes a "scene-driven, small-step fast-run" approach, prioritizing high-frequency and essential scenarios for development, successfully creating and testing five application scenarios including knowledge assistance and document translation [3] - The introduction of several domestic open-source large models has led to the construction of a multi-model collaborative intelligent office matrix, expanding the intelligent computing server cluster to meet current and future diverse application needs [3] Group 3: Departmental Collaboration - A virtual task force has been established to break down departmental barriers and promote deep integration of AI technology with business processes, led by the data management department [5] - A dynamic service feedback loop has been created, allowing business personnel to report issues encountered while using large model scenarios, facilitating timely feedback and iterative improvements to ensure business needs are met [5] Group 4: Data Security and Risk Management - The bank implements a tiered management system for its AI knowledge base, categorizing content into four levels: public, internal, sensitive, and core, to prevent unauthorized access and ensure appropriate management of sensitive information [7] - Strict network access controls are in place, including firewall rules and differentiated authorization strategies to mitigate external attacks and internal data leakage risks [7] - A comprehensive auditing mechanism is established to regularly review and assess the outputs of generative large models, ensuring compliance with ethical standards and value orientation [7]