Minsheng Bank Chief Information Officer Zhang Bin: 261 new AI application scenarios added in 2025

(Source: Beijing Business Today)

Beijing Business Today (Reporter: Meng Fanxia, Zhou Yili) On March 31, 2025, Minsheng Bank held its 2025 annual performance communication meeting. The bank’s Chief Information Officer, Zhang Bin, provided a detailed interpretation of the bank’s related plans and practices in the field of artificial intelligence. He said that since late 2023, when generative AI made breakthrough progress, Minsheng Bank has taken it as a goal to promote its high-quality development and to provide better and improved financial services for the real economy and the public at large. At the same time, he emphasized that the application of generative AI in banking is not merely about simple technical deployment. In 2025, Minsheng Bank’s AI applications were advanced in an orderly manner mainly across four dimensions: strategic leadership, organizational support, capability building, and application deepening.

In terms of strategic leadership, under the overall coordination of the bank’s Head Office Digital Leadership Group, Minsheng Bank has continued to deepen AI governance. It formulated an AI strategy to strengthen top-level design and clarified the basic principles of value orientation, performance co-creation, open collaboration, and safe and controllable operations. With intelligent banking as the target, it explored and promoted a transformation of AI applications from “tool empowerment” to “model remolding.”

Regarding organizational support, Minsheng Bank adopted a combination of external recruitment and internal cultivation to strengthen its AI talent team building. Zhang Bin acknowledged that starting from early 2024, the bank’s technology function hiring has focused on three major areas: AI, security, and architecture. In 2025, it established a standardized AI engineer training and certification system, and also developed a coordinated mechanism between business analysts and intelligent solution architecture engineers, supporting the transformation in which performance is integrated into performance co-creation.

In the area of capability building, Minsheng Bank has continued to optimize foundational capabilities such as computing power, data, knowledge, and models. In 2025, it focused on strengthening Agent engineering capabilities and AI security and risk prevention and control capabilities. As a primary form and carrier of large-model applications, the bank conducted research from multiple perspectives, including management mechanisms, platform capabilities, tool ecosystems, and application paradigms. It built an intelligent agent foundation that is manageable and controllable throughout the entire lifecycle, and that supports efficient internal and external collaboration, providing enterprise-level support for building complex applications in key business scenarios. At the same time, it also held an AI Agent application competition across the whole bank, successfully incubating thousands of AI agents, effectively boosting the enthusiasm for organization-wide AI application and innovation.

In terms of security and risk prevention and control, Minsheng Bank also established an AI application classification and grading framework and built an end-to-end management system covering model training, research and development, and production deployment. It also built an AI quantitative evaluation system, including technical capabilities such as an AI firewall and content safety guardrails.

Zhang Bin said that AI value is ultimately reflected through scenario-based application implementation. In 2025, the newly added AI application scenarios at Minsheng Bank showed a clear growth trend. In fields including marketing, risk control, operations, analysis and decision-making, and office R&D, around high-value scenarios, the bank completed more than 40 key applications. It added 261 more granular scenario applications. Daily average AI service calls exceeded 5 million times, including a year-over-year 16-fold increase in generative AI call volume.

Specifically, in the marketing field, it focused on building retail and corporate marketing assistants, including functions such as persona analysis, product holding analysis, and product recommendations, covering more than 20k customer managers across the whole bank. In the risk control field, it carried out AI enablement and remolding across the full lifecycle of credit business. In the review and approval stage, the adoption rate of intelligently assisted generated content exceeded 84%, saving more than 40k staff-hours annually. The time spent on legal contract review was compressed from the hour level to the minute level. The identification rate of financial analysis risk signals reached more than 50%. The efficiency of generating post-loan reports increased by 20%. In the operations field, it achieved a 95% accuracy rate for entering account opening information and a 90% accuracy rate for intelligently assisted review in loan disbursement, significantly improving service efficiency and customer satisfaction.

In the office field, the AI question-and-answer knowledge assistant had been used by 60k employees, with a monthly active user base of 20k. In the IT R&D field, Minsheng Bank promoted the construction of software engineering 3.0, enabling AI enablement throughout the entire R&D process. By the end of 2025, its AI code generation rate reached 20.68%, and R&D efficiency improved by 10%. Zhang Bin said, “In 2026, the Agent autonomous programming mode has already been put into production, and we expect R&D efficiency to increase by 30%.”

Massive information, precise interpretation—available in the Sina Finance APP