Everbright Bank's Yang Bingbing: Focus on four areas—computing power, algorithms, data, and functionality—and ensure every investment is used effectively.

March 31 financial front-line news: Everbright Bank held its 2025 annual performance briefing today. Deputy President Yang Bingbing said that over the past year, Everbright Bank concentrated financial technology resources on key areas to enable the transformation of business across the whole industry with technology. In 2025, full-year technology investment accounted for more than 5% of operating revenue. This level of investment fully demonstrates the bank’s emphasis on the development of financial technology. Rather than spreading resources blindly, the bank precisely focused on four core directions—computing power, algorithms, data, and functionalities—so that every dollar is put to its best use, providing strong support and a solid guarantee for high-quality business development across the entire industry.

First, in terms of computing power, the bank follows the principle of “phased implementation, priority to what’s urgently needed,” and advances both the construction of general-purpose computing power and intelligent computing power in parallel. They have different priorities and complement each other—general-purpose computing power is geared toward “all-around compatibility,” meeting the basic computing power needs of businesses across the industry; intelligent computing power pursues “specialized and refined,” focusing on core scenarios such as AI applications. 2025 is an important milestone for the bank’s computing power development. The growth rate of intelligent computing power has far outpaced general-purpose computing power. Compared with the previous year, the increase is close to 150%, laying a solid foundation for further deepening AI applications and improving the efficiency of technology enablement.

Second, in terms of algorithms, the bank focuses on building a “9×10” intelligent assistant matrix. “9” refers to covering nine types of job positions, and “10” refers to having ten general capabilities. The core purpose is to empower employees to improve work efficiency. At present, this intelligent assistant matrix has covered more than 15k employees, accounting for nearly 35% of all employees. For example, the assistant for corporate banking relationship managers has cumulatively generated more than 37,000 types of intelligent reports, covering the entire process such as pre-marketing for credit granting business, pre-loan investigations, and post-loan management—substantially reducing the workload of relationship managers. Meanwhile, the bank continuously strengthens the professional talent team. Across the whole bank, nearly 100 people have obtained model master qualification certifications, further enhancing the professional capability of model construction. In addition, the bank promotes large-scale application of “AI+RPA.” In 2025, it added 610 intelligent automation scenarios, saving more than 1,100 person-years of work hours, significantly improving operational efficiency. The bank treats the sustained growth in Token consumption scale as a core metric for measuring the depth of AI applications, driving intelligent applications to upgrade from “usable” to “good to use, widely used.”

Third, in terms of data, the deputy president believes that in the AI era, it is not only people using data; more importantly, AI is using data. Over time, high-quality datasets are the key to enabling AI to “understand, read, and interpret well” and to maximize its value. In 2025, the bank systematically advanced the building of high-quality datasets. It fully integrates various types of structured and unstructured data, such as business data, external data, and metadata, and promotes upgrading data from “accurate and usable” to “understandable, inferable, and executable,” so that data truly becomes a core asset driving business development. Also, when it comes to data applications for front-line branches, the bank independently developed an interactive intelligent Q&A tool called “Ask Data,” lowering the barrier to using data. “Ask Data” covers more than 6,000 indicators. Currently, the number of active users is nearly 5,000. It enables and improves efficiency for front-line business scenarios such as operational decision-making, corporate due diligence, and customer acquisition and marketing.

Fourth, in terms of functionalities, the bank focuses on upgrading core business systems and empowering key projects, using a dual-engine drive to enhance technology support capabilities. On one hand, it comprehensively promotes the upgrading and replacement of important business systems such as a new generation of core business system and a new generation of integrated teller systems, consolidating and strengthening core competitiveness in key business areas; on the other hand, it works with full effort to support special teams advancing inclusive finance, supply chain finance, and retail credit, continuously strengthening the quality and effectiveness across the entire process of online customer acquisition, activation, and retention. For example, the bank redesigned the online business process for inclusive corporate banking, improving the average approval time by more than 80% versus the traditional model, greatly enhancing customer experience; it delivered more than 30 supply chain benchmark chain projects. At the same time, it improves the online level of traditional retail credit products and builds a post-loan collection system for retail credit, enabling end-to-end online operations such as automatic generation of collection tasks, tracking, and recording, further enhancing business efficiency and risk management and control capabilities.

Yang Bingbing said that in 2026, the bank will continue to focus on the four dimensions of computing power, algorithms, data, and functionalities, deepen the integration of technology and business, and continuously enhance its core competitiveness.

A wealth of information and precise analysis—find it all on the Sina Finance APP

Responsible editor: Zhang Wen