Shaping Collaboration Culture and Driving Agile Transformation with Minsheng Bank’s Digital Platform

1. Overcoming Platform Construction Challenges

The platform had to integrate more than 20 core application modules and nearly 300 downstream systems, testing the bank’s ability to coordinate large‑scale, tightly coupled services.

Technical heterogeneity across branches—multiple programming languages, databases, and frameworks—required strict cross‑institution governance.

Enterprise‑level shared capabilities were immature; dozens of middle‑office services (office, data, grading, AI, etc.) were not yet systematized, limiting scenario‑driven collaboration.

Rapid parallel rollout demanded smooth, multi‑dimensional migration with strict disaster‑recovery, data consistency, and transaction integrity during gray‑release phases.

2. Building a Three‑Dimensional Architecture

To address these issues, Minsheng Bank constructed an “application‑deployment‑security” three‑dimensional architecture that mutually supports and unifies the platform.

Application Architecture – 6 layers, 8 domains : The bank adopted a six‑layer, eight‑domain enterprise model built on cloud‑native technologies. The layers are:

Infrastructure layer : IaaS, foundational PaaS, APaaS, integrated DevOps, security services.

Data layer : Lake‑warehouse integration, real‑time data, data acquisition and exchange.

Product layer : Feishu‑based collaborative products adapted to bank standards.

Middle‑office layer : Office, collaboration, decision‑making, AI, and data capabilities.

Scenario layer : Over 300 self‑built middle‑office integration applications via an open ecosystem.

Channel layer : Unified “i Minsheng” entry covering PC, mobile, Pad, and smart‑screen.

Deployment Architecture – Multi‑cloud, two‑site, three‑center : The solution uses “head‑office cloud, branch cloud, ecosystem cloud” across two data‑center locations (Ma Po, Peng Bo Shi) and three centers, providing separate zones for office, collaboration, and ecosystem workloads while satisfying domestic‑innovation (信创) requirements.

Network design employs SDN, distributed gateways, and SD‑WAN for intelligent traffic scheduling and multi‑link load balancing, delivering high availability, scalability, and elasticity.

Security Architecture – End‑to‑end deep defense : Based on a three‑layer security model (digital infrastructure protection, digital operation security, digital open security), the platform implements cryptographic security, network segmentation, container‑level interception, sandboxing, and full‑link data encryption. It isolates office, production, and internet networks, and adds DLP‑based email scanning and mobile data‑copy protection. Multiple red‑team/blue‑team exercises have shown no security incidents since launch.

3. Seven Innovation Breakthroughs

The platform achieved breakthroughs in seven dimensions:

Organizational & Mechanism Innovation : A leadership group chaired by the chairman and a PMO spanning 27 departments established agile governance (daily stand‑ups, monthly senior meetings, weekly reports) and matrix‑based performance assessment.

Infrastructure & Resource Innovation : Full‑stack domestic‑technology stack (hardware, OS, middleware, DB) enabled dual‑center, cross‑cloud deployment with high‑speed dedicated links, global video acceleration, and micro‑service design (2,000+ services on 200+ servers) supporting elastic scaling.

Technology & Implementation Innovation : A dual‑engine approach (“Volcano Engine + Minsheng Financial Cloud”) resolved Feishu and bank core differences, using data hooks and subscription listeners, CI/CD pipelines, and a phased rollout (head‑office pilot → branch rollout) with extensive training.

Standard & Enablement Innovation : Established standards for application, cloud‑native, open‑service, development security, and operations, reducing duplicate modules, simplifying heterogeneous integration, and delivering 30+ SaaS/service/terminal components via a “review‑test‑release” workflow.

Terminal & Security Innovation : End‑to‑end protection across terminals, links, and services (container interception, virtual paths, sandboxing, HTTPS/WSS/DTLS, identity authentication, security gateways, vulnerability scanning) meeting financial‑grade data protection and regulatory compliance.

AI Scenario Innovation : Integrated AI middle‑office (RAG, NLP, OCR, large models) to deliver four core scenarios—intelligent knowledge management (12,318 dictionary entries, 260k documents), intelligent operations, AI robot companion (LLM‑driven Q&A via event subscription), and AI‑assisted document proofreading.

Cultural & Management Innovation : Redefined collaboration culture with an All‑in‑One workbench, achieving a 50 % boost in collaboration efficiency, 98 % tool adoption, and a unified “platform + ecosystem” model.

4. Demonstrated Outcomes

Following an “explore‑extend‑deepen” roadmap, the platform progressed from design and pilot to full‑scale rollout, scenario enablement, AI upgrades, and value extension.

Key results include:

All‑in‑One model enabling rapid business iteration.

Over 80,000 users, saving ¥16 million in development costs and ¥5.5 million annually in operations.

Support for 100,000 TPS, 30 % faster approval processes, and 98 % digital tool usage.

Group‑level distributed architecture and solution serving cross‑region, cross‑department high‑concurrency access.

5. Future Direction

The platform will continue to deepen AI integration (large models, RAG), expand scenario coverage, and build a co‑creation ecosystem, positioning the platform as the core engine for agile organization and intelligent banking.