Hangzhou Bank：Process Middleware Drives High-Speed Business Processes

About Hangzhou Bank

Founded in September 1996, Hangzhou Bank is headquartered in Hangzhou and is among the Global 500 Banking Brands, Fortune China 500, and Forbes Global 2000. Currently, the bank has over 200 branches, covering developed economic zones such as the Yangtze River Delta, Pearl River Delta, and Bohai Bay. On October 27, 2016, the bank successfully listed its A-shares on the Shanghai Stock Exchange, with the stock code 600926.

Project Background

In the early stages of project implementation, Hangzhou Bank faced the following challenges:

1. Multiple in-house systems operated independently. Each in-house system was like a separate product line, with its own independent and complete product, development, and testing teams. Each system had complex but generic processes and integration scenarios, which were implemented separately by each team.

2. Lack of a flexible, efficient, and user-friendly process engine. Adjustments were cumbersome, and the functions did not fully meet the requirements. Some in-house systems used hard-coded solutions for process flow scenarios. Every adjustment to rules and flows meant a new development iteration cycle, which was inefficient and costly. Additionally, some in-house systems were based on the Activiti open-source BPM framework, and every adjustment still required developer intervention. Due to the limitations of the underlying framework, some detailed functionalities could not be fully met (e.g., conflict of interest principles, arbitrary rejection mechanisms, etc.).

3. Integration with heterogeneous systems required extensive secondary development. In the business processes of various in-house systems, there were many scenarios involving data integration with third-party systems (e.g., credit systems needed to integrate with multiple credit data sources). Currently, each integration point was implemented as a standalone feature through hard coding, resulting in a high workload and maintenance costs.

Project Objectives

The main goals of this project are:

1. To provide general process services to dozens of in-house systems through a robust and mature technical architecture.

2. To integrate the process engine as a pure backend service seamlessly with in-house systems.

3. To embed business processes into various banking operations easily through powerful and flexible process integration capabilities, achieving real-time data synchronization and automated process management.

4. To provide a modeling engine that is efficient, user-friendly, and maintainable by business personnel, reducing the cost of process development and maintenance.

5. To have a unified process monitoring and analysis tool.

6. To offer a professional process product that supports various process models and case management.

Application Scenarios

1. The process middleware seamlessly integrates with dozens of in-house systems as a pure backend process engine service. End users continue to use the front-end interfaces of the in-house systems, maintaining their usage habits without any changes in the interaction experience, ensuring a smooth rollout.

2. The process middleware serves as a unified entry point and data source for multiple systems, enabling unified process operation, monitoring, and analysis.

3. Deep application of the AlphaBot process robot, for example, in the in-house credit system of Hangzhou Bank, multiple credit products (such as village-wide credit, cloud mortgage loans, cloud microloans, tax loans, etc.) have corresponding loan approval processes. By configuring robots for tasks such as email, OCR scanning, and SMS, repetitive work in the process is automated, enhancing the automation capability.

4. Perfectly solving the "conflict of interest" principle and arbitrary rejection mechanisms. For instance, the process engine can automatically generate a list of "conflict of interest" users based on the personnel relationship model of the process initiator (e.g., relatives), ensuring the fairness and efficiency of the process.

5. The process approval phase provides two types of rejection mechanisms: "step-by-step approval" and "submit to the person who rejected," offering both strict rules and flexibility to cover a wider range of complex business scenarios.

6. An intelligent and intuitive "pre-run" view that combines "pre-run" data with actual historical process records to form a "complete process" view, providing a comprehensive overview of past, present, and future process data, helping handlers identify and mitigate risks.

7. A flexible yet rigorous hierarchical control and permission system, allowing different levels of administrators to have varying granularities of design permissions for process templates. Administrators from different branches and regions can maintain the process templates for their respective branches, with data being isolated.

Comprehensive performance monitoring and log tracking, providing 360-degree, no-dead-angle monitoring and analysis. This allows the client's operations team to manage and allocate resources uniformly, predict potential performance bottlenecks, and provide data support for continuous system optimization and iteration.