Process development system for customizing and constructing intelligent agent

Abstract

The invention relates to the technical field of workflow management, and discloses a process development system for customizing and constructing an intelligent agent, which comprises a general data component library, a visual editing module, an automatic vector marking module and a business system integration module, and is characterized in that the general data component library is used for storing reusable process node components; comprising a description component, an approval component and an evaluation component, and the approval and evaluation components can be flexibly added according to requirements; the visual editing module provides a dragging type operation interface and supports dynamic adjustment of form components and process nodes through operations such as dragging, clicking, inputting and selecting; the automatic vector marking module carries out vectorization processing on a text input by a user to generate a text feature vector supporting semantic retrieval and correlation analysis; and the business system integration module realizes synchronization of approval process execution records and business system data. The invention aims to solve the problems of repeated flow node configuration, low reusability and insufficient data integration and sharing capability in the existing flow approval system.

Description

Technical Field

[0001] This invention relates to the field of workflow management technology, and more specifically to a process development system for custom-built intelligent agents. Background Technology

[0002] With the development of internet technology, enterprise information systems are increasingly widely used in daily office work, becoming indispensable tools for efficient enterprise operation. Enterprise business systems include financial systems, human resource systems, customer relationship management systems, etc., covering all business areas of the enterprise. Process approval, as a core component of enterprise management, plays a crucial role in coordinating various tasks, optimizing resource allocation, and improving work efficiency. However, in practical application scenarios, existing process approval systems face numerous challenges.

[0003] Traditional workflow approval systems mostly employ pre-set fixed workflow templates and approval nodes. While this model can meet some standardized business processes to a certain extent, its lack of flexibility makes it difficult to adapt to the diverse and personalized business needs of enterprises. For example, when an enterprise needs to customize approval processes for specific projects or departments, traditional systems often cannot make rapid adjustments. This makes it difficult for enterprises to respond quickly and optimize processes when facing complex and ever-changing business scenarios. Customizing and modifying workflows in traditional systems requires the intervention of professional developers to rewrite and debug the system code. This is not only time-consuming and labor-intensive but also prone to introducing new errors, seriously affecting the enterprise's operational efficiency and responsiveness.

[0004] Furthermore, in terms of data integration and sharing, traditional workflow approval systems suffer from inconsistent data formats and standards across various business systems, hindering data flow and sharing. This not only impacts cross-system collaboration but also makes it difficult to ensure data management and process consistency. For example, the inability to synchronize data between the financial and sales systems in real time affects the efficiency and accuracy of corporate decision-making.

[0005] The core issue mentioned above lies in the insufficient reusability of processes. In traditional systems, each change in business requirements necessitates the redevelopment and reconfiguration of processes, lacking reusable process components and flexible customization capabilities. This model not only leads to low development efficiency but also makes it difficult for enterprises to quickly adapt to market changes and internal business adjustments. Therefore, a process development system capable of solving these problems is urgently needed. Summary of the Invention

[0006] The present invention aims to provide a customizable intelligent agent construction process development system to solve the problems of insufficient data integration and sharing capabilities of various business systems, repetitive and low reusability of process node configurations, difficulty in quickly responding to the diverse and personalized business needs of enterprises, resulting in low development efficiency and high costs in the existing technology.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: A custom intelligent agent construction process development system, comprising: The system includes a general data component library, a visual editing module, an automatic vector labeling module, and a business system integration module. The general data component library is used to store reusable process node components; The process node component includes a description component, an approval component, and an evaluation component. The description component is used to describe the event; the approval component is used for process approval; and the evaluation component is used to evaluate the event outcome. The approval component and the evaluation component are added to the process node component as needed. The visual editing module provides a drag-and-drop interface, which includes dragging, clicking, inputting, and selecting. Form components and process nodes can be dynamically adjusted through drag-and-drop operations. The automatic vector tagging module is used to vectorize the text content input by the user, generating text feature vectors that support semantic retrieval and association analysis; The business system integration module is used to synchronize the generated approval process execution records with the business system data.

[0008] The principles and advantages of this solution are as follows: In practical applications, by developing process nodes into general-purpose data components, the process achieves high flexibility and customizability, enabling adjustments and optimizations based on actual needs to adapt to constantly changing business environments. These general-purpose data components allow for the editing of both manual and automated business processes, improving the relevance and overall efficiency of process approvals. This solves the problem of limited data integration and information sharing capabilities between various business systems in traditional solutions, ensuring consistency in data management and processes. By using process nodes as general-purpose data components, the configuration process for process approvals is simplified, eliminating the need for professional personnel to modify business process code and significantly reducing implementation and maintenance costs. All records have automatic vector labeling capabilities, which enhances the intelligence level of the approval process and improves the efficiency and accuracy of data querying and retrieval by the AI ​​agent. Through business system integration and visual editing, the solution meets the specific business process needs of different enterprises, overcoming the lack of personalized settings and adjustments in existing technologies.

[0009] Preferably, as an improvement, the general data component library further includes a time component, a priority component, and an attachment component. The time component is used to configure the node processing time limit, the priority component is used to set the urgency of node processing, and the attachment component is used to define the required file types.

[0010] Technical benefits: Facilitates the standardization of time, priority, and attachment rules, further improving the configuration standardization of process nodes.

[0011] Preferably, as an improvement, the general data component library adopts tag classification management, the tags including business type tags and approval level tags, the business type tags are used to identify the business scenarios to which the component is applicable, and the approval level tags are used to identify the approval permissions required by the component.

[0012] Technical benefits: Facilitates rapid retrieval and matching of components, reducing component selection costs during cross-departmental process configuration.

[0013] Preferably, as an improvement, the general data component library further includes a dynamic component template generator, which includes: The business requirement parsing unit is used to analyze the business requirement text input by the user and identify key entities and actions through natural language processing technology. The component recommendation unit recommends matching combinations of process node components from the general data component library based on the key entities and actions, and generates an initial template; The adaptive adjustment unit automatically adjusts the component parameters in the template or replaces them with suitable components when it detects changes in business rules.

[0014] Technical benefits: It facilitates the configuration of auxiliary processes through intelligent generation, significantly improving the agility to respond to business changes, while maintaining the standardization advantages brought by component reuse.

[0015] Preferably, as an improvement, the visual editing module includes: The form change unit is used to respond to form change requests, display the database table structure, and receive user drag-and-drop adjustments, deletions, or additions of form components through a visual interface; The process change unit is used to respond to process change requests. It receives user drag-and-drop operations to adjust, delete, or add process nodes and configure their order and logical relationships through a visual interface, generating a new process execution path. Technical benefits: Separating form and process design reduces operational complexity and improves user-friendliness.

[0016] Preferably, as an improvement, the automatic vector tagging module includes a word segmentation unit and a vectorization unit. The word segmentation unit is used to perform semantic word segmentation on the user input text, and the vectorization unit is used to convert the word segmentation results into feature vectors in a high-dimensional vector space. The automatic vector tagging module uses a pre-trained language model for vectorization processing, and the pre-trained language model includes a BERT model or a GPT model.

[0017] Technical benefits: It facilitates improved accuracy in text semantic understanding, supports more precise association analysis, leverages large-scale corpora to enhance vectorization accuracy, and supports cross-domain semantic retrieval.

[0018] Preferably, as an improvement, the text feature vector generated by the automatic vector labeling module is used to support intelligent recommendation of process nodes. When a user creates a new process, the module recommends matching historical process nodes based on the input text.

[0019] Technical benefits: It facilitates intelligent assistance in process configuration and reduces the workload of manually searching for components.

[0020] Preferably, as an improvement, the business system integration module supports timed synchronization and event-triggered synchronization. The timed synchronization performs data synchronization at preset time intervals, and the event-triggered synchronization performs data synchronization immediately upon completion of a specific process node.

[0021] Technical benefits: It supports diverse synchronization strategies to meet the real-time requirements of different business scenarios.

[0022] Preferably, as an improvement, the business system integration module includes a data bus and a real-time event stream engine; the data bus is used to build an enterprise-level data transmission channel, supporting data standardization and format conversion between different business systems; the real-time event stream engine is used to monitor state change events of process nodes, and automatically triggers response actions of related business systems when a specific event occurs; the real-time event stream engine includes an event definition unit, a rule configuration unit, and an execution scheduling unit, the event definition unit is used to define triggerable event types, the rule configuration unit is used to configure the mapping relationship between events and response actions, and the execution scheduling unit is used to control the execution order and parameter passing of response actions.

[0023] Technical effects: By introducing a data bus and event stream mechanism, data integration is upgraded from static synchronization to dynamic collaboration, significantly improving the real-time performance, flexibility and maintainability of cross-system collaboration.

[0024] Preferably, as an improvement, it also includes a permission management module, which is used to control user access permissions to process node components and business system data; The permission management module includes a role-component permission mapping engine and an attribute-based dynamic permission engine; The role-component permission mapping engine is based on the RBAC model, which binds process node components in the general data component library with preset roles for permissions. The permission binding includes creation, viewing, editing and deletion operation permissions. The attribute-based dynamic permission engine is based on the ABAC model and dynamically calculates access permissions according to user attributes, resource attributes, and environment attributes. When a process node component is reused, it automatically inherits the basic permission configuration of its component library, and supports overriding and adjusting it in a specific process instance; when RBAC and ABAC rules conflict in terms of permissions, they are automatically adjudicated through preset priority rules.

[0025] Technical benefits: By reusing component-level permissions, redundant configurations are reduced, and by controlling both static roles and dynamic attributes, permission management efficiency is improved. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the structure of a custom intelligent agent construction process development system. Detailed Implementation

[0027] The following detailed description illustrates the specific implementation method: The basic implementation examples are as follows: Figure 1 As shown: A custom intelligent agent construction process development system includes a general data component library, a visual editing module, an automatic vector labeling module, and a business system integration module.

[0028] A general data component library stores reusable process node components. These components include description, approval, and evaluation components. The description component describes events; the approval component approves processes; and the evaluation component evaluates event outcomes. The approval and evaluation components are added to process node components as needed. The library also includes time, priority, and attachment components. The time component configures node processing time limits, the priority component sets the urgency of node processing, and the attachment component defines required file types. The library uses tag-based classification management. Tags include business type tags and approval level tags. Business type tags identify the applicable business scenarios for the component, and approval level tags identify the required approval permissions for the component.

[0029] For example, in the practical application of the office supplies procurement application process: First, the description component is used to collect basic information about the applicant, including the name, specifications, quantity, and estimated unit price of the required office supplies. Next, approval components are added sequentially, corresponding to the approval stages of the department head, administration department, and finance department. Each approval component has corresponding approval permissions and form fields, such as the maximum amount that the department head can approve, and the administration department focusing on the necessity of the office supplies. Finally, after the procurement is completed, an evaluation component is added, allowing the applicant to evaluate the quality of the office supplies and the procurement efficiency. The time component is used to set the processing time limit for the department head approval stage to 2 working days, the administration department's review stage to 1.5 working days, and the finance department's approval stage to 1 working day. Based on the urgency of the procurement, the priority component sets four levels: urgent, high, medium, and low, prioritizing approval messages in urgent situations. The attachment component defines the required file type as a PDF procurement list. Meanwhile, a tag-based classification management system is adopted, assigning the "Office Supplies Procurement" business type tag to components applicable to office supplies procurement, and further assigning approval level tags such as "Department Head Level," "Administration Department Level," and "Finance Department Level" based on approval authority. In subsequent office supplies procurement process configuration, the required components can be quickly located by filtering for the "Office Supplies Procurement" tag, improving process configuration efficiency. The specific execution process is as follows: After an applicant submits a procurement request, the system automatically identifies the procurement amount and urgency level, and pushes approval tasks according to pre-set priority rules. If the procurement request is marked as urgent, the system will prioritize sending an approval reminder to the department head. After the department head completes the approval within 2 working days, the process automatically flows to the administration department for review, which must complete the review within 1.5 working days. After approval, it enters the finance department's approval stage, which completes the approval within 1 working day. After approval, the procurement department executes the procurement task and notifies the applicant for evaluation upon completion. After the applicant submits the evaluation, the process ends. Throughout the entire process, the system automatically tracks the processing progress and time limits of each node to ensure timely completion.

[0030] The general data component library adopts a distributed storage architecture and can be used for editing any business process, including manual and automated business processes, such as ERP, MES, CRM, SRM, PLM, OA, etc. It can realize business flow replication, greatly reduce the development cost of business systems, and improve reusability.

[0031] The general data component library also includes a dynamic component template generator to facilitate intelligent generation of auxiliary process configurations, significantly improving agility in responding to business changes while maintaining the standardization advantages brought by component reuse. The dynamic component template generator includes a business requirement parsing unit, a component recommendation unit, and an adaptive adjustment unit. The business requirement parsing unit analyzes the business requirement text input by the user, identifying key entities and actions through natural language processing technology. The component recommendation unit, based on the key entities and actions, recommends matching process node component combinations from the general data component library and generates an initial template. The adaptive adjustment unit automatically adjusts the component parameters in the template or replaces them with more suitable components when it detects changes in business rules. It also includes a historical case learning unit to record historical successful configuration cases and establish a requirement-component mapping knowledge base to optimize recommendation accuracy. Furthermore, it includes an anomaly identification and optimization suggestion unit to periodically analyze process execution data, calculate the average processing time and delay rate of each node, mark nodes with abnormally long processing times or high delay rates as potential bottlenecks, and analyze the logical relationships between nodes through semantic retrieval to identify nodes that can be merged or processed in parallel. Generate process optimization suggestions, including node merging, parallel processing, and automated processing. The optimization suggestions are presented to the administrator in the form of a report, which includes data analysis charts and specific improvement measures.

[0032] The visual editing module provides a drag-and-drop interface, which includes dragging, clicking, inputting, and selecting. It allows for dynamic adjustment of form components and process nodes through drag-and-drop operations.

[0033] The visual editing module adopts a responsive design, adapting to display devices of different sizes. The interface is divided into three parts: a left-side component panel, a central editing area, and a right-side property panel. The left-side component panel displays available form and process node components, presented as icons and names, and supports searching and categorized filtering. The central editing area provides a grid layout, displaying the currently edited form or process. The right-side property panel dynamically displays editable properties based on the selected component, including basic information, style settings, data binding, and event handling. The visual editing module supports keyboard shortcuts to improve editing efficiency.

[0034] The visual editing module includes a form change unit and a process change unit.

[0035] The Form Change Unit responds to form change requests, displays the database table structure, and receives user drag-and-drop adjustments, deletions, or additions to form components through a visual interface. It provides a form preview function, displaying the editing effects in real time. The Form Change Unit supports importing and exporting form templates for easy cross-system reuse. Further, the Form Change Unit includes a dynamic database update module, which updates the database table structure in real time based on user editing operations on form components. This module uses ORM technology to map form components to database fields, supporting the setting of field types, lengths, and constraints. Before performing database changes, the dynamic database update module performs a security check to prevent data loss.

[0036] The process change unit receives user drag-and-drop adjustments, deletions, or additions to configure the order and logical relationships of process nodes through a visual interface, generating new process execution paths. It provides process simulation functionality to verify the rationality of the process design. The unit supports conditional branching, dynamically determining the process direction based on form data values. Further, the unit includes a process execution record generation module and an intelligent analysis module. The execution record generation module automatically generates paper documents based on the new process path, processes the execution records by title, and synchronizes data across systems. It supports multiple output formats, including but not limited to PDF, Word, and HTML. The module automatically generates process titles using the naming convention "[Process Type]-[Initiator]-[Initiation Time]". The intelligent analysis module performs semantic retrieval and correlation analysis on the process execution data, identifying the logical relationships between process nodes. It uses graph algorithms to analyze the process network structure, identifying key nodes and potential bottlenecks. The module generates process optimization suggestions, including node merging, parallel processing, and automation solutions.

[0037] An automatic vector tagging module is used to vectorize user-input text content, generating text feature vectors that support semantic retrieval and association analysis. The automatic vector tagging module includes a word segmentation unit and a vectorization unit. The word segmentation unit performs semantic word segmentation on the user-input text, and the vectorization unit converts the segmentation results into feature vectors in a high-dimensional vector space. The automatic vector tagging module uses a pre-trained language model for vectorization, including either a BERT model or a GPT model. The text feature vectors generated by the automatic vector tagging module support intelligent recommendation of process nodes. When a user creates a new process, matching historical process nodes are recommended based on the input text, facilitating intelligent assistance in process configuration and reducing the workload of manually searching for components.

[0038] The business system integration module is used to synchronize the generated approval process execution records with the data of the business systems. These business systems include ERP, MES, CRM, SRM, PLM, and OA systems, enabling cross-system collaboration. The module supports scheduled synchronization and event-triggered synchronization. Scheduled synchronization executes data synchronization at preset time intervals, while event-triggered synchronization executes data synchronization immediately upon completion of a specific process node, supporting diverse synchronization strategies to meet the real-time requirements of different business scenarios.

[0039] The business system integration module includes a data bus and a real-time event stream engine. The data bus is used to build an enterprise-level data transmission channel, supporting data standardization and format conversion between different business systems. The real-time event stream engine is used to monitor status change events of process nodes, and automatically triggers response actions from related business systems when specific events occur (such as approval or task completion). The real-time event stream engine includes an event definition unit, a rule configuration unit, and an execution scheduling unit. The event definition unit is used to define triggerable event types, the rule configuration unit is used to configure the mapping relationship between events and response actions, and the execution scheduling unit is used to control the execution order and parameter passing of response actions. By introducing the data bus and event stream mechanism, data integration is upgraded from static synchronization to dynamic collaboration, significantly improving the real-time performance, flexibility, and maintainability of cross-system collaboration.

[0040] The business system integration module employs both RESTful APIs and message queues for inter-system communication. It supports data mapping configuration to resolve differences in field names and data formats between different systems. The module provides data synchronization status monitoring, recording success rates and reasons for failures. It enables bidirectional data synchronization, allowing the pushing of process data to and from business systems to update process status. Furthermore, it supports batch data processing, improving the efficiency of synchronizing large amounts of data.

[0041] It also includes a permission management module, which controls user access permissions to process node components and business system data. This module includes a role-component permission mapping engine and an attribute-based dynamic permission engine. The role-component permission mapping engine, based on the RBAC model, binds process node components in a general data component library to preset roles for permissions, including creation, viewing, editing, and deletion permissions. The attribute-based dynamic permission engine, based on the ABAC model, dynamically calculates access permissions based on user attributes, resource attributes, and environment attributes. When a process node component is reused, it automatically inherits the basic permission configuration of its parent component library, while also supporting overriding adjustments in specific process instances. When RBAC and ABAC rules conflict, they are automatically resolved through preset priority rules. Component-level permission reuse reduces redundant configuration, and dual control through static roles and dynamic attributes improves permission management efficiency.

[0042] The above descriptions are merely embodiments of the present invention, and common knowledge such as specific technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solutions of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A custom intelligent agent construction process development system, characterized in that: It includes a general data component library, a visual editing module, an automatic vector labeling module, and a business system integration module; The general data component library is used to store reusable process node components; The process node component includes a description component, an approval component, and an evaluation component. The description component is used to describe the event; the approval component is used for process approval; and the evaluation component is used to evaluate the event outcome. The approval component and the evaluation component are added to the process node component as needed. The visual editing module provides a drag-and-drop interface, which includes dragging, clicking, inputting, and selecting. Form components and process nodes can be dynamically adjusted through drag-and-drop operations. The automatic vector tagging module is used to vectorize the text content input by the user, generating text feature vectors that support semantic retrieval and association analysis; The business system integration module is used to synchronize the generated approval process execution records with the business system data.

2. The custom intelligent agent construction process development system according to claim 1, characterized in that: The general data component library also includes a time component, a priority component, and an attachment component. The time component is used to configure the node processing time limit, the priority component is used to set the urgency of node processing, and the attachment component is used to define the required file types.

3. The custom intelligent agent construction process development system according to claim 1, characterized in that: The general data component library adopts tag-based classification management. The tags include business type tags and approval level tags. The business type tags are used to identify the business scenarios to which the component is applicable, and the approval level tags are used to identify the approval permissions required by the component.

4. The custom intelligent agent construction process development system according to claim 1, characterized in that, The general data component library also includes a dynamic component template generator, which includes: The business requirement parsing unit is used to analyze the business requirement text input by the user and identify key entities and actions through natural language processing technology. The component recommendation unit recommends matching combinations of process node components from the general data component library based on the key entities and actions, and generates an initial template; The adaptive adjustment unit automatically adjusts the component parameters in the template or replaces them with suitable components when it detects changes in business rules.

5. The custom intelligent agent construction process development system according to claim 1, characterized in that, The visual editing module includes: The form change unit is used to respond to form change requests, display the database table structure, and receive user drag-and-drop adjustments, deletions, or additions of form components through a visual interface; The process change unit is used to respond to process change requests. It receives user drag-and-drop adjustments, deletions, or additions of process nodes through a visual interface to configure the order and logical relationships of process nodes and generate new process execution paths.

6. The custom intelligent agent construction process development system according to claim 1, characterized in that: The automatic vector tagging module includes a word segmentation unit and a vectorization unit. The word segmentation unit is used to perform semantic word segmentation on the user input text, and the vectorization unit is used to convert the word segmentation results into feature vectors in a high-dimensional vector space. The automatic vector tagging module uses a pre-trained language model for vectorization processing, and the pre-trained language model includes a BERT model or a GPT model.

7. The custom intelligent agent construction process development system according to claim 1, characterized in that: The text feature vectors generated by the automatic vector labeling module are used to support intelligent recommendation of process nodes. When a user creates a new process, the module recommends matching historical process nodes based on the input text.

8. The custom intelligent agent construction process development system according to claim 1, characterized in that: The business system integration module supports scheduled synchronization and event-triggered synchronization. Scheduled synchronization performs data synchronization at preset time intervals, while event-triggered synchronization performs data synchronization immediately upon completion of a specific process node.

9. The custom intelligent agent construction process development system according to claim 1, characterized in that: The business system integration module includes a data bus and a real-time event stream engine; the data bus is used to build an enterprise-level data transmission channel, supporting data standardization and format conversion between different business systems. The real-time event flow engine is used to monitor state change events of process nodes. When a specific event occurs, it automatically triggers the response action of the associated business system. The real-time event flow engine includes an event definition unit, a rule configuration unit, and an execution scheduling unit. The event definition unit is used to define the types of events that can be triggered. The rule configuration unit is used to configure the mapping relationship between events and response actions. The execution scheduling unit is used to control the execution order and parameter passing of response actions.

10. A custom intelligent agent construction process development system according to claim 1, characterized in that: It also includes a permission management module, which is used to control user access permissions to process node components and business system data; The permission management module includes a role-component permission mapping engine and an attribute-based dynamic permission engine; The role-component permission mapping engine is based on the RBAC model, which binds process node components in the general data component library with preset roles for permissions. The permission binding includes creation, viewing, editing and deletion operation permissions. The attribute-based dynamic permission engine is based on the ABAC model and dynamically calculates access permissions according to user attributes, resource attributes, and environment attributes. When a process node component is reused, it automatically inherits the basic permission configuration of its component library, and supports overriding and adjusting it in a specific process instance; when RBAC and ABAC rules conflict in terms of permissions, they are automatically adjudicated through preset priority rules.

Images