Low-code configurable AI integration system and method based on dynamic intelligent agents

Abstract

This invention relates to the technical field of integrating office software with artificial intelligence applications, and in particular discloses a low-code configurable AI integration system and method based on dynamic intelligent agents. The system includes a text editing module, a parameterized interface module, a prompt word construction module, an intelligent agent module, a secure communication module, a summarization module, and a document writing module. The text editing module initiates an intelligent analysis request to the parameterized interface module based on target content. The parameterized interface module extracts the target content to obtain multiple variables to be analyzed. The prompt word construction module generates multiple prompt words based on the multiple variables to be analyzed. The secure communication module encapsulates the encrypted prompt words and authentication information to obtain encapsulated information. The intelligent agent module generates analysis results based on the encapsulated information. The summarization module converts the analysis results into JSON data. The document writing module, in conjunction with preset macro injection instructions, fills in or modifies the target content.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of office software and artificial intelligence application integration, and particularly to a low-code configurable AI integration system and method based on dynamic agents. BACKGROUND

[0002] With the popularity of large language models (LLM) in the office field, it has become a mainstream trend to deeply integrate AI capabilities into office software. WPS AI, Dingding intelligent assistant, Feishu intelligent partner and other general-purpose products have shown potential in document drafting, table analysis, meeting minutes and other aspects; however, in the power grid, water conservancy and other key information infrastructure industries, general solutions have natural limitations in data security and domain knowledge adaptation, so building a private and controllable office assistant using a large model deployed in the enterprise intranet has become a frontier and blank in the industry.

[0003] Therefore, based on the above technical problems, the present application proposes a low-code configurable AI integration system and method based on dynamic agents to fill the gap in existing intranet AI assistants and improve security. SUMMARY

[0004] The present application aims to overcome the shortcomings of the prior art and provide a low-code configurable AI integration system and method based on dynamic agents to fill the gap in existing intranet AI assistants and improve security.

[0005] To achieve the above objectives, this invention provides a low-code configurable AI integration system based on dynamic intelligent agents, comprising a text editing module, a parameterized interface module, a prompt word construction module, an intelligent agent module, an intelligent agent configuration module, a secure communication module, a summary module, and a document writing module. The text editing module initiates an intelligent analysis request to the parameterized interface module after the user selects target content. The parameterized interface module, upon receiving the intelligent analysis request from the user terminal module, extracts the target content and its associated contextual content to obtain multiple variables to be analyzed, and sends these variables to the prompt word construction module. The prompt word construction module fills the multiple variables to be analyzed into a preset search template to generate multiple prompt words recognizable by the intelligent agent, and sends these prompt words to the secure communication module. The intelligent agent configuration module outputs information to the intelligent agent module based on the type of business being processed by the user. A request is made to invoke a corresponding agent to provide services. The request includes a service instruction to drive the agent and authentication information corresponding to that agent. A secure communication module encrypts the prompt word to obtain an encrypted prompt word and simultaneously acquires the authentication information from the request. The encrypted prompt word and the authentication information are then encapsulated to obtain encapsulated information, which is sent to the agent module. The agent module includes multiple agents capable of handling various business processes. Based on the encapsulated information, the agent module performs concurrent analysis in conjunction with a pre-defined concurrent request module to generate analysis results. A summary module converts the analysis results into JSON data containing key-value pairs and sends the JSON data to a document writing module. The document writing module parses the JSON data and, in conjunction with pre-defined macro injection instructions, fills in or modifies the target content in the text editing module.

[0006] Furthermore, the text editing module includes an editing unit and a candidate column unit. The editing unit is used for users to process business in an internal office environment. The candidate column unit is used to simulate the state of the target content after it has been filled or modified after receiving the content to be filled or modified from the document writing module. The candidate column unit is equipped with a direct fill mode and an interactive confirmation fill mode.

[0007] Furthermore, the text editing module also includes a backup unit, which is used to back up the target content before the text editing module fills in or modifies the target content.

[0008] Furthermore, the system also includes a web front-end module connected to the text editing module, which allows users to process business through the text editing module in an external office environment.

[0009] Furthermore, the authentication information includes the service address, application ID, and key of a certain intelligent agent.

[0010] Furthermore, the encapsulated information is assembled in the following format: authentication information, encrypted prompts, target URL, request method, and content type.

[0011] Furthermore, the concurrent request module includes a request distribution unit, at least one data plugin unit, a business interface unit, and a knowledge base unit. The request distribution unit is used to parse the encrypted prompt words and concurrently send the prompt words to the data plugin unit, the business retrieval unit, and the knowledge base unit. The data plugin unit is connected to a pre-set internal information module and is used to retrieve historical business data based on the prompt words in the internal information module, and input the historical business data into the intelligent agent module for analysis. The business interface unit is used to acquire real-time business data or perform logical operations on specific businesses to obtain logical operation results, and input the real-time business data or the above logical operation results into the intelligent agent module for analysis. The knowledge base unit pre-stores internal standard document data, procedure data, and case data, and is used to retrieve data related to the prompt words based on RAG technology, and input the data into the intelligent agent module for analysis.

[0012] A low-code, configurable AI integration method based on dynamic agents includes the following steps: Step S1: When the user selects the target content in the text editing module, an intelligent analysis request is sent to the parameterized interface module; Step S2: After receiving the intelligent analysis request sent by the user terminal module, the parameterization interface module extracts the target content and the context content associated with the target content to obtain multiple variables to be analyzed, and sends the multiple variables to be analyzed to the prompt word construction module. At the same time, the intelligent agent configuration module outputs a mobilization request to the intelligent agent module based on the business type processed by the user to call the corresponding intelligent agent to provide services. The mobilization request includes a service instruction to drive a certain intelligent agent to run and the authentication information corresponding to the intelligent agent. Step S3: After the prompt word construction module obtains multiple variables to be analyzed, the prompt word construction module fills the multiple variables to be analyzed into the preset search template to generate multiple prompt words that the agent can recognize, and sends the multiple prompt words to the secure communication module. Step S4: After the secure communication module obtains multiple prompt words, it encrypts the prompt words to obtain encrypted prompt words. At the same time, it obtains the authentication information in the mobilization request, encapsulates the encrypted prompt words and the authentication information in the mobilization request to obtain encapsulated information, and then sends the encapsulated information to the intelligent agent module. Step S5: After the agent module obtains the encapsulation information, based on the encapsulation information, the agent module cooperates with the preset concurrent request module to perform concurrent analysis to generate analysis results; Step S6: After obtaining the analysis results in the summary module, instruct the summary module to convert the analysis results into JSON data containing key-value pairs, and send the JSON data to the document writing module; Step S7: After the document writing module obtains the JSON data, it parses the JSON data and, in conjunction with the preset macro injection instructions, fills in or modifies the target content in the text editing module.

[0013] Based on the same inventive concept, this application also provides a data processing device for a low-code configurable AI integrated system based on dynamic intelligent agents, including: a memory and a processor, wherein the memory is used to store a computer program; and the processor is used to execute the computer program to implement the steps of the aforementioned low-code configurable AI integrated system based on dynamic intelligent agents.

[0014] Based on the same inventive concept, this application also provides a computer-readable storage medium storing a computer program, and based on the same inventive concept, this application also provides the steps of a low-code configurable AI integrated system based on dynamic intelligent agents.

[0015] The present invention adopts the above-described solution, and its beneficial effects are as follows: 1. By establishing a closed-loop intranet for AI calls across the entire process, the entire retrieval and analysis of target content is conducted within a closed intranet. This avoids data leakage of sensitive data in the target content due to routine network searches, thus resolving the data security risks of public data cloud solutions. At the same time, by setting up a secure communication module to confirm the authentication information of relevant data, the security of data transmission is improved.

[0016] 2. Each intelligent agent integrates knowledge from specific domains (such as highly specialized knowledge of power grid operation and maintenance, production scheduling, and safety procedures). At the same time, by setting up an intelligent agent configuration module, relevant intelligent agent services can be dynamically configured for different business types, thereby increasing the types of business that the system can handle and solving the problems of knowledge barriers and weak business adaptability of general models.

[0017] 3. By setting a parameterized interface module, the target content and its associated context can be completely transmitted, ensuring a complete analysis of the target content text. It also enables the agent to concurrently aggregate multi-source data during analysis, referencing existing cases and other reference data, thereby improving the accuracy of the analysis results. This solves the problems of inefficient context transmission and overly biased expert analysis results in existing technologies.

[0018] 4. This invention can be replicated and promoted at low cost within an intranet, and is suitable for scenarios such as power grids, energy, finance, and government affairs that have high requirements for data security and need to deeply integrate domain knowledge with office processes, thus having clear industrial application value. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of the low-code configurable AI integration system based on dynamic agents in this embodiment.

[0020] Figure 2 This is a flowchart illustrating the low-code configurable AI integration method based on dynamic agents in this embodiment. Detailed Implementation

[0021] To facilitate understanding of the present invention, a more complete description is given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention. Example

[0022] See appendix Figure 1As shown in this embodiment, a low-code configurable AI integration system based on dynamic intelligent agents includes a text editing module, a parameterized interface module, a prompt word construction module, an intelligent agent module, an intelligent agent configuration module, a secure communication module, a summary module, and a document writing module. The text editing module interacts with the user and, after the user selects target content, initiates an intelligent analysis request to the parameterized interface module. This text editing module is designed for internal office environments (e.g., WPS) and can process text, image tables, and data tables. The parameterized interface module, upon receiving the intelligent analysis request from the user terminal module, extracts the target content and its associated contextual content to obtain multiple variables to be analyzed and sends these variables to the prompt word construction module. The prompt word construction module fills these variables into a preset search template to generate multiple prompt words recognizable by the intelligent agent and sends these prompt words to the secure communication module. The intelligent agent configuration module is used to configure the system based on the type of business being processed by the user (e.g., regulatory review). For tasks such as business processing, text summarization, and data verification, a request is sent to the agent module to invoke the agent corresponding to the business type. The request includes the service instruction that drives the agent and the authentication information corresponding to that agent, thereby assisting the user in handling related business. The secure communication module encrypts the prompt word to obtain the encrypted prompt word and the authentication information in the request. The encrypted prompt word and the authentication information in the request are then encapsulated to obtain encapsulated information, which is then sent to the agent module. The encapsulated information is formatted as authentication information, encrypted prompt word, target URL, request method, and content type. Specifically, the authentication information is added to the HTTP request header, the encrypted prompt word is placed in the request body, and the target URL, request method (POST), and content type (application / json) are specified. By setting the above encapsulation information, the security and protocol compliance of the data at the transport layer are ensured.

[0023] The intelligent agent module includes multiple intelligent agents capable of handling various types of business (each intelligent agent integrates knowledge from specific domains, such as highly specialized knowledge domains like power grid operation and maintenance, production scheduling, and safety procedures). The intelligent agent module is used to perform concurrent analysis based on encapsulated information, in conjunction with a pre-defined concurrent request module, to generate analysis results. The aggregation module is used to convert the analysis results into JSON data containing key-value pairs and send the JSON data to the document writing module. The document writing module is used to parse the JSON data and, in conjunction with pre-defined macro injection instructions, to populate or modify the target content in the text editing module.

[0024] Furthermore, the text editing module includes an editing unit and a candidate column unit. The editing unit is used by users to process business in an internal office environment. The candidate column unit is used to simulate the state of the target content after it has been filled or modified, after receiving the content to be filled or modified from the document writing module. The candidate column unit has a direct fill mode and an interactive confirmation fill mode. The direct fill mode is used when processing general business to directly trigger macro commands to modify the target content (bookmark replacement) without requiring secondary confirmation from the user, thereby improving processing efficiency. For business with higher risk or requiring manual review, the interactive confirmation fill mode is used, allowing the user to see the state of the text / data / table after the target content has been filled or modified in the candidate column unit. The user then confirms that the state of the filled or modified content meets their needs. Then, the target content is filled / replaced to avoid misjudgment or omission, thus ensuring the accuracy of the system's business processing. Secondly, the above-mentioned text editing module also includes a backup unit. The backup unit is used to back up the target content (i.e., snapshot data) before the text editing module fills in or modifies the target content. When an error occurs during the batch writing process (for example, a JSON data containing 10 fields, when filling in the 5th replacement content, an error occurs because the bookmark is not present or the user is not satisfied with the generated result), the target content is restored to the state before modification by rereading the backup data saved by the above-mentioned backup unit (i.e., rollback), avoiding incomplete data that is only half modified, thus ensuring the integrity and consistency of document data.

[0025] Furthermore, the system also includes a web front-end module connected to the text editing module. This web front-end module allows users to process business in an external office environment through the text editing module, thus enabling users to process business in an external office environment even when the system is on an intranet. This avoids sensitive data leaving the network and resolves the data security risks of public cloud solutions. At the same time, the web front-end module enables the system to support external office environments, which is conducive to large-scale operation and maintenance and expansion. It also lowers the threshold and cost of applying AI in multiple scenarios and improves its scalability and cost-effectiveness.

[0026] In this embodiment, the authentication information includes the service address (URL - network access address / interface endpoint of the agent service), application ID (AppId - unique identifier of the application), and key (AppSecret - access key used to request signature) of a certain agent.

[0027] In this embodiment, the concurrent request module includes a request distribution unit, at least one data plug-in unit, a business interface unit, and a knowledge base unit. The request distribution unit is used to parse the encrypted prompt words and concurrently send the prompt words to the data plug-in unit, the business retrieval unit, and the knowledge base unit. The data plug-in unit is connected to a preset internal information module and is used to perform retrieval in the internal information module based on the prompt words (i.e., using the prompt words as query conditions) to obtain historical business data (including equipment operation and maintenance status, production plans, etc.). The historical business data is then input into the intelligent agent module for analysis to achieve data-driven intelligent analysis and increase substantive value. Furthermore, the specific number of the above-mentioned data plug-in units can be set according to the actual situation (for example, setting two data plug-in units respectively connected to e-report and power grid management platform). The business interface unit is used to acquire real-time business data or perform logical operations for specific business operations to obtain the results of the logical operations. It then inputs the real-time business data or the results of the above logical operations (compliance verification results) into the intelligent agent module for analysis, thereby realizing real-time analysis / compliance analysis. The knowledge base unit pre-stores internal standard document data, procedure data, and case data. It is used to retrieve relevant information data based on RAG technology and input the information data into the intelligent agent module for analysis, thereby generating normative reference data and generating evidence-based responses, reducing data illusion.

[0028] The following description is based on specific embodiments: Taking the scenario of equipment defect assisted audit as an example, the specific process is as follows: Intelligent analysis request initiated: The user selects the defect description text in the text editing module; Parameter extraction and agent invocation: The agent configuration module outputs a request to the agent module based on the type of business handled by the user, so as to invoke an agent in the domain that can handle defects to provide services; Prompt word construction: The defect description text (target content) is obtained through the parameterized interface module, and multiple variables to be analyzed are generated. The prompt word construction module fills them into the preset template to generate multiple complete prompt words, and sends the multiple prompt words to the secure communication module. Secure communication transmission: The secure communication module encrypts the prompt words and attaches authentication information to generate encapsulated information, which is then sent to the agent module. Agent Analysis: Based on encapsulated information, the agent calls the concurrent request module to perform concurrent analysis (e.g., querying the defect standard system library and the defect historical case knowledge base to obtain the corresponding report specification format, defect judgment standard, etc.) to generate analysis results, and then sends the analysis results to the aggregation module; Results Summary: The summary module organizes the analysis results into JSON data (which may include defect classification, defect causes, defect consequence analysis, acceptance criteria, etc.) and sends the JSON data to the document writing module. Document Writing: The document writing module parses the JSON data and triggers a preset macro injection instruction to fill the key-value pairs in the JSON into the selected defect description text, thereby automatically generating a standardized defect analysis report.

[0029] Table 1 shows a comparison of this system with traditional processing methods in different application scenarios. Table 1. Typical Application Scenarios and Comparative Analysis:

[0030] In summary, this embodiment has the following characteristics: Security: By establishing a closed-loop intranet for the entire AI call process, sensitive data is prevented from leaving the network, thus solving the data security risks of public data cloud solutions. At the same time, by setting up a secure communication module to confirm the authentication information of relevant data, the security of data transmission is improved. Specialization and adaptability: Each intelligent agent integrates knowledge of specific domains (such as highly specialized knowledge of power grid operation and maintenance, production scheduling, and safety procedures). At the same time, by setting up an intelligent agent configuration module, relevant intelligent agent services can be dynamically configured according to business types, which relatively increases the types of business that the system can handle and solves the problems of knowledge barriers and weak business adaptability of general models. Ease of use and accuracy: By setting parameterized interface modules, the target content and the context associated with the target content can be completely transmitted, ensuring a complete analysis of the target content text. It also enables the agent to concurrently aggregate multi-source data during analysis, refer to existing cases and other reference data, improve the accuracy of the analysis results, and solve the problems of inefficient context transmission and overly one-sided expert analysis results in existing technologies. Industrial applicability: This invention can be replicated and promoted at low cost on intranets, and is suitable for scenarios such as power grids, energy, finance, and government affairs that have high requirements for data security and need to deeply integrate domain knowledge with office processes, thus having clear industrial application value. Example

[0031] See appendix Figure 2 As shown, a low-code configurable AI integration method based on dynamic intelligent agents includes the following steps: Step S1: When the user selects the target content in the text editing module, an intelligent analysis request is sent to the parameterized interface module; Step S2: After receiving the intelligent analysis request sent by the user terminal module, the parameterized interface module extracts the target content and the context content associated with the target content to obtain multiple variables to be analyzed. The multiple variables to be analyzed are then sent to the prompt word construction module. At the same time, the intelligent agent configuration module outputs a mobilization request to the intelligent agent module based on the business type processed by the user to call the corresponding intelligent agent to provide services. The mobilization request includes the service instruction that drives a certain intelligent agent to run and the authentication information corresponding to that intelligent agent. Step S3: After the prompt word construction module obtains multiple variables to be analyzed, the prompt word construction module fills the multiple variables to be analyzed into the preset search template to generate multiple prompt words that the agent can recognize, and sends the multiple prompt words to the secure communication module. Step S4: After the secure communication module obtains multiple prompt words, it encrypts the prompt words to obtain encrypted prompt words. At the same time, it obtains the authentication information in the mobilization request, encapsulates the encrypted prompt words and the authentication information in the mobilization request to obtain encapsulated information, and then sends the encapsulated information to the intelligent agent module. Step S5: After the agent module obtains the encapsulation information, based on the encapsulation information, the agent module cooperates with the preset concurrent request module to perform concurrent analysis to generate analysis results; Step S6: After obtaining the analysis results in the summary module, instruct the summary module to convert the analysis results into JSON data containing key-value pairs, and send the JSON data to the document writing module; Step S7: After the document writing module obtains the JSON data, it parses the JSON data and, in conjunction with the preset macro injection instructions, fills in or modifies the target content in the text editing module. Example

[0032] Based on the same inventive concept, this application also provides a data processing device for a low-code configurable AI integrated system based on dynamic intelligent agents, including: a memory and a processor, wherein the memory is used to store a computer program; and the processor is used to execute the computer program to implement the steps of the low-code configurable AI integrated system based on dynamic intelligent agents as described above. Example

[0033] Based on the same inventive concept, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the aforementioned low-code configurable AI integrated system based on the same inventive concept.

[0034] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Any modifications or variations made by those skilled in the art, without departing from the scope of the present invention, using the disclosed technical content, are equivalent embodiments of the present invention. Therefore, all equivalent changes made based on the concept of the present invention without departing from the scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A low-code configurable AI integration system based on dynamic intelligent agents, characterized in that: It includes a text editing module, a parameterized interface module, a prompt word construction module, an agent module, an agent configuration module, a secure communication module, a summary module, and a document writing module. The text editing module is used to send an intelligent analysis request to the parameterized interface module after the user selects the target content; The parameterized interface module is used to extract the target content and the context content associated with the target content after receiving the intelligent analysis request sent by the user terminal module, obtain multiple variables to be analyzed, and send the multiple variables to be analyzed to the prompt word construction module. The prompt word construction module is used to fill multiple variables to be analyzed into a preset search template to generate multiple prompt words that the intelligent agent can recognize, and send the multiple prompt words to the secure communication module; The intelligent agent configuration module is used to output a dispatch request to the intelligent agent module based on the type of business processed by the user, so as to call the corresponding intelligent agent to provide services. The dispatch request includes a service instruction to drive a certain intelligent agent to run and authentication information corresponding to the intelligent agent. The secure communication module is used to encrypt the prompt word to obtain the encrypted prompt word, and simultaneously obtain the authentication information in the mobilization request. The encrypted prompt word and the authentication information in the mobilization request are then encapsulated to obtain encapsulated information, which is then sent to the intelligent agent module. The intelligent agent module includes multiple intelligent agents capable of handling various types of different services. The intelligent agent module is used to perform concurrent analysis based on the encapsulated information in conjunction with a preset concurrent request module to generate analysis results. The aggregation module is used to convert the analysis results into JSON data containing key-value pairs and send the JSON data to the document writing module; The document writing module is used to parse JSON data and, in conjunction with preset macro injection instructions, to fill in or modify the target content in the text editing module.

2. A low-code configurable AI integration system based on dynamic intelligent agents as described in claim 1, characterized in that: The text editing module includes an editing unit and a candidate column unit. The editing unit is used for users to process business in an internal office environment. The candidate column unit is used to simulate the state of the target content after it has been filled or modified after receiving the content to be filled or modified from the document writing module. The candidate column unit has a direct fill mode and an interactive confirmation fill mode.

3. A low-code configurable AI integration system based on dynamic intelligent agents as described in claim 2, characterized in that: The text editing module also includes a backup unit, which is used to back up the target content before the text editing module fills in or modifies the target content.

4. A low-code configurable AI integration system based on dynamic intelligent agents as described in claim 1, characterized in that: The system also includes a web front-end module connected to the text editing module, which allows users to process business through the text editing module in an external office environment.

5. A low-code configurable AI integration system based on dynamic intelligent agents as described in claim 1, characterized in that: The authentication information includes the service address, application ID, and key of a certain intelligent agent.

6. A low-code configurable AI integration system based on dynamic intelligent agents as described in claim 1, characterized in that: The encapsulated information is assembled in the following format: authentication information, encrypted prompt, target URL, request method, and content type.

7. A low-code configurable AI integration system based on dynamic intelligent agents as described in claim 1, characterized in that: The concurrent request module includes a request distribution unit, at least one data plugin unit, a business interface unit, and a knowledge base unit. The request distribution unit is used to parse the encrypted prompt words and concurrently send the prompt words to the data plugin unit, the business retrieval unit, and the knowledge base unit. The data plugin unit is connected to a pre-set internal information module and is used to retrieve historical business data based on the prompt words in the internal information module, and input the historical business data into the intelligent agent module for analysis. The business interface unit is used to acquire real-time business data or perform logical operations on specific business operations to obtain the logical operation results, and input the real-time business data or the above logical operation results into the intelligent agent module for analysis. The knowledge base unit pre-stores internal standard document data, procedure data, and case data, and is used to retrieve data related to the prompt words based on RAG technology, and input the data into the intelligent agent module for analysis.

8. A low-code configurable AI integration method based on dynamic intelligent agents, characterized in that, Includes the following steps: Step S1: When the user selects the target content in the text editing module, an intelligent analysis request is sent to the parameterized interface module; Step S2: After receiving the intelligent analysis request sent by the user terminal module, the parameterization interface module extracts the target content and the context content associated with the target content to obtain multiple variables to be analyzed, and sends the multiple variables to be analyzed to the prompt word construction module. At the same time, the intelligent agent configuration module outputs a mobilization request to the intelligent agent module based on the business type processed by the user to call the corresponding intelligent agent to provide services. The mobilization request includes a service instruction to drive a certain intelligent agent to run and the authentication information corresponding to the intelligent agent. Step S3: After the prompt word construction module obtains multiple variables to be analyzed, the prompt word construction module fills the multiple variables to be analyzed into the preset search template to generate multiple prompt words that the agent can recognize, and sends the multiple prompt words to the secure communication module. Step S4: After the secure communication module obtains multiple prompt words, it encrypts the prompt words to obtain encrypted prompt words. At the same time, it obtains the authentication information in the mobilization request, encapsulates the encrypted prompt words and the authentication information in the mobilization request to obtain encapsulated information, and then sends the encapsulated information to the intelligent agent module. Step S5: After the agent module obtains the encapsulation information, based on the encapsulation information, the agent module cooperates with the preset concurrent request module to perform concurrent analysis to generate analysis results; Step S6: After obtaining the analysis results in the summary module, instruct the summary module to convert the analysis results into JSON data containing key-value pairs, and send the JSON data to the document writing module; Step S7: After the document writing module obtains the JSON data, it parses the JSON data and, in conjunction with the preset macro injection instructions, fills in or modifies the target content in the text editing module.

9. A data processing device for a low-code configurable AI integrated system based on dynamic intelligent agents, characterized in that, include: A memory and a processor, the memory being used to store a computer program; the processor being used to execute the computer program to implement the steps of a low-code configurable AI integration system based on a dynamic agent as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of a low-code configurable AI integration system based on a dynamic agent as described in any one of claims 1 to 7.

Images