Software development method and system based on natural language arrangement and readable storage medium

Abstract

The invention relates to a software development method and system based on natural language arrangement and a readable storage medium. The method comprises the steps that a programmable system based on natural language arrangement is deployed, basic resources are registered in the programmable system, and registered resources and system resources are subjected to natural language description, then stored in a relation library and synchronized to a vector library; the method comprises the following steps: creating a service application in a programmable system and arranging the service application, performing natural language description on arrangement content, publishing the arrangement content as an intelligent agent, storing the intelligent agent into a relation library and synchronizing the intelligent agent to a vector library; performing vector retrieval on the questioning content of the user by using a large language model, and selecting the most conforming agent; and calling the most conforming agent to execute the corresponding arrangement content according to the paragraph sequence, and outputting a final result. According to the software development method provided by the invention, a user does not need to know the technical principle, the demand is disassembled through a natural language, detailed steps are formed, and execution is completed through a large language model, so that the communication time consumption is reduced, and the possibility of demand deviation is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of software development, and particularly to a software development method, system and readable storage medium based on natural language orchestration. Background Art

[0002] With the acceleration of the digital transformation process, enterprises' demand for automated and intelligent workflow management is becoming increasingly urgent. In the process of software development and business system implementation, how to efficiently convert complex human instructions into executable program instructions to meet the needs of various software development operations has always been a core issue to be solved. The current mainstream implementation methods have the following significant drawbacks:

[0003] 1. Inefficiency of traditional software development models

[0004] In traditional software engineering practices (such as development with programming languages like Java and Python), the realization of business requirements highly depends on professional developers. The complete cycle from requirement analysis to code writing, testing and deployment often takes weeks or even months, involving a large number of cross-departmental requirement communication meetings and repeated modifications. Even with detailed requirement documents, developers still need to gradually translate the logic described in natural language into a code structure recognizable by a computer. This model not only results in high labor costs, but also requires re-coding and re-deployment of the system when requirements change, with severely insufficient flexibility and response speed.

[0005] 2. Applicability limitations of low-code platforms

[0006] In recent years, emerging low-code / no-code development platforms (such as OutSystems and Bubble) have significantly reduced the development threshold through visual component dragging and pre-set logical rules. However, such platforms still require users to have basic technical cognitive abilities, and the main users are still technical users, and there is still the work of communicating with business requirement parties. Summary of the Invention

[0007] The purpose of the present invention is to provide a software development method, system and readable storage medium based on natural language orchestration to solve problems such as high development threshold, inefficiency and high cost existing in mainstream software development methods.

[0008] To achieve the above purpose, the present invention provides a software development method based on natural language orchestration, including the following steps:

[0009] S1. Deploy a programmable system based on natural language orchestration, register basic resources in the programmable system, and store the registered resources and system resources in a relational database after natural language description and synchronize them to a vector database;

[0010] S2. Create a business application in the programmable system, perform orchestration, and then publish the orchestration content as an agent in the relational database and synchronize it to the vector database after natural language description;

[0011] S3. Use a large language model to perform vector retrieval on the user's question content and select the most suitable agent;

[0012] S4. Invoke the most suitable agent to execute the corresponding orchestration content in paragraph order and output the final result.

[0013] Optionally, in S1, the registration of the basic resources includes interface resource registration and database resource publication.

[0014] Optionally, in S1, first convert the registered resources and system resources after natural language description into SQL statements, and then publish the SQL statements as standard restFul interfaces and store them in the relational database and synchronize them to the vector database.

[0015] Optionally, the steps of S3 specifically include:

[0016] Use the large language model to perform vector retrieval on the question content and find multiple agents;

[0017] Put the multiple agents into the parameter block of the Function calling module of the large language model, and the large language model selects the most suitable agent.

[0018] Optionally, the steps of S4 specifically include:

[0019] S41. Invoke the most suitable agent to perform vector retrieval on the content of each paragraph in the paragraph order and find multiple interface resources;

[0020] S42. Put the multiple interface resources into the parameter block of the Function calling module, and the large language model selects the most suitable interface resource;

[0021] S43. Extract parameters from the question content and the return content of the previous paragraphs, and the large language model uses the extracted parameters to call the most suitable interface resource;

[0022] S44. Repeat S41 - S43 until all paragraphs are executed;

[0023] S45. Output the final result.

[0024] Based on the same inventive concept, the present invention also provides a programmable system based on natural language orchestration, including:

[0025] A data management module for registering basic resources;

[0026] A resource management module for storing the registered resources and system resources in a relational database after natural language description and synchronizing them to a vector database;

[0027] A human-computer interaction interface for interacting with users and creating business applications;

[0028] An orchestration module for orchestrating the business application, and then storing the orchestrated content in the relational database as an agent after natural language description and synchronizing it to the vector database;

[0029] A large language model for performing vector retrieval on the user's question content, selecting the most suitable agent, and calling the most suitable agent to execute the corresponding orchestrated content in paragraph order to output the final result.

[0030] Optionally, the programmable system further includes a text2sql module, which is used to convert the registered resources and system resources after natural language description into sql statements, and publish the sql statements as standard restFul interfaces and then store them in the relational database and synchronize them to the vector database.

[0031] Optionally, the large language model includes a Function calling module, and the large language model is specifically used for:

[0032] Performing vector retrieval on the question content to find multiple agents, and putting the multiple agents into the parameter block of the Function calling module to select the most suitable agent for calling and execution.

[0033] Optionally, the large language model is further specifically used for:

[0034] Calling the most suitable agent to perform vector retrieval on the content of each paragraph in the paragraph order to find multiple interface resources;

[0035] Putting the multiple interface resources into the parameter block of the Function calling module to select the most suitable interface resource;

[0036] Extracting parameters from the question content and the return content of the previous paragraphs, and using the extracted parameters to call the most suitable interface resource until all paragraphs are executed, and then outputting the final result to the human-computer interaction interface.

[0037] Based on the same inventive concept, the present invention also provides a readable storage medium, on which a computer program is stored, and when the computer program is executed, it can implement the software development method based on natural language arrangement as described above.

[0038] In the software development method, system and readable storage medium based on natural language arrangement provided by the present invention, users do not need to understand the technical principles. The business requirements are disassembled through natural language, and after forming detailed steps, they are executed by a large language model. Through this method, natural language arrangement is realized, which can lower the threshold, enable users to complete software development by themselves, reduce communication time consumption and reduce the possibility of requirement deviation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Those of ordinary skill in the art should understand that the provided drawings are used to better understand the present invention and do not constitute any limitation to the scope of the present invention. Among them:

[0040] Figure 1 is the overall flowchart of the software development method based on natural language arrangement provided by an embodiment of the present invention;

[0041] Figure 2 is the flowchart of application creation provided by an embodiment of the present invention;

[0042] Figure 3 is the flowchart of application invocation provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] To make the objectives, advantages and features of the present invention clearer, the following further describes the present invention in detail with reference to the drawings and specific embodiments. It should be noted that the drawings are in a very simplified form and are all drawn using non-precise scales, only for the purpose of facilitating and clearly assisting in explaining the embodiments of the present invention. To make the objectives, features and advantages of the present invention more obvious and understandable, please refer to the drawings. It should be noted that the structures, scales, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those skilled in this technology to understand and read, and are not used to limit the limited conditions for implementing the present invention. Any modification of the structure, change of the proportional relationship or adjustment of the size, in the case of being the same or similar to the effects that the present invention can produce and the objectives that can be achieved, should still fall within the scope covered by the technical content disclosed by the present invention.

[0044] As used in the present invention, the singular forms "a", "an" and "the" include plural objects unless the context clearly indicates otherwise. As used in the present invention, the term "or" is generally used in the sense of including "and / or" unless the context clearly indicates otherwise.

[0045] Please refer to Figures 1 - 3, this embodiment provides a software development method based on natural language orchestration, including the following steps:

[0046] S1. Deploy a programmable system based on natural language orchestration, register basic resources in the programmable system, and after natural language description of the registered resources and system resources, store them in a relational database and synchronize them to a vector database;

[0047] S2. Create a business application in the programmable system and perform orchestration, and then after natural language description of the orchestration content, publish it as an agent and store it in the relational database and synchronize it to the vector database;

[0048] S3. Use a large language model to perform vector retrieval on the user's question content and select the most suitable agent;

[0049] S4. Invoke the most suitable agent to execute the corresponding orchestration content in paragraph order and output the final result.

[0050] It should be noted that natural language orchestration refers to describing specific work content through language, literalizing the communication and guidance methods between humans, and disassembling complex tasks, just as detailed as the step descriptions in various specification operation manuals. Before the application of the software development method based on natural language orchestration provided by the present invention, software development usually required professional programmers to implement. From the proposal of requirements to communication and then to coding and deployment, it required a lot of time and effort.

[0051] From a macro perspective, software products are all about data processing and display. In order to lower the development threshold and reduce communication time consumption, enabling various users to quickly develop application services, the present invention provides a software development method based on natural language orchestration, enabling users to disassemble business requirements through natural language without understanding technical principles, forming detailed steps and then having a large language model complete the execution. Through this method, natural language orchestration is achieved, which can lower the threshold, enable users to complete software development by themselves, reduce communication time consumption, and reduce the possibility of requirement deviation.

[0052] First, execute step S1, deploy a programmable system based on natural language orchestration, register basic resources in the programmable system, and after natural language description of the registered resources and system resources, store them in a relational database and synchronize them to a vector database.

[0053] In this embodiment, the registration of the basic resources includes the registration of interface resources and the publication of database resources. For example, the required standard restFul interface information and data tables and fields in different databases can be described in natural language, stored in the relational database, and synchronized to the vector database.

[0054] In this embodiment, the system resources are basic resources that come with the system, such as a knowledge base, which can support the input and vector retrieval of documents and question-answer pairs for user arrangement and use.

[0055] Preferably, after the registration resources and system resources are described in natural language and before they are stored in the relationship library and synchronized to the vector library, the registration resources and system resources after the natural language description are first converted into SQL statements, and then the SQL statements are published as standard restFul interfaces and then stored in the relationship library and synchronized to the vector library to facilitate subsequent large language model calls.

[0056] Then, S2 is executed to create and orchestrate a business application in the programmable system, and then the orchestration content is described in natural language and published as an intelligent agent, stored in the relationship library, and synchronized to the vector library.

[0057] In this embodiment, the orchestration can be managed by "agents". After creating an agent, the business content can be orchestrated under the agent. When arranging, the business steps are described in rows. Each step needs to have clear logic, that is, it is clear that resources need to be called or data processing requirements are clearly described. After the arranging is completed, the agent is released to the market, and the front-end user can call the agent through human-computer interaction.

[0058] Then execute S3, such as Figure 3 As shown, the large language model is used to perform vector retrieval on the user's question content and select the most suitable agent. In this embodiment, the step S3 specifically includes:

[0059] Using the large language model to perform vector retrieval on the question content to find multiple agents;

[0060] The multiple agents are placed into the parameter block of the Function calling module of the large language model, and the large language model selects the most suitable agent.

[0061] The Function Calling module is used by the large language model to convert the text instructions output by the model into specific function calling functions during the interaction process. The Function calling module is configured as a json string, which is mainly configured in the tools attribute, mainly including the function name, description, parameters, prompt words, etc. When the large language model interacts with the user, it will identify the intention in the user's question content and select the matching function for call execution according to the intention.

[0062] In this embodiment, function calls are implemented in combination with vector retrieval. When communicating with the user, first match the resource description and call information in the vector library, then combine the matched description information into the parameter block content of the function calling module, and then the large language model is called to execute. With such a configuration, it is more accurate than directly putting all resources into the parameter block of the function calling module, can reduce the number of parameters, and improve the recognition effect and response speed of the model.

[0063] Finally, execute S4, call the most suitable intelligent agent to execute the corresponding choreographed content in paragraph order, and output the final result. In this embodiment, please continue to refer to Figure 3 , the steps of S4 specifically include:

[0064] S41. Call the most suitable intelligent agent to perform vector retrieval on the content of each paragraph in the paragraph order to find multiple interface resources;

[0065] S42. Put the multiple interface resources into the parameter block of the Function calling module, and the large language model selects the most suitable interface resource;

[0066] S43. Extract parameters from the question content and the return content of the previous paragraphs, and the large language model uses the extracted parameters to call the most suitable interface resource;

[0067] S44. Repeat S41 - S43 until all paragraphs are executed;

[0068] S45. Output the final result.

[0069] In some specific embodiments, when a paragraph is executed, the function calling module is used to match the user's question content in the vector library to query the interfaces that may be used. To ensure accuracy, each paragraph description is matched with at least 5 pieces of data in the vector library, and there is no similarity limit. Then, the restFul interfaces matched in the vector library are put into the parameter block of the function-calling module, and the large language model selects the most suitable interface resource.

[0070] The same method is adopted when executing the next paragraph, but the execution result of the previous paragraph is added to the conversation context. Repeat S41 - S43 until all paragraphs are executed, and return the execution result of the last paragraph. Through this method, the assembly of services can be realized in natural language, and other types of service access can be extended in the future.

[0071] Based on the same inventive concept, an embodiment of the present invention also proposes a programmable system based on natural language choreography, including:

[0072] Data management module, used to register basic resources;

[0073] Resource management module, used to describe registered resources and system resources in natural language, store them in the relationship library and synchronize them to the vector library;

[0074] Human-computer interaction interface, used to interact with users and create business applications;

[0075] An orchestration module, used to orchestrate the business application, and then publish the orchestration content as an intelligent body after describing it in natural language, store it in the relationship library, and synchronize it to the vector library;

[0076] The large language model is used to perform vector retrieval on the user's question content, select the most suitable intelligent agent, and call the most suitable intelligent agent to execute the corresponding arrangement content in paragraph order to output the final result.

[0077] In this embodiment, the data management module can implement different codes for different databases based on J2EE interfaces and abstract concepts to support access to multiple databases. Then, the table and field are described through the data dictionary to complete the preliminary preparation for data query.

[0078] In this embodiment, the registration of the basic resources includes interface resource registration and database resource release. The system resources are basic resources that come with the system, such as a knowledge base. The resource management module manages all accessed resources, text2sql resources, and system resources. The accessed resources are described in natural language and stored in the vector library. When used, information is provided through a vector retrieval service.

[0079] In this embodiment, the human-computer interaction interface may adopt a dialog interaction mode to render different results in the interaction interface, such as displaying statistical data in a table or bar chart after querying the database.

[0080] In this embodiment, the orchestration module can be managed by "agents". After creating an agent, business content can be orchestrated under the agent. When arranging, the business steps are described in rows. Each step needs to have clear logic, that is, it is clear that resources need to be called or data processing requirements are clearly described. After the arranging is completed, the agent is released to the market, and the front-end user can call the agent through the human-computer interaction interface.

[0081] Preferably, the programmable system also includes a text2sql module, which is used to convert the registration resources and system resources after the natural language description into SQL statements, and publish the SQL statements into a standard restFul interface and then store them in the relationship library and synchronize them to the vector library.

[0082] In this embodiment, the Text2sql module is based on the data data source access and data dictionary of the data management module. Users can put forward data query requirements through dialogue. After the large language model conducts reasoning and analysis, sql statements are generated based on the data dictionary, and then the data management module completes the execution.

[0083] When calling the text2sql module, the user's question is retrieved in the data table vector library to find the corresponding table information, so as to determine the general range of the user's table. Then, according to the user's question, the table and field information are retrieved in the data field vector library. Through the reasoning ability of the large language model, all the retrieved table and field information is put into the context, and the large language model generates sql statements, and the generated sql statements are put into the corresponding data source for execution and the results are returned. After the execution of the sql is confirmed to be correct, the standard restFul interface can be published with this sql.

[0084] In this embodiment, by combining the text2sql module with vector retrieval technology, the table information and field information are respectively stored in the vector library. When in use, the user's dialogue is first vector-matched with the table information vector, and then the dialogue is vector-matched with the table name and field information, and a set of table and field descriptions that more meet the user's needs can be obtained.

[0085] In this embodiment, the large language model includes a Function calling module, and the large language model is specifically used for:

[0086] Perform vector retrieval on the question content, find multiple agents, and put the multiple agents into the parameter block of the Function calling module to select the most suitable agent for calling and execution.

[0087] In this embodiment, the large language model is also specifically used for:

[0088] Call the most suitable agent to perform vector retrieval on the content of each paragraph in the order of the paragraph to find multiple interface resources;

[0089] Put the multiple interface resources into the parameter block of the Function calling module to select the most suitable interface resource;

[0090] Extract parameters from the question content and the return content of the previous paragraphs, and use the extracted parameters to call the most suitable interface resource until all paragraphs are executed, and then output the final result to the human-computer interaction interface.

[0091] In some specific embodiments, when a paragraph is executed, the function calling module is used to match the content of the user's question in the vector library to query the interfaces that may be used. To ensure accuracy, each paragraph description is matched with at least 5 data in the vector library, and there is no similarity limit. Then, the restFul interfaces matched in the vector library are placed in the parameter block of the function-calling module, and the large language model selects the most suitable interface resource.

[0092] The same method is adopted when the next paragraph is executed, but the execution result of the previous paragraph is added to the conversation context. This is repeated until all paragraphs are executed, and the execution result of the last paragraph is returned. Through this method, the assembly of services can be realized in natural language, and other types of service access can be extended subsequently.

[0093] Based on the same inventive concept, an embodiment of the present invention also provides a readable storage medium, on which a computer program is stored. When the computer program is executed, it can implement the software development method based on natural language orchestration as described above.

[0094] A readable storage medium can be a tangible device that can hold and store instructions used by an instruction execution device. For example, it can be, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punched card or raised structures in grooves storing instructions thereon, and any suitable combination of the above. The computer programs described herein can be downloaded from the readable storage medium to various computing / processing devices, or downloaded to an external computer or external storage device through a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter or network interface in each computing / processing device receives the computer program from the network and forwards the computer program for storage in the readable storage medium in each computing / processing device. The computer program for performing the operations of the present invention can be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages - such as Smalltalk, C++, etc., and conventional procedural programming languages - such as the "C" language or similar programming languages. The computer program can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., using an Internet service provider to connect through the Internet). In some embodiments, by using the state information of the computer program to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute computer-readable program instructions to implement various aspects of the present invention.

[0095] Aspects of the present invention are described herein with reference to the flowcharts and / or block diagrams of methods, systems, and computer program products according to embodiments of the present invention. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer programs. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, thereby producing a machine such that when these programs are executed by the processor of the computer or other programmable data processing apparatus, a device is produced that implements the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams. These computer programs can also be stored in a readable storage medium, and these computer programs cause a computer, a programmable data processing apparatus, and / or other devices to work in a specific manner. Thus, the readable storage medium storing the computer programs includes a manufactured article that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams.

[0096] The computer programs can also be loaded onto a computer, other programmable data processing apparatus, or other devices, such that a series of operation steps are executed on the computer, other programmable data processing apparatus, or other devices to produce a computer-implemented process, so that the computer programs executed on the computer, other programmable data processing apparatus, or other devices implement the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams.

[0097] In summary, embodiments of the present invention provide a software development method, system, and readable storage medium based on natural language orchestration, enabling users to disassemble business requirements through natural language without understanding technical principles, form detailed steps, and then have a large language model execute them. Through this method, natural language orchestration is achieved, which can lower the threshold, allowing users to complete software development by themselves, reducing communication time consumption, and lowering the possibility of requirement deviation.

[0098] In addition, it should also be recognized that although the present invention has been disclosed above with preferred embodiments, the above embodiments are not intended to limit the present invention. For any person skilled in the art, without departing from the scope of the technical solution of the present invention, many possible changes and modifications can be made to the technical solution of the present invention by using the disclosed technical content above, or modified into equivalent embodiments with equivalent changes. Therefore, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A software development method based on natural language orchestration, characterized in that, It includes the following steps: S1. Deploy a programmable system based on natural language orchestration, register basic resources in the programmable system, and after natural language description of the registered resources and system resources, store them in a relational database and synchronize them to a vector database; S2. Create a business application in the programmable system and perform orchestration, and then after natural language description of the orchestration content, publish it as an agent and store it in the relational database and synchronize it to the vector database; S3. Use a large language model to perform vector retrieval on the user's question content and select the most suitable agent; S4. Invoke the most suitable agent to execute the corresponding orchestration content in paragraph order and output the final result.

2. The software development method based on natural language arrangement according to claim 1, characterized in that, In S1, the registration of the basic resources includes interface resource registration and database resource publication.

3. The software development method based on natural language layout according to claim 1, characterized in that In S1, first convert the registered resources and system resources after the natural language description into SQL statements, and then publish the SQL statements as standard RESTful interfaces and then store them in the relational database and synchronize them to the vector database.

4. The software development method based on natural language arrangement according to claim 1, characterized in that The steps of S3 specifically include: Use the large language model to perform vector retrieval on the question content and find multiple agents; Put the multiple agents into the parameter block of the Function calling module of the large language model, and the large language model selects the most suitable agent.

5. The software development method based on natural language layout according to claim 4, characterized in that The steps of S4 specifically include: S41. Invoke the most suitable agent to perform vector retrieval on the content of each paragraph in the paragraph order and find multiple interface resources; S42. Put the multiple interface resources into the parameter block of the Function calling module, and the large language model selects the most suitable interface resource; S43. Extract parameters from the question content and the return content of the previous paragraphs, and the large language model uses the extracted parameters to invoke the most suitable interface resource; S44. Repeat S41 - S43 until all paragraphs are executed; S45. Output the final result.

6. A programmable system based on natural language orchestration, characterized in that, It includes: A data management module for registering basic resources; A resource management module for storing the registered resources and system resources in a relational database after natural language description and synchronizing them to a vector database; A human - machine interaction interface for interacting with users and creating business applications; An orchestration module for orchestrating the business application, and then after natural language description of the orchestration content, publishing it as an agent and storing it in the relational database and synchronizing it to the vector database; A large language model for performing vector retrieval on the user's question content, selecting the most suitable agent, and invoking the most suitable agent to execute the corresponding orchestration content in paragraph order and output the final result.

7. The programmable system based on natural language arrangement according to claim 6, wherein The programmable system further includes a text2sql module, and the text2sql module is used to convert the registered resources and system resources after the natural language description into SQL statements, and publish the SQL statements as standard RESTful interfaces and then store them in the relational database and synchronize them to the vector database.

8. The programmable system based on natural language arrangement according to claim 6, characterized in that, The large language model includes a Function calling module, and the large language model is specifically configured to: Perform vector retrieval on the question content, find multiple agents, and place the multiple agents into the parameter block of the Function calling module to select the most suitable agent for invocation and execution.

9. The programmable system based on natural language arrangement according to claim 8, characterized in that, The large language model is specifically further configured to: Invoke the most suitable agent to perform vector retrieval on the content of each paragraph in the paragraph order to find multiple interface resources; Place the multiple interface resources into the parameter block of the Function calling module to select the most suitable interface resource; Extract parameters from the question content and the return content of the previous paragraphs, and use the extracted parameters to call the most suitable interface resource until all paragraphs are executed, and then output the final result to the human-computer interaction interface.

10. A readable storage medium, on which a computer program is stored, characterized in that, When the computer program is executed, it can implement the software development method based on natural language arrangement according to any one of claims 1-5.

Images