Method for multi-al coordination and aggregation scheduling and full-automatic execution based on large model

Abstract

This invention discloses a multi-AI collaborative aggregation scheduling and fully automated execution method based on a large model, belonging to the fields of artificial intelligence technology and task automation processing technology. This method receives user natural language requests, utilizes a large model for intent recognition, task decomposition and structured orchestration, constructs a multi-AI capability matching library to automatically match and adapt AI tools, generates standardized task execution links, and schedules multi-AI collaborative work through a unified interface gateway. Simultaneously, it automatically completes third-party platform interface calls, resource configuration, task iteration optimization, and result integration and output, eliminating the need for users to manually operate multiple AI tools or perform cumbersome configuration steps. This invention solves the technical problems of existing technologies where users need to download, switch, and operate multiple AI tools separately, resulting in high technical barriers, cumbersome operations, and the inability of multiple AIs to work collaboratively. It achieves a one-click trigger and fully automated task processing effect, significantly reducing user operating costs and improving the execution efficiency of complex tasks. It is suitable for one-stop AI processing of various complex tasks such as APP development, content generation, and platform configuration.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method for multi-AI collaborative aggregation scheduling and fully automated execution based on a large model, which falls under the category of AI task automation processing and intelligent scheduling technology. Background Technology

[0002] With the rapid development of artificial intelligence technology, various AI tools have emerged, covering multiple fields such as code development, design creation, content generation, and platform configuration. When dealing with complex tasks (such as batch development of apps or generation of multiple types of content), users often need to use multiple different AI tools at the same time.

[0003] In existing technologies, users need to download, install, register accounts, configure interfaces, manually input requirements, and switch between multiple AI tools one by one, which has the following technical drawbacks: 1. The operation process is cumbersome, requiring a lot of time to switch tools and configure parameters, with a high technical threshold, making it impossible for users without professional background to complete complex tasks; 2. Data cannot be shared between different AI tools, making collaborative work impossible and resulting in low task execution efficiency; 3. There is a lack of a unified task scheduling mechanism, which cannot automatically match the optimal AI combination according to user needs, requiring users to select manually, resulting in poor adaptability; 4. Auxiliary operations such as interface debugging and resource configuration during task execution need to be completed manually by users, further increasing the operating cost.

[0004] Currently, there is no existing technology that can achieve unified scheduling of multiple AI tools, fully automated collaborative execution, and one-click triggering by users to complete the entire process of tasks. Therefore, developing an AI execution method that can aggregate multiple AI capabilities, automatically schedule and coordinate, and automate the entire process has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a multi-AI collaborative aggregation scheduling and fully automated execution method based on a large model. The aim is to solve the technical problems of existing multi-AI tools being cumbersome to use, unable to work collaboratively, having high technical barriers for users, and low efficiency in executing complex tasks, thereby achieving a task processing effect that can be triggered by a single click and automated throughout the entire process.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for multi-AI collaborative aggregation scheduling and fully automated execution based on a large model includes the following steps: 1. Request Reception and Parsing: The system receives user-inputted natural language task requests via the terminal interface. A pre-trained large language model is used to perform intent recognition, entity extraction, and task decomposition on these requests. This transforms the unstructured natural language requests into structured task instructions containing a list of subtasks, subtask dependencies, execution priorities, and task objective parameters. In this step, the large language model uses either open-source or commercial models and is optimized through a dedicated prompt project to achieve accurate decomposition of complex tasks and adapt to the fuzzy natural language descriptions of non-professional users.

[0007] 2. AI Capability Matching and Filtering: A pre-built multi-AI capability matching library is retrieved. This library stores data on the function types, interface parameters, calling permissions, applicable scenarios, and execution efficiency of various AI tools. Based on the sub-task types in the structured task instructions, the optimal AI tool combination for each sub-task is automatically matched and filtered, establishing a one-to-one correspondence between sub-tasks and AI tools. The matching library supports real-time updates, allowing for dynamic addition or removal of AI tools to ensure matching accuracy.

[0008] 3. Task Link Orchestration: Based on the subtask dependencies and execution priorities, a topology sorting algorithm is used to generate a task execution link that combines serial and parallel processing. The task execution link is then serialized and encapsulated to generate standardized task data packages that are compatible with various AI tool interfaces, thus solving the problem of inconsistent parameters across different AI interfaces.

[0009] 4. Multi-AI Collaborative Scheduling: Through a unified API gateway interface, standardized task data packets are distributed to the corresponding matching AI tools. Each AI tool is controlled to execute sub-tasks sequentially or in parallel according to the task execution chain. The execution progress and intermediate results of each AI tool are obtained in real time. At the same time, an abnormal retry mechanism is set up to ensure stable task execution.

[0010] 5. Automated Processing: For interface configuration, resource application, platform integration, and result verification during task execution, the automated execution module automatically completes parameter configuration, interface debugging, and result verification within the scope of user authorization and official open APIs of third-party platforms. All operations are completed based on the official open APIs of third-party platforms and user authorization, without the need for manual user intervention.

[0011] 6. Results Integration and Output: The intermediate results output by various AI tools are integrated, verified, and standardized in format to generate the final task results and feed them back to the user terminal. At the same time, a task execution log is generated, which includes the execution steps, execution time, and the participation of each AI tool, so that users can view and trace the results.

[0012] Furthermore, this method also supports iterative optimization of requirements. Users can submit modification instructions for the output results, and the system will re-parse the modification requirements, adjust the task chain, and execute it again until the user's requirements are met. Beneficial effects

[0013] 1. Through large models, accurate parsing of natural language requirements and task decomposition are achieved. Users do not need to have professional technical knowledge. They only need to input simple requirements to complete complex tasks, which greatly reduces the user threshold. 2. A multi-AI capability matching library and unified scheduling mechanism were built to automatically match the optimal AI combination and realize the collaborative operation of multiple AI tools, solving the technical problem of data interoperability and collaborative execution of different AI tools; 3. Fully automated execution: Automatically completes interface configuration, task scheduling, result integration, and other operations. Users only need to trigger it with one click, without having to manually switch or operate multiple AI tools, which greatly simplifies the operation process and improves task execution efficiency. 4. An exception handling and iterative optimization mechanism is set up, which ensures high task execution stability and allows for real-time adjustments based on user feedback to adapt to the personalized needs of different users. 5. All automated operations are based on official APIs and user authorization, ensuring strong compliance and making them widely applicable to various complex scenarios such as APP development, content creation, and platform configuration. Detailed Implementation

[0014] User requirements: Generate 50 lightweight apps in batches, including basic interfaces, functional code, and deployment configurations. No professional development background is required, and no manual operation of any AI tools or platforms is necessary.

[0015] 1. Requirements Reception and Parsing: When a user enters "Generate 50 lightweight apps in batches, including basic functions such as login, list, and details, and complete the code writing and deployment without my manual operation" in the terminal, the system parses this natural language requirement through a large model and breaks it down into: Subtask 1 (App Requirements Analysis), Subtask 2 (Interface Prototype Design), Subtask 3 (UI Design), Subtask 4 (Front-end Code Generation), Subtask 5 (Back-end Code Generation), Subtask 6 (Code Packaging), and Subtask 7 (Platform Deployment), clarifying the dependencies and execution priorities of each subtask.

[0016] 2. AI Capability Matching and Filtering: The system retrieves multiple AI capability matching libraries to match the AI ​​corresponding to subtask 1 (requirements analysis), subtask 2 (prototype design), subtask 3 (UI generation), subtask 4 (front-end development), subtask 5 (back-end development), subtask 6 (code packaging), and subtask 7 (cloud platform deployment), thus establishing the correspondence between subtasks and AI tools.

[0017] 3. Task chain orchestration: A topology sorting algorithm is used to generate a serial chain of "requirements analysis → prototype design → UI design → front-end and back-end code generation → code packaging → platform deployment". At the same time, task logic for 50 apps to be executed in parallel is set up to generate standardized task data packages.

[0018] 4. Multi-AI Collaborative Scheduling: Through a unified API gateway, task data packets are distributed to the corresponding AI tools, and the AI ​​tools are controlled to execute sub-tasks in sequence. The execution progress is obtained in real time. If a single AI fails to execute, it will automatically retry twice, and the progress will be synchronized to the user terminal at the same time.

[0019] 5. Automated Processing: The automation module is based on the official interfaces of various AI and the open APIs of the cloud platform, and automatically completes operations such as interface calls, parameter configuration, deployment and debugging. The entire process does not require users to manually register, pay or configure, and all operations are completed within the scope of user authorization.

[0020] 6. Results Integration and Output: The system integrates the prototype files, design drawings, code packages, and deployment links output by each AI to generate a complete result package of 50 apps, which is then pushed to the user terminal, while generating a complete execution log.

[0021] If a user requests modifications, such as adjusting the UI style, the system receives the modification instructions, re-parses the requirements, adjusts the task chain, and re-schedules the AI ​​tool to execute the modifications until the user is satisfied.

Claims

1. A method for multi-AI collaborative aggregation scheduling and fully automated execution based on a large model, characterized in that, Includes the following steps: (1) Request reception and parsing: Receive natural language task requests input by users through the terminal interface, and use a pre-trained large language model to perform intent recognition, entity extraction and task decomposition on the natural language task requests, transforming unstructured natural language requests into structured task instructions containing a list of sub-tasks, sub-task dependencies, execution priorities and task target parameters; (2) AI capability matching and screening: Retrieve a pre-built multi-AI capability matching library, which stores the function types, interface parameters, calling permissions, applicable scenarios and execution efficiency data of various AI tools. Based on the sub-task types in the structured task instructions, automatically match and screen the optimal AI tool combination for the corresponding sub-tasks, and establish a one-to-one correspondence between sub-tasks and AI tools; (3) Task link orchestration: Based on the sub-task dependencies and execution priorities, use a topology sorting algorithm to generate a task execution link that combines serial and parallel processing, serialize and encapsulate the task execution link, and generate a standardized task data package that is compatible with the interfaces of each AI tool; (4) Multi-AI collaborative scheduling: Distribute the standardized task data package to the corresponding matched AI through a unified API gateway interface. Tools, control each AI tool to execute sub-tasks sequentially or in parallel according to the task execution link, and obtain the execution progress and intermediate results of each AI tool in real time; (5) Automated process processing: for interface configuration, resource application, platform docking, and result verification operations in the task execution process, the automated execution module automatically completes parameter configuration, interface debugging and result verification within the scope of user authorization and official open API of third-party platforms, without the need for manual intervention by the user; (6) Result integration and output: integrate, verify and standardize the intermediate results output by each AI tool, generate the final task results and feed them back to the user terminal, and generate task execution logs for the user to view.

2. The method according to claim 1, characterized in that, The construction method of the multi-AI capability matching library mentioned in step (2) is as follows: collect the official interface documents and function parameters of various publicly available AI tools, classify the function types in a normalized manner, establish the mapping relationship between function tags and AI tools, and update the calling status and interface availability data of AI tools in real time.

3. The method according to claim 1, characterized in that, In the multi-AI collaborative scheduling process described in step (4), an abnormal retry mechanism and a progress feedback mechanism are set up. If a single AI tool fails to execute, it will automatically retry according to a preset number of times, and at the same time, the real-time execution progress will be synchronized to the user terminal.

4. The method according to claim 1, characterized in that, The automated process described in step (5) is executed only based on the official open API of the third-party platform, including interface calls, parameter configuration, and resource deployment operations after user authorization. All operations are executed within the scope of user authorization and the rules of the open interface of the third-party platform.

5. The method according to claim 1, characterized in that, It also includes a requirement iteration and optimization step: receiving user modification instructions for the final task results, feeding the modification instructions back to the large model parsing module, readjusting the structured task instructions and task execution links, and triggering multi-AI collaborative execution again until the user's needs are met.