Natural language-driven agent creation method and system

CN120909636AInactive Publication Date: 2025-11-07SHENZHEN PEMI TECHNOLOGY CO LTD
-1 Cites -1 Cited by

Patent Information

Application Number
CN202511060464.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-30
Publication Date
2025-11-07
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

When dealing with complex user expressions, existing technologies generate agents with chaotic behavior transitions and component adaptation failures, affecting system stability and scalability.

Method used

By acquiring the task description information input by the user, performing text standardization and semantic parsing, extracting task target information, behavioral intent information and constraint information, constructing functional module information, and based on this, searching and fitting the functional component library to generate a structured execution graph, and finally automatically generating an executable script and task description file.

Benefits of technology

It achieves fully automated construction from natural language input to agent deployment, improving construction efficiency and the stability of task adaptability, and improving the problems of chaotic behavior transitions and component adaptation failures of agents when expressing complex expressions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120909636A_ABST
    Figure CN120909636A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of agents, and provides a natural language driven agent creation method and system.The method comprises the steps that after task description information input by a user is obtained, text standardization is conducted, and constraint condition information, a candidate task type set and a functional behavior set are obtained; and screening the candidate task type set and the function behavior set by using the constraint condition information, after obtaining function module information, performing retrieval and fitting, after obtaining an execution graph, performing structured labeling and intermediate code generation, and obtaining an executable script and a task description file for agent deployment. By automatically generating the deployable executable script and the task description file, automatic construction of the whole process from natural language input to agent deployment is realized, the construction efficiency and the stability of the task adaptation capability are improved, and the situation that the task adaptation capability is more stable when facing complex user expression is improved. And the generated intelligent agent has the problems of behavior jump chaos and component adaptation failure.
Need to check novelty before this filing date? Find Prior Art