Method, device and equipment for dynamically managing and controlling capability of enterprise business intelligence agent, and medium

Abstract

This application discloses a method, apparatus, device, and medium for dynamic management and control of the capabilities of enterprise business intelligent agents, relating to the field of intelligent agent management and control technology. The method includes: after the target enterprise business intelligent agent receives a target enterprise business task to be processed, performing structured processing on the enterprise business task information to obtain structured use cases; filtering several pre-defined planning domains based on the structured use cases to determine a matching target planning domain; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business intelligent agent in different enterprise business tasks; activating the defined target tool capabilities within the intelligent agent execution context corresponding to the target planning domain, so that the target enterprise business intelligent agent can use the target tool capabilities to execute the target enterprise business task; continuously monitoring the capability invocation behavior of the target enterprise business intelligent agent during task execution, and triggering corresponding exception handling operations when abnormal capability invocation behavior is detected.

Description

Technical Field

[0001] This invention relates to the field of intelligent agent management technology, and in particular to a method, apparatus, equipment and medium for dynamic management of the capabilities of enterprise business intelligent agents. Background Technology

[0002] With the widespread application of large-scale modeling technology and Agentic AI in enterprise-level scenarios, intelligent agents are gradually evolving into enterprise-level execution entities capable of directly participating in business operations, data processing, and process control. Intelligent agents need to connect to various business systems, database interfaces, and automation tools, and their behavior directly impacts business outcomes. Therefore, managing the execution capabilities of intelligent agents has become a pressing technical problem. In existing enterprise-level intelligent agent systems, enterprise business intelligent agents are typically equipped with a large number of executable capabilities (Tools) during the initialization phase, and these are permanently invoked as inherent attributes of the agent. In complex business environments, static binding can easily lead to capability abuse, permission violations, and task interference. In addition, prompt words or rule tables can be used to control tool usage, but relying on model inference, in complex scenarios involving cross-system and multi-task parallel execution, enterprise business intelligent agents may invoke capabilities across tasks, resulting in unexpected operations or system state anomalies, leading to unpredictable execution results. Furthermore, adding new tools or functions may affect existing task behavior, making it difficult to assess and control the overall system risk, and compromising the security of capability evolution.

[0003] In summary, optimizing the capability management methods of enterprise business intelligence agents to ensure the security and controllability of agent behavior and improve system stability are urgent problems that need to be solved. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a method, apparatus, device, and medium for dynamic management of the capabilities of enterprise business intelligent agents, which can optimize the capability management method of enterprise business intelligent agents to ensure the security and controllability of intelligent agent behavior and improve system stability. The specific solution is as follows: Firstly, this application provides a method for dynamic management and control of the capabilities of an enterprise business intelligence agent, including: After the target enterprise business intelligence agent receives the target enterprise business task to be processed, it performs structured processing on the corresponding enterprise business task information to obtain the corresponding structured use cases. Based on the structured use cases, several pre-defined planning domains are filtered to determine the target planning domain that matches the target enterprise business task; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business agent in different enterprise business tasks. Within the agent execution context corresponding to the target planning domain, the defined target tool capabilities are activated so that the target enterprise business agent can utilize the target tool capabilities to execute the target enterprise business tasks. During the execution of the target enterprise's business tasks, the capability invocation behavior of the target enterprise's business intelligence agent is continuously monitored, and when an anomaly is detected in the capability invocation behavior, corresponding anomaly handling operations are triggered.

[0005] Optionally, the step of filtering a pre-defined number of planning domains based on the structured use cases to determine the target planning domain that matches the target enterprise's business tasks includes: Determine the constraint-related information and business execution objectives in the structured use cases; Based on the constraint-related information, availability filtering is performed on several pre-defined planning domains, and intent context matching is performed between the business execution target and the several planning domains to obtain several initial candidate planning domains. The activation risk and resource consumption cost of the initial candidate planning domains are evaluated, and the target candidate planning domains are obtained by screening the initial candidate planning domains based on the evaluation results. Among the target candidate planning domains, the planning domains that pass the compliance audit verification are determined as the target planning domains that match the business tasks of the target enterprise.

[0006] Optionally, activating the defined target tool capability within the agent execution context corresponding to the target planning domain includes: Determine the set of target tool capabilities and target behavior constraint rules defined in the target planning domain; Within the agent execution context corresponding to the target planning domain, the target tool capabilities in the target tool capability set are activated according to the structured use case and the target behavior constraint rules.

[0007] Optionally, the process of utilizing the target tool capabilities to execute the target enterprise's business tasks includes: If the execution context of the first agent corresponding to the current first target planning domain ends, the first target tool capability defined in the first target planning domain is unloaded, and a second target planning domain is determined; the second target planning domain is the target planning domain that matches the execution context of the corresponding second agent and the current agent execution context. Determine the second objective tool capability defined by the second objective planning domain, and activate the second objective tool capability.

[0008] Optionally, the unloading of the first target tool capability defined by the first target planning domain includes: Remove the first objective tool capability defined in the first objective planning domain from the current agent execution context; The current execution status of the tool capability and the cached temporary data generated during the invocation of the first target tool capability are cleaned up, and corresponding tool capability invocation logs are generated.

[0009] Optionally, after utilizing the target tool capabilities to execute the target enterprise business task, the method further includes: The execution of the target enterprise's business tasks is summarized, and the planning domains are adjusted and optimized based on the summary results.

[0010] Optionally, when an anomaly is detected in the capability invocation behavior, triggering a corresponding exception handling operation includes: If it is detected that the currently invoked tool capability is different from the target tool capability defined in the target planning domain, then the capability invocation behavior is determined to be abnormal; Uninstall the currently activated tool capability, or terminate the execution of the target enterprise business task, or roll back the target enterprise business task to the target state; the target state is the task state that was backed up in advance before the capability call behavior became abnormal.

[0011] Secondly, this application provides a device for dynamic management and control of the capabilities of an enterprise business intelligence agent, comprising: The information processing module is used to perform structured processing on the corresponding enterprise business task information after the target enterprise business intelligence agent receives the target enterprise business task to be processed, so as to obtain the corresponding structured use cases. The planning domain filtering module is used to filter several pre-defined planning domains based on the structured use cases to determine the target planning domain that matches the target enterprise business task; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business agent in different enterprise business tasks; The capability activation module is used to activate the defined target tool capabilities within the agent execution context corresponding to the target planning domain, so that the target enterprise business agent can use the target tool capabilities to execute the target enterprise business tasks. The operation triggering module is used to continuously monitor the capability invocation behavior of the target enterprise business agent during the execution of the target enterprise business task, and to trigger corresponding exception handling operations when an abnormality is detected in the capability invocation behavior.

[0012] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute the computer program to implement the aforementioned method for dynamic management and control of the capabilities of enterprise business intelligence agents.

[0013] Fourthly, this application provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned method for dynamic management and control of the capabilities of an enterprise business intelligence agent.

[0014] In this application, after the target enterprise business agent receives the target enterprise business task to be processed, it performs structured processing on the corresponding enterprise business task information to obtain corresponding structured use cases; based on the structured use cases, it filters several pre-defined planning domains to determine the target planning domain that matches the target enterprise business task; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business agent in different enterprise business tasks; within the agent execution context corresponding to the target planning domain, the defined target tool capabilities are activated so that the target enterprise business agent can use the target tool capabilities to execute the target enterprise business task; during the execution of the target enterprise business task, the capability invocation behavior of the target enterprise business agent is continuously monitored, and when an abnormality is detected in the capability invocation behavior, a corresponding exception handling operation is triggered. As can be seen from the above, after the target enterprise business intelligence agent receives the target enterprise business task to be processed, this application first performs structured processing on the business task information to generate corresponding structured use cases, and then selects a matching target planning domain from a number of preset planning domains based on the structured use cases. The planning domain is used to define the set of tool capabilities and behavioral constraint rules of the intelligence agent under different business tasks. Subsequently, the required target tool capabilities are activated in the execution context corresponding to the target planning domain to execute the target enterprise business task through the target tool capabilities. During the task execution process, the capability call behavior of the target enterprise business intelligence agent is continuously monitored, and if an anomaly is detected, the corresponding anomaly handling operation is triggered. In this way, through the process described above in this application, structured task processing, precise matching of planning domains, on-demand activation of tool capabilities, and full-process behavior monitoring achieve the standardization, controllability, and efficiency of the execution process of enterprise business intelligence agents. This ensures that the intelligence agents operate stably under preset constraints, while also promptly detecting and handling execution anomalies, significantly improving the reliability and security of automated execution of enterprise business. Dynamically activating capabilities and strictly binding them to contexts prevents the abuse of cross-task capabilities and unexpected operations. Adding new planning domains or tools will not interfere with the operation of existing planning domains, achieving capability isolation and secure iteration. This optimizes the capability management method of enterprise business intelligence agents to ensure the security and controllability of intelligence agent behavior and improve system stability. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 This application discloses a flowchart of a method for dynamic management and control of the capabilities of an enterprise business intelligence agent. Figure 2 This is a schematic diagram of the structure of a dynamic capability control device for an enterprise business intelligent agent disclosed in this application; Figure 3 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] In existing enterprise-level intelligent agent systems, enterprise business agents are typically equipped with a large number of executable capabilities (Tools) during the initialization phase, and these are permanently invoked as inherent attributes of the agent. In complex business environments, static binding can easily lead to capability abuse, permission violations, and task interference. In addition, prompt words or rule tables can be used to control tool usage, but relying on model inference, in complex scenarios involving cross-system and multi-task parallel execution, enterprise business agents may invoke capabilities across tasks, resulting in unexpected operations or system state anomalies, leading to unpredictable execution results. Furthermore, adding new tools or functions may affect existing task behavior, making it difficult to assess and control the overall system risk, and compromising the security of capability evolution.

[0019] To overcome the aforementioned technical problems, this application provides a method for dynamic control of the capabilities of enterprise business intelligent agents, which can optimize the capability management method of enterprise business intelligent agents to ensure the security and controllability of agent behavior and improve system stability.

[0020] See Figure 1 As shown in the figure, this invention discloses a method for dynamic management and control of the capabilities of an enterprise business intelligence agent, including: Step S11: After the target enterprise business intelligence agent receives the target enterprise business task to be processed, it performs structured processing on the corresponding enterprise business task information to obtain the corresponding structured use cases.

[0021] In this embodiment, after the target enterprise business intelligence agent receives the target enterprise business task to be processed, it performs structured processing on the enterprise business task information to generate corresponding structured use cases (User Cases).

[0022] It should be noted that the target enterprise business intelligence agent is an autonomous computing entity running in an enterprise-level system, capable of perceiving task context, executing planning decisions, and completing business operations. It performs tasks by accessing various business systems and tools, and its behavior is controlled by the planning domain, possessing a dynamic capability activation and deactivation mechanism. The structured use case is an abstract unit that structurally represents user input information and task requirements, including task objectives, constraints and priorities, contextual information (such as user permissions, system status, and business environment), risk level, and compliance requirements. It serves as the entry point for MoPD (Mixture of Planning Domains, a runtime capability management mechanism) capability activation, guiding planning domain selection, capability activation, and behavior review. The planning domain is the logical unit for the enterprise-level business intelligence agent to execute tasks. As the minimum governance and runtime container for capabilities, it defines the set of tool capabilities and behavioral constraint rules for the enterprise business intelligence agent in different enterprise business tasks. It not only defines the agent's behavioral space in a specific task but also clarifies the boundaries of capability application. It is understood that this application proposes a MoPD mechanism for on-demand activation of agent capabilities in a multi-planning domain environment. This mechanism enables dynamic activation, context binding, explicit unloading, lifecycle management, and behavior review of capabilities, forming a systematic review and reflection loop. This provides a secure, controllable, and auditable operating mechanism for enterprise-level agents. The core idea is that agents do not inherently possess capabilities; capabilities only exist temporarily within the selected planning domain context. MoPD includes the following characteristics: capabilities are dynamically activated based on task context and User Case; inactive planning domains and their tools are unavailable in the execution context; the capability lifecycle is bound to the planning domain, achieving an activation-execution-unloading closed loop; and it supports multi-task parallelism, capability isolation, risk control, and audit traceability. Thus, this embodiment, through standardized task parsing and structured transformation, allows agents to clearly understand task content and execution elements, providing a data foundation for subsequent task planning and execution.

[0023] Step S12: Based on the structured use cases, filter several pre-defined planning domains to determine the target planning domain that matches the target enterprise business task; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business agent in different enterprise business tasks.

[0024] In this embodiment, based on the structured use cases, a target planning domain matching the target enterprise's business tasks is selected from a pre-defined number of planning domains. This planning domain, serving as a capability carrier, defines the tool capabilities and behavioral constraints of the enterprise's business intelligence agent under different enterprise business tasks, clearly defining capability activation conditions, tool callability, and lifecycle, thus achieving dynamic capability management.

[0025] It should be noted that the planning domain specifically defines the agent's following in a particular task: GoalState: clearly defines the final goal and standard for task completion, as well as intermediate checkpoints; State Space: encompasses task-related data structures, object states, business states, and environmental variables; ActionModel: defines the sequence of operations that the agent can take and the corresponding triggering conditions; ConstraintModel: includes operation permissions, execution order restrictions, and system business strategy requirements; RiskModel: defines and assesses the risks that tool capabilities may cause, used to quantify the risks that each tool capability may generate; and ToolSet: a list of tools or interfaces that are allowed to be called within the planning domain context. Tools are business functional units or interfaces that the agent can call, including automation scripts, API (Application Programming Interface) calls, data processing modules, etc. Under the MoPD mechanism, they are not inherent attributes of the agent, but only exist temporarily when the planning domain is activated, and their lifecycle is strictly bound to the task or planning domain context.

[0026] Understandably, User Cases are used to determine which planning domains can be selected and activated, serving as a unified entry point for capability review and risk control. The system filters, sorts, and activates candidate planning domains based on User Cases, task intents, context, and risk assessment results for Meta-Planning. Meta-Planning is a key component of MoPD, enabling on-demand capability activation, task isolation, and risk control. During runtime, the system performs multi-dimensional evaluation and trade-offs on candidate planning domains, forming a dynamic, task-driven capability activation strategy: specifically, it determines constraint-related information and business execution objectives in the structured use cases; based on the constraint-related information, it performs availability filtering on several pre-defined planning domains and uses the business execution objectives to match the intent context of these planning domains, obtaining several initial candidate planning domains; it evaluates the activation risk and resource consumption cost of these initial candidate planning domains, and filters them based on the evaluation results to obtain target candidate planning domains; and it identifies the planning domains among the target candidate planning domains that pass compliance audit verification as the target planning domains that match the target enterprise's business tasks. The constraint-related information includes user permissions, object status, business constraints, and system policies. Specifically, first, constraint-related information and business execution objectives are determined based on the structured use cases. Then, based on the constraint-related information, availability filtering is performed on several pre-defined planning domains to ensure that only planning domains matching the task context are selected, preventing capability abuse or unauthorized calls. Simultaneously, intent context matching is performed using the business execution objectives, comparing user intent, task objectives, and planning domain objectives to ensure that the agent's execution behavior is highly consistent with actual business needs, resulting in initial candidate planning domains. This allows for effective isolation of different task capabilities in multi-task or asynchronous task environments. Next, the system activation risks and resource consumption costs that may be introduced by activating the planning domain are evaluated and filtered. Planning domains that may cause business conflicts or system anomalies are prioritized or rejected for activation. Finally, the planning domain behavior is checked to ensure compliance with enterprise policies, regulatory requirements, and internal audit standards for compliance audit verification. Planning domains that fail the review will be rejected for activation or transferred to manual approval. Planning domains that pass the compliance audit verification may be a single domain or may form a candidate domain list for further consideration, thereby determining the target planning domain that matches the target enterprise's business tasks.In this way, this embodiment performs preliminary filtering and evaluation of candidate planning domains based on structured use cases before each task execution, allowing only eligible planning domains to enter the activation phase. This clarifies the capability boundaries of each task, generates traceable capability boundary descriptions, and provides a technical foundation for auditing, backtracking, and task optimization. It makes business execution more aligned with the task scenario, ensuring predictable and explainable agent behavior, standardized agent behavior, and reasonable invocation, thereby improving the reliability and adaptability of task processing and enhancing system controllability. Availability filtering ensures that only planning domains matching the task context are selected, preventing capability abuse or unauthorized invocation. Intent and context matching ensures that the agent's execution behavior is highly consistent with actual business needs. In multi-task or asynchronous task environments, it enables effective isolation of different task capabilities. The multi-layered screening mechanism takes into account task matching, risk control, resource rationality, and compliance requirements, ensuring the refinement, controllability, and auditability of MoPD capability activation, and improving the stability and security of enterprise business agent execution. Adding new planning domains or tools will not interfere with the operation of existing planning domains, achieving capability isolation and secure iteration. The risk assessment object is elevated from within the model to the combination of planning domains and tools, forming a structured review and rollback closed loop.

[0027] Step S13: Within the agent execution context corresponding to the target planning domain, activate the defined target tool capabilities so that the target enterprise business agent can utilize the target tool capabilities to execute the target enterprise business tasks.

[0028] In this embodiment, a specified target tool capability is activated in the agent execution context corresponding to the target planning domain, so that the target enterprise business agent can execute the target enterprise business task based on the matched target tool capability.

[0029] It should be noted that the process of activating the defined target tool capabilities within the agent execution context corresponding to the target planning domain is as follows: The set of target tool capabilities and target behavior constraint rules defined by the target planning domain are determined; within the agent execution context corresponding to the target planning domain, the target tool capabilities in the set of target tool capabilities are activated according to the structured use cases and the target behavior constraint rules. That is, the enterprise business agent does not inherently possess any business capabilities; tool capabilities are only temporarily activated within the selected planning domain context and are strictly bound to the task context. Specifically, after the target planning domain is selected, the set of target tool capabilities and target behavior constraint rules defined by the target planning domain are determined. Then, within the corresponding agent execution context, the set of target tool capabilities is temporarily loaded according to the structured use cases and the target behavior constraint rules to activate the target tool capabilities within it. It can be understood that the activation behavior is strictly controlled by the task context, permissions, and constraint rules. The activated tool capabilities are bound to the context and can only be used within the current task and its subtasks. Isolation control is also implemented; inactive tools are invisible and cannot be called in the current execution context, achieving least privilege control and ensuring security. In addition, the system can activate multiple planning domains simultaneously. Subtasks inherit the planning domain context of their parent tasks or operate independently. The system dynamically allocates activation capability combinations based on task complexity, task level, and business objectives to ensure the controllability of complex business processes.

[0030] It should be noted that during the process of the target enterprise business agent using the target tool capability to execute the target enterprise business task, the processing flow is as follows: If the execution context of the first agent corresponding to the current first target planning domain ends, the first target tool capability defined by the first target planning domain is unloaded, and a second target planning domain is determined; the second target planning domain is the target planning domain that matches the execution context of the corresponding second agent; the second target tool capability defined by the second target planning domain is determined, and the second target tool capability is activated. That is, this embodiment performs lifecycle management of tool capabilities, managing them throughout the activation, execution, and unloading processes to ensure that tool capabilities have a clear scope of effectiveness, execution status, and revocation path in the task context, thereby achieving dynamic capability control and system security. Specifically, when the execution context corresponding to the current first target planning domain ends, the target tool capability defined therein is explicitly unloaded, unbound from the current context, and then a second target planning domain adapted to the current context is determined, and the tool capability corresponding to that planning domain is activated.

[0031] It should be further noted that the process of unloading the first target tool capability defined by the first target planning domain is as follows: The first target tool capability defined by the first target planning domain is removed from the current agent execution context; the current tool capability execution state and the cached temporary data generated during the invocation of the first target tool capability are cleaned up, and a corresponding tool capability invocation log is generated. That is, after the execution context of the first target planning domain ends, its corresponding target tool capability is removed from the current agent execution context, the tool execution state and temporary cached data are cleaned up, and a complete tool invocation log is generated, including the tool usage cycle, activation time, and uninstallation time, providing a basis for backtracking analysis. It is understood that this embodiment can also introduce a reflection mechanism, that is, a mechanism for analyzing and optimizing the actual behavior and planning domain selection results after executing the target enterprise business task and capability uninstallation. Its processing flow is as follows: The execution status of the target enterprise business task is summarized, and the several planning domains are adjusted and optimized based on the summary results. In other words, the execution of the target enterprise's business tasks is summarized, deviations and abnormal behaviors are identified, deviations in individual tasks are summarized and optimized, and each planning domain is iteratively adjusted and optimized based on the summary results. Specifically, the planning domain definition, tool allocation, and selection strategy are iteratively adjusted to optimize future planning domain definitions and capability allocation strategies. In this way, this embodiment loads the tool capabilities defined in the planning domain into the agent's execution context based on structured use cases and selected target planning domains. Only capabilities related to the task objective are activated and strictly bound to the task context, making them available in the current task or subtask. Inactive tools are invisible in the execution context, ensuring a high degree of matching between capability calls and tasks, implementing the principle of least privilege, avoiding cross-task capability abuse, allowing the agent to operate stably under standardized constraints, and improving the accuracy and security of business execution. Dynamically controlling the capability lifecycle to achieve a closed loop of activation and deactivation can efficiently release resources and achieve the above... The resource release and state reset during the switching process, along with the unloading mechanism, ensure that capabilities exist temporarily and disappear upon task completion. This supports security control in multi-task and multi-scenario environments, guarantees stable connection of the agent at different business stages, improves the continuity of task execution and resource utilization, ensures a clean and stable execution environment, and retains traceable logs, thereby enhancing the reliability and standardization of agent operation. The introduction of a reflection mechanism supports the continuous improvement and controllable evolution of agent behavior, forming a long-term operational optimization closed loop. This continuously improves the agent's matching degree and execution effect for business tasks, enhancing the controllability, auditability, and long-term security of agent behavior.

[0032] Step S14: During the execution of the target enterprise's business task, continuously monitor the capability invocation behavior of the target enterprise's business intelligence agent, and trigger corresponding exception handling operations when an anomaly is detected in the capability invocation behavior.

[0033] In this embodiment, during the execution of the target enterprise's business tasks, the capability invocation behavior of the intelligent agent is continuously monitored, the capability invocation behavior is reviewed in real time, and it is compared and verified with the planning domain definition in real time to ensure that the behavior conforms to the planning domain definition. When an abnormality is detected in the capability invocation behavior, the corresponding abnormality handling operation is triggered immediately.

[0034] It should be noted that the processing flow for triggering corresponding exception handling operations when an anomaly is detected in the capability invocation behavior is as follows: If the currently invoked tool capability is detected to be different from the target tool capability defined in the target planning domain, the capability invocation behavior is determined to be abnormal; the currently activated tool capability is uninstalled, or the execution operation of the target enterprise business task is terminated, or the target enterprise business task is rolled back to the target state; the target state is the task state that was backed up in advance before the capability invocation behavior became abnormal. That is, when it is detected that the currently invoked tool capability does not conform to the definition of the target planning domain, that is, it deviates from the planning domain definition or is called abnormally, the capability invocation behavior is determined to be abnormal, and the following processing methods are taken: uninstalling the current tool, automatically terminating the execution of the abnormal task, or rolling back the task to the backup state before the abnormality, rolling back the capabilities and operations already activated in the task, updating the risk model and planning domain selection strategy, improving the security of future tasks, and recording audit logs for post-event analysis and reflection mechanism optimization to achieve a closed loop of capability governance.

[0035] Understandably, in enterprise-level intelligent agent systems, different tasks can simultaneously activate different planning domains, ensuring capability isolation and the principle of least privilege between tasks. Dynamic switching of task contexts is supported, allowing the uninstallation of tool capabilities from the previous task before activating a new planning domain, avoiding cross-task residual risks. The system can summarize planning domain activation status and risk indicators in real time during execution, providing dynamic risk monitoring for long-cycle tasks. In this way, this embodiment combines a reflection mechanism with planning domain management, performing real-time verification based on the difference between the actual execution state and the planning domain definition. It supports systematic risk assessment, behavior review, and rollback operations. Deviations or abnormal calls trigger capability uninstallation, rollback, or execution interruption, thereby achieving full-process constraints on intelligent agent behavior, enabling rapid response to anomalies and safe task recovery, effectively ensuring task execution stability and system security. This not only guarantees the security of a single task but also supports behavior optimization during long-term system operation, thus ensuring the security, controllability, and auditability of intelligent agent behavior, and improving capability isolation and system stability in multi-task, multi-scenario environments.

[0036] As can be seen from the above, in this embodiment, after the target enterprise business intelligence agent receives the target enterprise business task to be processed, it first performs structured processing on the business task information to generate corresponding structured use cases, and then selects a matching target planning domain from a number of preset planning domains based on the structured use cases. The planning domain is used to define the set of tool capabilities and behavioral constraint rules of the intelligence agent under different business tasks. Subsequently, the required target tool capabilities are activated in the execution context corresponding to the target planning domain to execute the target enterprise business task through the target tool capabilities. During the task execution process, the capability call behavior of the target enterprise business intelligence agent is continuously monitored, and if an anomaly is detected, the corresponding anomaly handling operation is triggered. In this way, through the above-described process in the embodiments of this application, structured task processing, precise matching of planning domains, on-demand activation of tool capabilities, and full-process behavior monitoring achieve the standardization, controllability, and efficiency of the execution process of enterprise business intelligence agents. This ensures that the intelligence agent operates stably under preset constraints, and can promptly detect and handle execution anomalies, significantly improving the reliability and security of automated execution of enterprise business. Dynamically activating capabilities and strictly binding contexts prevents the abuse of cross-task capabilities and unexpected operations. Adding new planning domains or tools will not interfere with the operation of existing planning domains, achieving capability isolation and secure iteration. This optimizes the capability management method of enterprise business intelligence agents to ensure the security and controllability of intelligence agent behavior and improve system stability.

[0037] Accordingly, see Figure 2 As shown in the illustration, this application also provides a device for dynamic management and control of the capabilities of an enterprise business intelligence agent, comprising: The information processing module 11 is used to perform structured processing on the corresponding enterprise business task information after the target enterprise business intelligence agent receives the target enterprise business task to be processed, so as to obtain the corresponding structured use cases. The planning domain filtering module 12 is used to filter several pre-defined planning domains based on the structured use cases to determine the target planning domain that matches the target enterprise business task; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business agent in different enterprise business tasks. Capability activation module 13 is used to activate defined target tool capabilities within the agent execution context corresponding to the target planning domain, so that the target enterprise business agent can use the target tool capabilities to execute the target enterprise business tasks; The operation triggering module 14 is used to continuously monitor the capability invocation behavior of the target enterprise business agent during the execution of the target enterprise business task, and to trigger corresponding exception handling operations when an abnormality is detected in the capability invocation behavior.

[0038] In some specific embodiments, the planning domain filtering module 12 may specifically include: The target determination unit is used to determine the constraint-related information and business execution targets in the structured use cases; The context matching unit is used to perform availability filtering on a number of pre-defined planning domains based on the constraint-related information, and to perform intent context matching between the business execution target and the number of planning domains to obtain a number of corresponding initial candidate planning domains. The cost evaluation unit is used to evaluate the activation risk and resource consumption cost of the initial candidate planning domain, so as to filter the initial candidate planning domains according to the evaluation results and obtain the target candidate planning domain. The planning domain determination unit is used to determine the planning domains that have passed the compliance audit verification among the target candidate planning domains as the target planning domains that match the business tasks of the target enterprise.

[0039] In some specific embodiments, the capability activation module 13 may specifically include: The rule determination unit is used to determine the set of target tool capabilities and target behavior constraint rules defined in the target planning domain; The first capability activation unit is used to activate the target tool capabilities in the target tool capability set according to the structured use cases and the target behavior constraint rules within the agent execution context corresponding to the target planning domain.

[0040] In some specific embodiments, the capability activation module 13 may specifically include: The capability unloading submodule is used to unload the first target tool capability defined in the first target planning domain if the first agent execution context corresponding to the current first target planning domain ends, and to determine the second target planning domain; the second target planning domain is the target planning domain that matches the corresponding second agent execution context with the current agent execution context; The second capability activation unit is used to determine the second target tool capability defined by the second target planning domain and activate the second target tool capability.

[0041] In some specific implementations, the capability offloading submodule may specifically include: The capability removal unit is used to remove the first target tool capability defined in the first target planning domain from the current agent execution context; The log generation unit is used to clean up the current execution status of the tool capability, the cached temporary data generated during the invocation of the first target tool capability, and generate the corresponding tool capability invocation log.

[0042] In some specific embodiments, the enterprise business intelligence agent's capability dynamic management and control device may further include: The situation summary unit is used to summarize the execution status of the target enterprise's business tasks, so as to adjust and optimize the several planning domains based on the summary results.

[0043] In some specific embodiments, the operation triggering module 14 may specifically include: The condition determination unit is used to determine that the capability invocation behavior is abnormal if it detects that the currently invoked tool capability is different from the target tool capability defined in the target planning domain. The task rollback unit is used to unload the currently activated tool capability, or terminate the execution operation of the target enterprise business task, or roll back the target enterprise business task to the target state; the target state is the task state that was backed up in advance before the capability call behavior became abnormal.

[0044] Furthermore, embodiments of this application also disclose an electronic device, Figure 3 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the dynamic capability management method for enterprise business intelligence disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be a computer.

[0045] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0046] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0047] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the dynamic management method for the enterprise business intelligence agent executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.

[0048] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned method for dynamic management and control of the capabilities of an enterprise business intelligent agent. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0049] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0050] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0051] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0052] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0053] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for dynamic management and control of the capabilities of an enterprise business intelligence agent, characterized in that, include: After the target enterprise business intelligence agent receives the target enterprise business task to be processed, it performs structured processing on the corresponding enterprise business task information to obtain the corresponding structured use cases. Based on the structured use cases, several pre-defined planning domains are filtered to determine the target planning domain that matches the target enterprise business task; the planning domain is used to define the set of tool capabilities and behavioral constraint rules of the enterprise business agent in different enterprise business tasks. Within the agent execution context corresponding to the target planning domain, the defined target tool capabilities are activated so that the target enterprise business agent can utilize the target tool capabilities to execute the target enterprise business tasks. During the execution of the target enterprise's business tasks, the capability invocation behavior of the target enterprise's business intelligence agent is continuously monitored, and when an anomaly is detected in the capability invocation behavior, corresponding anomaly handling operations are triggered.

2. The method for dynamic control of the capabilities of an enterprise business intelligence agent according to claim 1, characterized in that, The step of filtering several pre-defined planning domains based on the structured use cases to determine the target planning domain that matches the target enterprise's business tasks includes: Determine the constraint-related information and business execution objectives in the structured use cases; Based on the constraint-related information, availability filtering is performed on several pre-defined planning domains, and intent context matching is performed between the business execution target and the several planning domains to obtain several initial candidate planning domains. The activation risk and resource consumption cost of the initial candidate planning domains are evaluated, and the target candidate planning domains are obtained by screening the initial candidate planning domains based on the evaluation results. Among the target candidate planning domains, the planning domains that pass the compliance audit verification are determined as the target planning domains that match the business tasks of the target enterprise.

3. The method for dynamic control of the capabilities of an enterprise business intelligence agent according to claim 1, characterized in that, Activating the defined target tool capabilities within the agent execution context corresponding to the target planning domain includes: Determine the set of target tool capabilities and target behavior constraint rules defined in the target planning domain; Within the agent execution context corresponding to the target planning domain, the target tool capabilities in the target tool capability set are activated according to the structured use case and the target behavior constraint rules.

4. The method for dynamic control of the capabilities of an enterprise business intelligence agent according to claim 1, characterized in that, The process of utilizing the target tool capabilities to execute the target enterprise's business tasks includes: If the execution context of the first agent corresponding to the current first target planning domain ends, the first target tool capability defined in the first target planning domain is unloaded, and a second target planning domain is determined; the second target planning domain is the target planning domain that matches the execution context of the corresponding second agent and the current agent execution context. Determine the second objective tool capability defined by the second objective planning domain, and activate the second objective tool capability.

5. The method for dynamic control of the capabilities of an enterprise business intelligent agent according to claim 4, characterized in that, The unloading of the first target tool capability defined by the first target planning domain includes: Remove the first objective tool capability defined in the first objective planning domain from the current agent execution context; The current execution status of the tool capability and the cached temporary data generated during the invocation of the first target tool capability are cleaned up, and corresponding tool capability invocation logs are generated.

6. The method for dynamic control of the capabilities of an enterprise business intelligent agent according to claim 1, characterized in that, After utilizing the target tool's capabilities to execute the target enterprise's business tasks, the process further includes: The execution of the target enterprise's business tasks is summarized, and the planning domains are adjusted and optimized based on the summary results.

7. The method for dynamic control of the capabilities of an enterprise business intelligence agent according to any one of claims 1 to 6, characterized in that, When an anomaly is detected in the capability invocation behavior, the corresponding exception handling operation is triggered, including: If it is detected that the currently invoked tool capability is different from the target tool capability defined in the target planning domain, then the capability invocation behavior is determined to be abnormal; Uninstall the currently activated tool capability, or terminate the execution of the target enterprise business task, or roll back the target enterprise business task to the target state; the target state is the task state that was backed up in advance before the capability call behavior became abnormal.

8. A device for dynamic control of the capabilities of an enterprise business intelligence agent, characterized in that, include: The information processing module is used to perform structured processing on the corresponding enterprise business task information after the target enterprise business intelligence agent receives the target enterprise business task to be processed, so as to obtain the corresponding structured use cases. The planning domain filtering module is used to filter several pre-defined planning domains based on the structured use cases in order to determine the target planning domain that matches the target enterprise's business tasks. The planning domain is used to define the set of tool capabilities and behavioral constraint rules of enterprise business intelligence agents in different enterprise business tasks; The capability activation module is used to activate the defined target tool capabilities within the agent execution context corresponding to the target planning domain, so that the target enterprise business agent can use the target tool capabilities to execute the target enterprise business tasks. The operation triggering module is used to continuously monitor the capability invocation behavior of the target enterprise business agent during the execution of the target enterprise business task, and to trigger corresponding exception handling operations when an abnormality is detected in the capability invocation behavior.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the method for dynamic management and control of the capabilities of an enterprise business intelligence agent as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when the computer programs are executed by a processor, they implement the dynamic management method for the capabilities of an enterprise business intelligence agent as described in any one of claims 1 to 7.

Images