Management device, management method, and management program

Abstract

Ensure that the timing, content, and recipients of the presentation are appropriate. [Solution] A management device having a processor that executes a program, a storage device that stores the program, and a communication interface that can communicate with a group of target computers which is a collection of target computers for each of a group of users and a group of monitored systems which is a collection of monitored systems used by at least one of the group of users, executes a collection process which collects information from the group of monitored systems, a generation process which generates a task that the user should execute and a presentation timing for presenting the task based on the collection results from the collection process, and a transmission process which sends the task to the user's target computer at the presentation timing.

Description

Technical Field

[0001] The present invention relates to a management device, a management method, and a management program for managing information.

Background Art

[0002] The supply chain of the manufacturing industry consists of multiple processes such as procurement, production planning, manufacturing equipment management, manufacturing, quality assurance, and maintenance. At the site of each process, regular operations are carried out along the business flow towards smooth manufacturing. Also, at the site, responses to abnormal situations such as failures, accidents, and disasters also occur. Further, at the site, adjustment operations are carried out between the processes to which the employees belong and the processes before and after them.

[0003] The following Patent Document 1 discloses a method for automatically detecting irrelevant events unrelated to operations and deleting them from the business flow. In this method, a value of an evaluation formula for determining an irrelevant event that is extracted as a business event despite being unrelated to business processing and occurs independently of other business events related to business processing is calculated using the occurrence probability and conditional probability of the business event. A business event class to be judged whose evaluation value is below a predetermined threshold is specified, and an event instance belonging to the event class is deleted from the original process instance to generate a corrected process instance.

[0004] Patent Document 2 below discloses a text mining method for extracting useful information from a large amount of electronically stored documents (hereinafter referred to as "text") that are necessary for quality control operations such as design specifications and failure investigation reports, and for using this information for business improvement. In this method, words or attribute values ​​extracted from the text to be analyzed are classified into two or more categories and displayed in a list. The user is provided with these two or more relationships, and when a word or attribute value is specified by the user, the list is narrowed down and displayed in conjunction with the specified word or attribute value. That is, because the list display and the specified narrowing are linked, it is easy to further narrow down the list while maintaining the relationships while overviewing the list display, and the narrowed-down results can then be displayed in a list. As a result, data analysis becomes easier. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2010-20634 [Patent Document 2] Japanese Patent Publication No. 2006-244298 [Overview of the project] [Problems that the invention aims to solve]

[0006] However, the conventional technologies described above do not take into account when, what, and to whom the information should be presented.

[0007] The present invention aims to optimize the timing, content, and target audience of its presentation. [Means for solving the problem]

[0008] A management device representing one aspect of the invention disclosed in this application is a management device having a processor that executes a program, a storage device that stores the program, a group of destination computers which is a collection of destination computers for each of a plurality of users, and a group of monitored systems which is a collection of monitored systems used by at least one of the plurality of users, wherein the processor is characterized by executing a collection process for collecting information from the group of monitored systems, a generation process that generates a task to be executed by the user and a presentation timing for presenting the task based on the collection results of the collection process, and a transmission process that transmits the task to the user's destination computer at the presentation timing. [Effects of the Invention]

[0009] According to a typical embodiment of the present invention, the timing, content, and target audience of the presentation can be optimized. Problems, configurations, and effects other than those mentioned above will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0010] [Figure 1] Figure 1 is an explanatory diagram showing an example of the system configuration of a work support system. [Figure 2] Figure 2 is an explanatory diagram showing an example of the operation of the work support system. [Figure 3] Figure 3 is a block diagram showing an example of a computer hardware configuration. [Figure 4] Figure 4 is an explanatory diagram showing an example of a business process flow database. [Figure 5] Figure 5 is an explanatory diagram showing an example of a work instruction database. [Figure 6] Figure 6 is an explanatory diagram showing an example of a work report database. [Figure 7] Figure 7 is an explanatory diagram showing an example of a database of results from routine operations. [Figure 8] Figure 8 is an explanatory diagram showing an example of a non-routine work case database. [Figure 9]FIG. 9 is an explanatory diagram showing an example of an employee information DB. [Figure 10] FIG. 10 is an explanatory diagram showing an example of an attendance information DB. [Figure 11] FIG. 11 is an explanatory diagram showing an example of a schedule DB. [Figure 12] FIG. 12 is a block diagram showing a functional configuration example of an AI agent server. [Figure 13] FIG. 13 is a sequence diagram showing an example of crawling by a work support system. [Figure 14] FIG. 14 is a flowchart showing an example of a work support processing procedure by an AI agent. [Figure 15] FIG. 15 is a flowchart showing a detailed processing procedure example of a presentation information determination process (step S1402). [Figure 16] FIG. 16 is a flowchart showing a detailed processing procedure example of a presentation information determination process (step S1402). [Figure 17] FIG. 17 is a flowchart showing a detailed processing procedure example of an execution target determination process (step S1403). [Figure 18] FIG. 18 is a flowchart showing a detailed processing procedure example of an execution target execution process (step S1404). [Figure 19] FIG. 19 is an explanatory diagram showing a first example of a display screen of a terminal. [Figure 20] FIG. 20 is an explanatory diagram showing a second example of a display screen of a terminal. [Figure 21] FIG. 21 is an explanatory diagram showing a third example of a display screen of a terminal. [Figure 22] FIG. 22 is an explanatory diagram showing a fourth example of a display screen of a terminal.

MODE FOR CARRYING OUT THE INVENTION

[0011] <FIGS. 1 and 2 Work Support System> FIG. 1 is an explanatory diagram showing a system configuration example of a work support system. FIG. 2 is an explanatory diagram showing an operation example of the work support system 100.

[0012] The work support system 100 includes an AI (Artificial Intelligence) agent system 101, process management systems 102-1 to 102-n (where n is an integer greater than or equal to 1), an employee management system 103, and one or more terminals 104. These are connected via a network 106 such as the Internet, LAN (Local Area Network), or WAN (Wide Area Network) to enable communication.

[0013] [AI Agent System 101] The AI ​​agent system 101 comprises an AI agent server 110 and storage 111. The AI ​​agent server 110 is a management device that manages the work support system 100, and is a computer on which the AI ​​agent system 101 behaves as an AI agent 200.

[0014] AI Agent 200 is a software agent that uses generative AI. A software agent is a virtual executing entity that functions as a proxy for a user to achieve a certain objective, behaves autonomously with a certain degree of judgment ability, and operates continuously. When a goal is set, it autonomously generates task proposals to achieve that goal. Specifically, for example, AI Agent 200 performs various tasks such as monitoring and control within the work support system 100, data collection and processing, and dialogue (chatbot).

[0015] AI Agent 200 is composed of (1) personality, (2) memory, (3) planning, and (4) behavior, and these interact with each other.

[0016] (1) Personality is the role in task execution. In this example, the personality of AI agent 200 is the role of a manager who oversees the process management systems 102-1 to 102-n and the employees who work with them. In the case of a manufacturing supply chain, a series of processes such as procurement process P1 ⇒ production planning process P2 ⇒ manufacturing equipment management process P3 ⇒ manufacturing process P4 ⇒ quality assurance process P5 ⇒ maintenance process P6 are carried out by process management systems 102-1 to 102-6.

[0017] In this case, the AI ​​agent 200 is aware that process management system 102-1 performs the procurement process P1, process management system 102-2 performs the production planning process P2, ..., and process management system 102-6 performs the maintenance process P6. The AI ​​agent 200 has the role of a manager to autonomously predict the next action that employees engaged in process management systems 102-1 to 102-n should take and to present information to employees at the appropriate time.

[0018] (2) Memory includes both short-term and long-term memory. In this example, the memory of AI agent 200 includes, for example, business flow DB121, work order DB122, work report DB123, routine work execution result DB124, non-routine work example DB125, employee information DB131, attendance information DB132, schedule DB133 and message DB134, as well as crawling results from external systems 105 and conversation logs between AI agent 200 and employees, and is stored in storage 111. Newer memories are valued in the plan.

[0019] (3) Planning is the process by which (1) an AI agent 200 with its own personality (2) relies on its memory to formulate tasks in order to achieve the goal. In this example, the planning of AI agent 200 is, for example, the process of defining the completion of the goal as the achievement of the task, and breaking down the task into one or more tasks (the goal itself may be a single task) based on prompts.

[0020] For example, consider a case where AI agent 200 instructs employee x to perform a certain task at a specific time. The necessary information is obtained from storage 111 where crawling results are stored, but to clarify the information source, we will describe it as various databases rather than storage 111.

[0021] Here, as an example, let's assume that the current time is before the employees arrive at work, and that the night shift team performed routine tasks within manufacturing process P4 last night.

[0022] AI agent 200 retrieves information about employee x from employee information DB 131. According to the retrieved information, employee x's job title is section chief in charge of manufacturing process P4, and the job definition in employee information DB 131 is supervision within manufacturing process P4. Therefore, AI agent 200 identifies employee x as the entity responsible for performing the following task b1.

[0023] For example, task b1 may be defined in the business flow DB121 as one of the manager's daily tasks, stored in storage 111 via crawling, and made accessible to the AI ​​agent 200. Alternatively, the AI ​​agent 200 may retrieve task b1 by giving the generating AI a prompt to query the tasks that the employee x with the acquired aptitude should perform.

[0024] (b1) In manufacturing process P4, confirm the routine operation r1 from last night.

[0025] Furthermore, the AI ​​agent 200, by referring to the attendance information DB 132, detects that employee x is not currently at work. Therefore, the AI ​​agent 200 determines the timing for notifying employee x of a task after their arrival at work (for example, when the AI ​​agent 200 detects that employee x's terminal 104 has been activated) from the attendance information DB 132 and the schedule DB 133.

[0026] Since task b1 includes "Last night's routine task r1", AI agent 200 identifies employees y and z, who perform routine task r1 in manufacturing process P4, from employee information DB 131. Since task b1 includes "Last night", AI agent 200 checks the attendance status of employees y and z last night from attendance information DB 132 and identifies from schedule DB 133 that employees y and z performed routine task r1 last night as part of the night shift.

[0027] Furthermore, since task b1 includes "confirm," AI agent 200 specifies work report DB123 as the information source necessary for confirming routine task r1, and retrieves work report d of routine task r1 performed last night by employees y and z from work report DB123. AI agent 200 analyzes the retrieved work report d using natural language processing. Suppose that "short stoppage of device A" is detected as irregular information from the analysis results. Sentences containing terms such as "stoppage" in the NG word list in storage 111 are detected as irregular information.

[0028] Then, the AI ​​agent 200 generates a prompt requesting task decomposition using the results of the above processing and provides it to the generating AI, causing the generating AI to decompose task b1. Here, we assume that task b1 has been decomposed into tasks t1 and t2.

[0029] (t1) Please check the work report from the night shift team. (t2) This work report d includes the phrase "minor shutdown of device A". Please consider the need to repair device A.

[0030] The AI ​​agent 200 will send tasks t1 and t2 to employee x's terminal 104 at the notification timings mentioned above.

[0031] The AI ​​agent 200 generates and breaks down tasks b2, b3, ... other than task b1 for each employee as needed. Alternatively, these tasks b1, b2, b3, ... may be stored in storage 111 beforehand as basic tasks. In this case, the assignment of each task to each employee for task breakdown is also pre-configured and stored in storage 111.

[0032] (4) Actions are specific actions, instructions, and information presentations for the execution of each task broken down in the plan. In this example, the actions of the AI ​​agent 200 include, for example, selecting a control target system from the process management system 102-1 to 102-n, the employee management system 103, the terminal group 104, and the external system group 105, and controlling the selected control target system to execute the next task, or instructing the selected control target system or the employee working on the selected control target system what task should be performed next, or presenting the information necessary to perform the task.

[0033] When employee x gives approval for the above task t1 from terminal 104, the AI ​​agent 200 selects the process management system 102-i that performed the routine work r1 as the system to be controlled, retrieves the work report d from the work report DB 123 of the process management system 102-4 of manufacturing process P4 (if it has already been retrieved and is stored in storage 111 when task t1 was generated, it may be retrieved from storage 111), and sends the work report d or link information that allows access to the work report d to employee x's terminal 104 (this may also be sent when task t1 is sent).

[0034] Regarding the above task t2, if employee x gives the instruction, for example, "Tell me the operating history and repair history of device A" in order to determine the need to repair device A, the AI ​​agent 200 generates tasks t21 and t22 by breaking down the task based on that instruction.

[0035] (t21) "Acquisition of operating history of device A" (t22) "Obtaining the repair history of device A"

[0036] Task t21 has an "operation history," and AI agent 200 recognizes that device A operates in the process control system 102-4 of manufacturing process P4. Therefore, AI agent 200 selects the process control system 102-4 of manufacturing process P4 as the system to be controlled, and retrieves the "operation history of device A" from task t21, using "device A" as the key, from the regular operation execution result DB124 and non-regular operation example DB125 of the process control system 102-4 of manufacturing process P4 that uses device A.

[0037] Task t22 is titled "Repair History," but AI agent 200 recognizes that "repair" is a task handled by the process management system 102-6 of maintenance process P6. Therefore, AI agent 200 selects the process management system 102-6 of maintenance process P6 as the system to be controlled, requests the process management system 102-6 of maintenance process P6 to obtain the repair history of device A, and obtains the repair history of device A from the process management system 102-6 of maintenance process P6.

[0038] As a result of executing tasks t21 and t22, the AI ​​agent 200 sends the operation history and repair history of device A to employee x's terminal 104.

[0039] In this way, the AI ​​agent 200 generates tasks from the information in the work support system 100 and takes actions to accomplish those tasks. Note that the task decomposition described above is just one example, and the method of decomposition will differ depending on the type of work.

[0040] In the example above, the AI ​​agent 200 set the notification timing based on the employee x arrival time, even when employee x is not yet at work. However, even when employee x is at work, the AI ​​agent 200 will set the notification timing as appropriate if it detects tasks that employee x should be performing.

[0041] For example, regarding the confirmation of routine task r2 completed in the morning, the AI ​​agent 200 will identify employee x as the entity performing task b2, which is "to confirm routine task r2 completed this morning in manufacturing process P4," and, similar to task b1, will break it down into tasks and take action according to those tasks.

[0042] Furthermore, if tasks b1, b2, b3, ... are already stored in storage 111 as basic tasks, the AI ​​agent 200 may, in advance, use the generated AI to attempt task decomposition for each task b1, b2, b3, ... and learn to achieve appropriate task decomposition.

[0043] Furthermore, the generative AI used by the AI ​​agent 200 generates text using a language model. Specifically, for example, the generative AI refers to the crawling results 1251 (see Figure 12) stored in the storage 111 to generate text according to the request. Specifically, for example, the generative AI generates answer texts to questions, summarizes input texts in response to input text summarization requests, and generates texts proposing solutions in response to requests for solutions.

[0044] A language model is a trained model that, for example, tokenizes an input string and predicts the next token from that sequence of tokens. Examples of language models include transfer models such as GPT (Generative Pre-trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers).

[0045] The AI ​​agent 200 uses this generative AI to converse with employees through the employee's terminal 104. The conversation log 1252 (see Figure 12) between the AI ​​agent 200 and the employee is stored in the storage 111. The conversation log 1252 consists of text data of the message that constitutes the conversation, including the sender, date and time of sending, recipient, date and time of receiving, and message content.

[0046] [Process Management System 102-1~102-n] In the process management systems 102-1 to 102-n, i is an integer satisfying 1 ≤ i ≤ n, and is a number that identifies a process. A series of processes P1 to Pn are supported by the work support system 100. For example, in the case of a manufacturing supply chain, each of the processes in a series such as procurement process P1 ⇒ production planning process P2 ⇒ manufacturing equipment management process P3 ⇒ manufacturing process P4 ⇒ quality assurance process P5 ⇒ maintenance process P6 is supported by the work support system 100.

[0047] Figure 2 shows the operation of the work support system 100 in the manufacturing process P4, which takes place after the production planning process P2 and the manufacturing equipment management process P3. It shows a worker W in the manufacturing process P4 interacting with the AI ​​agent 200 using a terminal 104.

[0048] Each of the process management systems 102-i includes a process management server 120, a business flow DB 121, a work instruction DB 122, a work report DB 123, a regular work execution result DB 124, and an irregular work example DB 125. The process management server 120 is a computer that manages process Pi and is a managed computer that is controlled by the AI ​​agent 200.

[0049] Business Flow DB121 is a database that stores information about the business flow of routine tasks related to process Pi. A business flow is a set of procedures for routine tasks that should be performed within process Pi. Details of Business Flow DB121 will be described later in Figure 4.

[0050] Work Instruction DB122 is a database that stores work instructions related to process Pi. A work instruction is a document that describes the content of instructions given to worker W, based on the business flow of process Pi, outlining when and what tasks to perform in advance. Worker W performs the work based on the work instructions created in advance. Details of Work Instruction DB122 will be described later in Figure 5.

[0051] Work Report DB123 is a database that stores work reports related to process Pi. A work report is document data that describes the content of reports regarding work related to process Pi. Details of Work Report DB123 will be described later in Figure 6.

[0052] The Routine Work Execution Results DB124 is a database that stores the results of routine work execution related to process Pi. Routine work execution results are the results of performing routine work. Routine work refers to the work described in the work instructions for process Pi. The Routine Work Execution Results DB124 is updated in real time by the process management system 102-i. Details of the Routine Work Execution Results DB124 will be described later in Figure 7.

[0053] Non-routine work case DB125 is a database that stores non-routine work cases related to process Pi. Non-routine work cases are events that occurred in precedent non-routine work in process Pi. Non-routine work refers to all work that is not specified in the business flow DB121. For example, it includes work that needs to be handled exceptionally outside of routine work and is described in the work instruction sheet for process Pi, or work that responds to sudden emergencies such as power outages or accidents that are not described in the work instruction sheet. Details of non-routine work case DB125 will be described later in Figure 8.

[0054] [Employee Management System 103] The employee management system 103 includes an employee management server 130, an employee information database 131, an attendance information database 132, a schedule database 133, and a message database 134. The employee management server 130 is a computer that manages employees engaged in process Pi, and is a managed computer controlled by the AI ​​agent 200. The employee management system 103 is operated on one or more computers. Worker W refers to an employee engaged in process Pi.

[0055] Employee Information DB131 is a database that stores basic information about employees.

[0056] The attendance information DB132 is a database that stores information about employees' attendance.

[0057] Schedule DB133 is a database that stores employee schedules.

[0058] Message DB134 is a database that manages messages sent and received between employee terminals 104. Message DB134 stores the sender, date and time of sending, recipient, date and time of receiving, and message content of each message.

[0059] [Terminal 104] Terminal 104 is a computer used by employee W engaged in process Pi, and is the computer to which information is presented by AI agent 200.

[0060] [External Systems Group 105] The external systems group 105 is a collection of external systems that can communicate with the work support system 100 via the network 106, and each of them is a managed computer controlled by the AI ​​agent 200. The external systems group 105 includes, for example, a power information system 151 and a weather information system 152. The power information system 151 is, for example, a computer owned by a power company and transmits power information to the work support system 100. The weather information system 152 is, for example, a computer owned by the Japan Meteorological Agency or a private company that handles weather information and transmits weather information to the work support system 100.

[0061] <Figure 3: Example of computer hardware configuration> Next, we will describe an example of the hardware configuration of the computer shown in Figure 1 (AI agent server 110, process management server 120, employee management server 130, terminal 104, power information system 151, weather information system 152).

[0062] Figure 3 is a block diagram showing an example of the hardware configuration of a computer. Computer 300 includes a processor 301, a storage device 302, an input device 303, an output device 304, and a communication interface (communication IF) 305. The processor 301, storage device 302, input device 303, output device 304, and communication IF 305 are connected by a bus 306. The processor 301 controls computer 300. The storage device 302 serves as the work area for the processor 301. The storage device 302 is a non-temporary or temporary recording medium that stores various programs and data. Examples of storage devices 302 include ROM (Read Only Memory), RAM (Random Access Memory), HDD (Hard Disk Drive), and flash memory. The input device 303 inputs data. Examples of input devices 303 include a keyboard, mouse, touch panel, numeric keypad, scanner, microphone, and sensor. The output device 304 outputs data. Output devices 304 include, for example, displays, printers, and speakers. The communication interface 305 connects to the network 106 and sends and receives data.

[0063] <Figure 4 Business Flow DB121> Figure 4 is an explanatory diagram showing an example of the business flow DB121. The business flow DB121 has the following fields: process control number 401, process name 402, intermediate process name 403, intermediate process content 404, standard work time 405, equipment used 406, standard process conditions 407, high-risk work applicability 408, and comments 409. The combination of values ​​for each field in the same row constitutes an entry that defines the business of intermediate process Qj.

[0064] Process control number 401 is a control number that uniquely identifies process Pi. Process Pi is subdivided into one or more sub-processes, and process control number 401 is different for each sub-process. In this example, process control number 401 is represented as Pi-Qj, where Pi is the process name 402 and Qj is the sub-process name 403.

[0065] Process name 402 is identification information that uniquely identifies process Pi.

[0066] The intermediate process name 403 is identification information that uniquely identifies intermediate process Qj. In this example, the intermediate process names 403 are "Q1", "Q2", and "Q3".

[0067] Intermediate process content 404 is a string of characters indicating the work content of intermediate process Qj. The work content of intermediate process Qj is defined based on laws and guidelines concerning, for example, hazardous work such as high voltage, handling of hazardous chemicals, and occupational safety and health.

[0068] The standard work time of 405 is the standard work time required for the intermediate process Qj.

[0069] Equipment 406 is used in the intermediate process Qj.

[0070] Standard process conditions 407 define the parameter settings for the equipment 406 used in intermediate process Qj, the procedure for operating the equipment, the time required for each procedure, and precautions.

[0071] High-risk work status 408 indicates whether the work in intermediate process Qj is a high-risk task. High-risk tasks are those performed only by experienced workers W.

[0072] Comment 409 is a string describing the intermediate process Qj. Comment 409 can be updated from terminal 104.

[0073] <Figure 5 Work Instructions DB122> Figure 5 is an explanatory diagram showing an example of the work order DB122. The work order DB122 has the following fields: work order number 501, creation date 502, creator ID 503, instruction summary 504, link to detailed content 505, comments 506, and related work report number 507. The combination of values ​​for each field in the same row constitutes an entry that defines the work order information.

[0074] Work order number 501 is an identification number that uniquely identifies the work order.

[0075] The creation date 502 is the date the work order was created.

[0076] Creator ID 503 is the creator's employee ID 901 (see Figure 9), which uniquely identifies the creator of the work order. Employee ID 901 is identification information that uniquely identifies the employee.

[0077] Instruction Summary 504 is a string of characters that shows a summary of the instructions in the work order.

[0078] The link 505 to the detailed information is access information to the location where the work order is saved, for example, a URL (Uniform Resource Locator). By specifying the link 505 to the detailed information, the work order can be downloaded or opened as a file on terminal 104.

[0079] Comment 506 is a string of text describing the work instructions. Comment 506 can be updated from terminal 104.

[0080] Related work report number 507 is work report number 601 of the work report associated with the work order. When worker W is required to promptly submit a work report after completing the work, work order number 501 of the work order for that work is associated with related work report number 507.

[0081] <Figure 6 Work Report DB123> Figure 6 is an explanatory diagram showing an example of the work report DB123. The work report DB123 has the following fields: work report number 601, creation date 602, creator ID 603, work summary 604, link to detailed content 605, comments 606, and related work order number 607. The combination of values ​​for each field in the same row constitutes an entry that defines the work report information.

[0082] Work report number 601 is an identification number that uniquely identifies the work report.

[0083] The creation date 602 is the date the work report was created.

[0084] Creator ID 603 is employee ID 901, which uniquely identifies the creator of the work report.

[0085] "Work Summary 604" is a string of characters that indicates the summary of the contents of the work report.

[0086] The link 605 to the detailed content is access information to the location where the work report is saved, for example, a URL (Uniform Resource Locator). By specifying the link 605 to the detailed content, terminal 104 can download or open the work report file.

[0087] Comment 606 is a string of text describing the work report. Comment 606 can be updated from terminal 104.

[0088] Related work order number 607 is work order number 501 of the work order associated with the work report. When worker W is required to promptly submit a work report after the completion of the work, work report number 601 of the work report for that work is associated with related work order number 607.

[0089] <Figure 7: Results of routine work execution DB124> Figure 7 is an explanatory diagram showing an example of the regular work execution result DB124. The regular work execution result DB124 has the following fields: lot number 701, process control number 401, process name 402, sub-process name 403, worker ID 702, start date and time 703, end date and time 704, status 705, and comment 706. The combination of values ​​of each field in the same row constitutes an entry that defines the regular work execution result.

[0090] Lot number 701 is an identification number that uniquely identifies a lot in process Pi. A lot is the smallest unit of a certain quantity of workpieces (e.g., manufactured goods) in process Pi.

[0091] Worker ID 702 is the employee ID of worker W, who is performing work on lot number 701.

[0092] The start date and time 703 is the date and time when work began on lot number 701.

[0093] The completion date and time 704 is the date and time when work on lot number 701 was completed.

[0094] Status 705 indicates the status of work (e.g., before start, in progress, completed) for lot number 701. Status 705 is updated in real time by the process control system 102-i.

[0095] Comment 706 is a string of text describing the work done on lot number 701. Comment 706 is updatable.

[0096] In this way, the regular work execution results DB124 aggregates the regular work execution results, such as which lot passed through which intermediate process Qj of which process Pi, and what the result was, starting from the lot. Note that in Figure 7, the regular work execution results are aggregated starting from the lot (lot number 701), but the regular work execution results may also be aggregated using the process (process control number 401) as the starting point. In this case, the regular work execution results DB124 aggregates the regular work execution results, such as which lot passed through which intermediate process Qj of which process Pi, and what the result was, starting from the process.

[0097] <Figure 8 Non-routine work example DB125> Figure 8 is an explanatory diagram showing an example of the Non-Routine Work Case DB125. The Non-Routine Work Case DB125 has the following fields: Case Number 801, Case Name 802, Type 803, Location 804, Work Content 805, Worker ID 806, Occurrence Date and Time 807, Completion Date and Time 808, and Status 809. The combination of values ​​for each field in the same row constitutes an entry that defines a non-routine work case.

[0098] Case number 801 is an identification number that uniquely identifies a non-routine work case.

[0099] Case name 802 is the name of the non-routine work case identified by case number 801.

[0100] Category 803 is an item that groups together multiple non-routine work cases, including the non-routine work case identified by case number 801, based on common characteristics (for example, malfunction, power outage, maintenance, etc.).

[0101] Location 804 is the location where the non-routine work incident identified by incident number 801 occurred.

[0102] Task 805 describes the work performed to address the non-routine task identified in case number 801.

[0103] Worker ID 806 is employee ID 901 of worker W, who handled the non-routine work case identified by case number 801.

[0104] The occurrence date and time 807 is the date and time when the non-routine work incident identified by incident number 801 occurred.

[0105] The end date and time 808 is the date and time when the non-routine work case identified by case number 801 was completed.

[0106] Status 809 indicates the status of a non-routine work instance (e.g., before start, during work, or after completion) identified by its instance number 801. Status 809 is updated in real time by the process control system 102-i.

[0107] Thus, in the Non-Routine Work Examples DB125, non-routine work is aggregated as irregular cases. In Figure 8, non-routine work is recorded chronologically, but it could also be categorized by type 803.

[0108] <Figure 9 Employee Information DB131> Figure 9 is an explanatory diagram showing an example of the employee information DB131. The employee information DB131 has the following fields: employee ID 901, employee name 902, age 903, job title 904, work history 905, work experience 906, job definition 907, qualifications 908, and preferred language 909. The combination of values ​​for each field in the same row constitutes an entry that defines the employee information of an employee engaged in process Pi. The employee information DB131 is updated over time. This reflects changes in employee information such as personnel changes, qualification acquisition, and additions to work history.

[0109] Employee ID 901 is an identifier that uniquely identifies the employee.

[0110] Employee name 902 is the name of the employee identified by employee ID 901.

[0111] Age 903 represents the number of years elapsed since the date of birth of the employee identified by employee ID 901.

[0112] Job title 904 represents the current position, role, and status of the employee identified by employee ID 901 in the workplace, and indicates the employee's suitability. Job title 904 is referenced when there are restrictions on the performance of work and tasks that fall under job title 904. For example, if there are work and tasks required for job title 904, "Section Manager," AI agent 200 will determine that the employee is suitable and generate work and tasks for the employee whose job title 904 corresponds to "Section Manager."

[0113] Experience 905 is the employee's length of service and previous job titles 904, identified by their employee ID 901. Experience 905 is referenced when there are restrictions on the performance of tasks and duties. For example, if there are tasks and duties that require experience 905 for a supervisor with one year or more of experience, AI agent 200 will determine that the employee is suitable and generate tasks and duties for employees whose experience 905 matches "supervisor with one year or more of experience".

[0114] Work experience 906 represents the work experience of the employee identified by employee ID 901, and indicates the employee's suitability. Work experience 906 is referenced when there are restrictions on the execution of work and tasks. For example, if there are work and tasks that require work experience 906 of "20 years or more of service," the AI ​​agent 200 will determine that the employee is suitable and generate work and tasks for employees whose work experience 906 falls under "20 years or more of service."

[0115] Job definition 907 is a job defined for the employee identified by employee ID 901 (e.g., job content, scope of responsibilities, required skills), and indicates the employee's suitability. Job definition 907 is referenced when there are restrictions in job definition 907 that apply to the performance of the work and tasks. For example, if there are work and tasks required by job definition 907 for "supervisor," AI agent 200 will determine that the employee is suitable and generate work and tasks for the employee whose job definition 907 corresponds to "supervisor."

[0116] Qualification 908 refers to the requirements for a position or role that an employee identified by employee ID 901 is deemed necessary or suitable for to perform a certain action, or the requirements for performing a job, and indicates the employee's suitability (more specifically, eligibility). Qualifications include national qualifications recognized by the government, public qualifications recognized by public institutions, and internal qualifications defined in company regulations.

[0117] Qualification 908 is referenced when there are restrictions on the performance of work and tasks that fall under qualification 908. For example, if there are work and tasks that require qualification 908 for "Hazardous Materials Handling Supervisor," AI Agent 200 will determine that the employee is suitable and generate work and tasks for the employee for whom qualification 908 corresponds to "Hazardous Materials Handling Supervisor."

[0118] The preferred language 909 is the language preferred by the employee identified by employee ID 901, and indicates the employee's aptitude. The preferred language 909 is referenced when there are restrictions on the execution of tasks and duties. For example, if there are tasks and duties that require a preferred language 909 of "Vietnamese," the AI ​​agent 200 will determine that the employee is aptitude and generate tasks and duties for the employee whose preferred language 909 is "Vietnamese."

[0119] Furthermore, AI Agent 200 can present information to and receive information from an employee identified by employee ID 901, using the employee's preferred language. For example, it can propose tasks and duties in Vietnamese to an employee whose preferred language is Vietnamese, and can receive information input in Vietnamese from that employee.

[0120] <Figure 10 Attendance Information DB132> Figure 10 is an explanatory diagram showing an example of the attendance information DB132. The attendance information DB132 has the following fields: attendance date 1001, employee ID 901, employee name 902, job title 904, attendance time 1002, departure time 1003, and work type 1004. The combination of values ​​for each field in the same row constitutes an entry that defines the attendance information of an employee engaged in process Pi.

[0121] "Workday 1001" is the date (year, month, and day) when the employee identified by employee ID 901 came to work.

[0122] The clock-in time 1002 is the time when the employee identified by employee ID 901 clocked in on workday 1001, and is recorded, for example, by an operation from terminal 104.

[0123] The clock-out time 1003 is the time when the employee identified by employee ID 901 clocked out on workday 1001, and is recorded, for example, by an operation from terminal 104.

[0124] Work arrangement 1004 is the work arrangement that applies to the employee identified by employee ID 901.

[0125] When AI Agent 200 generates tasks and operations, it refers to the attendance information it has crawled to identify the current attendance status of employees.

[0126] <Figure 11 Schedule DB133> Figure 11 is an explanatory diagram showing an example of Schedule DB 133. Schedule DB 133 stores the schedules of each employee on a daily basis. Specifically, the section labeled "On-site" contains information that identifies the work of intermediate process Qj (for example, process management number 401, process name 402, intermediate process name 403, intermediate process content 404). Similarly, the section labeled "Meeting" contains information such as agenda, location, attendees, and past meeting minutes. Schedule DB 133 is updated by operations from terminal 104.

[0127] When AI Agent 200 generates tasks and operations, it refers to crawled schedules to identify employees' past, present, and future plans.

[0128] <Figure 12 Example of the functional configuration of AI agent server 110> Figure 12 is a block diagram showing an example of the functional configuration of the AI ​​agent server 110. The AI ​​agent server 110 includes an input unit 1201, a communication unit 1202, an output unit 1203, a control unit 1204, and a storage unit 1205.

[0129] The input unit 1201 receives data input from the input device 303. The communication unit 1202 sends and receives data via the communication IF 305. The output unit 1203 outputs data to the output device 304.

[0130] The control unit 1204 controls the AI ​​agent server 110 by causing the processor 301 to execute a program stored in the memory device 302. The control unit 1204 makes the AI ​​agent server 110 function as an AI agent 200. The control unit 1204 includes an information collection unit 1241 and an AI agent execution unit 1242.

[0131] [Information Gathering Unit 1241] The information gathering unit 1241, acting as an AI agent 200, crawls the process management systems 102-1 to 102-n, the employee management system 103, and the external system group 105, and stores the crawling results 1251 in the storage unit 1205.

[0132] The information gathering unit 1241 retrieves entries defining business flows from each of the business flow DBs 121 of the process management systems 102-1 to 102-n by crawling them.

[0133] The information gathering unit 1241 crawls the work instruction DB 122 of each process management system 102-1 to 102-n to obtain entries that define work instruction information from each work instruction DB 122 of each process management system 102-1 to 102-n.

[0134] The information gathering unit 1241 retrieves entries that define work report information from each of the work report DBs 123 of the process management systems 102-1 to 102-n by crawling them.

[0135] The information gathering unit 1241 crawls the regular work execution result DB 124 of each of the process management systems 102-1 to 102-n to obtain entries that define the regular work execution results from each of the regular work execution result DB 124 of the process management systems 102-1 to 102-n.

[0136] The information gathering unit 1241 retrieves entries defining non-routine work cases from each non-routine work case DB 125 of each process management system 102-1 to 102-n by crawling the non-routine work case DB 125 of each process management system 102-1 to 102-n.

[0137] The information gathering unit 1241 retrieves entries from the employee information DB 131 of the employee management system 103 that define the employee information of employees engaged in process Pi, by crawling the employee information DB 131 of the employee management system 103.

[0138] The information gathering unit 1241 retrieves entries from the employee management system 103's attendance information DB 132 that define the attendance information of employees engaged in process Pi, by crawling the attendance information DB 132 of the employee management system 103.

[0139] The information gathering unit 1241 retrieves the schedules of employees engaged in process Pi from the schedule DB 133 of the employee management system 103 by crawling the schedule DB 133 of the employee management system 103.

[0140] The information gathering unit 1241 obtains power supply information from the power information system 151 and weather information from the weather information system 152 by crawling the external system group 105.

[0141] The information gathering unit 1241 performs repeated crawling, for example, at predetermined time intervals. The information gathering unit 1241 may perform repeated crawling at the same time interval for each crawling destination, or it may perform repeated crawling at different time intervals for each crawling destination.

[0142] The information gathering unit 1241 may perform crawling on updated databases when databases within the work support system 100 and the external systems group 105 are updated.

[0143] The information gathering unit 1241 may perform crawling on databases within the work support system 100 and the external systems group 105 in response to instructions from the terminal 104.

[0144] [AI Agent Execution Unit 1242] The AI ​​agent execution unit 1242, as the AI ​​agent 200, has a natural language processing function, a business content extraction function, an employee experience value calculation function, a regular / non-regular function, a goal proposal generation function, a task proposal generation function, a related system calling function, and an answer generation function.

[0145] The natural language processing function uses a generation AI to perform morphological analysis, syntactic analysis, semantic analysis, and contextual analysis on natural language strings input from the input unit 1201 or the communication unit 1202, and passes the results to the business content extraction function, employee experience value calculation function, regular / non-regular determination function, goal proposal generation function, task proposal generation function, related system calling function, and answer generation function.

[0146] The task content extraction function uses a generating AI to extract the content of tasks that employee W should perform from the crawling results 1251. Specifically, for example, the AI ​​agent 200 checks the employee's basic information, attendance, schedule, work flow, tasks of intermediate process Qj, and messages between employees within the crawling results 1251, and also checks the instructions from employee W, which are the result of natural language processing, to identify the content of tasks that the employee should perform now, within a specified time from now, or in the future after a specified time.

[0147] The employee experience calculation function calculates employee W's experience by dividing it into routine tasks and non-routine tasks. For example, for each employee W, AI agent 200 extracts the number of times that employee has executed intermediate process Qj from the routine task execution results DB124. Specifically, for example, AI agent 200 counts the number of times that employee W's worker ID 702 and the intermediate process name 403 of the intermediate process Qj have co-occurred as the number of times that employee has executed intermediate process Qj.

[0148] Furthermore, AI agent 200 extracts the number of times each employee W has performed non-routine tasks from the non-routine task case database 125. Specifically, for example, AI agent 200 extracts the number of occurrences of worker ID 806 for that employee W as the number of times they have performed non-routine tasks. AI agent 200 also counts the number of times each employee W has performed non-routine tasks. In addition, AI agent 200 extracts the number of non-routine tasks that each employee W might have performed due to the lack of qualification 908.

[0149] Furthermore, for each employee W, AI agent 200 extracts job title 904, career history 905, work experience 906, and qualifications 908 from employee information DB 131.

[0150] Based on the information extracted in this way, AI agent 200 calculates the experience points for routine tasks and non-routine tasks for each employee W.

[0151] For example, AI agent 200 calculates a higher experience score for routine tasks the more times employee W has performed intermediate process Qj. AI agent 200 also calculates a higher experience score for routine tasks the higher the job rank 904 corresponding to the routine task is. Furthermore, AI agent 200 calculates a higher experience score for routine tasks the longer the career history 905 corresponding to that routine task is.

[0152] Furthermore, for example, AI Agent 200 calculates a higher experience score for non-routine tasks the more times it has engaged in non-routine tasks. It also calculates a higher experience score for non-routine tasks the more different types of non-routine tasks it has performed. Additionally, AI Agent 200 calculates a higher experience score for non-routine tasks the higher the job rank 904 corresponding to the non-routine tasks. Finally, AI Agent 200 calculates a higher experience score for non-routine tasks the longer the career history 905 corresponding to the non-routine tasks.

[0153] AI Agent 200 classifies employee W's skill level based on the experience points calculated for each employee W. For example, let's assume that each employee is classified into Level 1 and Level 4.

[0154] Level 1: Experience in routine tasks is below the first threshold, and experience in non-routine tasks is below the second threshold. Level 2: Experience in routine tasks is above the first threshold, and the number of unexperienced non-routine tasks (803 types) is above a specified number (simply put, experience is limited to only some non-routine tasks). Level 3: The experience points for routine tasks are above the first threshold, and the number of types 803 that cannot be performed due to lack of qualification 908 is above a certain number. Level 4: The experience points for routine tasks are above the first threshold, and the number of types of tasks (803) that can be performed is above a certain number due to having qualification 908.

[0155] Furthermore, if a Level 4 employee W is able to be present at the work site or give instructions remotely, employees W at Levels 1-3 may also be allowed to engage in "non-routine work."

[0156] The routine / non-routine task determination function refers to information crawled from the non-routine task example DB125 and determines whether the content of the tasks extracted by the task content extraction function falls under routine or non-routine tasks.

[0157] The goal proposal generation function uses a generating AI to generate goal proposals for employee W based on the content of the work extracted by the work content extraction function. For example, AI agent 200 generates a prompt asking employee W, who is performing the work extracted by the work content extraction function, what goals they should be working towards, inputs this prompt into the generating AI, and obtains the answer from the generating AI as a goal proposal.

[0158] In this case, if there is information about other processes related to process Pi in which employee W is engaged, that information will also be taken into consideration when generating the proposed goal. For example, if process Pi in which employee W is engaged is manufacturing process P4, and another process related to process Pi is maintenance process P6, then when generating the proposed goal for employee W, the maintenance period or emergency maintenance period of equipment B in manufacturing process P4 by maintenance process P6 will be taken into consideration.

[0159] For example, it becomes possible to generate goal proposals that take into account the status of multiple processes, such as "Report the status of device B to the worker in maintenance process P6 before maintenance of device B begins," "Conduct a test run of device B after maintenance of device B is completed and report the results to the worker in maintenance process P6," and "Start manufacturing lot L using device B from [Month] [Day]" (where "[Month] [Day]" is a date after the maintenance period has ended).

[0160] The task proposal generation function generates task proposals for employee W to work on, using a generating AI based on the content of the work extracted by the work content extraction function. For example, the AI ​​agent 200 generates one or more task proposals by breaking down the goal proposal into tasks, based on the work of employee W who performs the work extracted by the work content extraction function.

[0161] In this case, if there is any information related to employee W and other employees U, the task proposal for employee W will be generated taking that information into consideration. For example, if there is a "minor shutdown of equipment B", a task proposal appropriate to employee W's job title 904 (for example, "Identify the cause of the minor shutdown of equipment") is required, but this identification may require a specific qualification 908 that employee W does not possess (for example, "Hazardous Materials Handling Supervisor").

[0162] In this case, AI agent 200 retrieves the attendance information and schedule of employee U, who holds the qualification 908, from the crawling results, identifies the date △ / △ which is a date that employees W and U can accommodate, and generates a task "Identify the cause of the minor equipment downtime on △ / △".

[0163] Furthermore, other employees U may be employees belonging to a different process than employee W, or employees belonging to the same process as employee W but in a different department than employee W.

[0164] In this way, the goal proposal generation function and task proposal generation function enable the AI ​​agent 200 to generate goal proposals and task proposals by acquiring and analyzing relevant information between different processes, relevant information between employees belonging to different processes, and relevant information between employees belonging to different departments but belonging to the same process, from the crawling results collected in the storage 111. In other words, when multiple related pieces of information are acquired from the crawling results, the AI ​​agent 200 generates valid goal proposals and task proposals without changing any of the information.

[0165] The related system calling function is a function that calls other systems necessary for executing the proposed goals and tasks.

[0166] The response generation function uses a generation AI to generate responses to employee inquiries, referencing crawling results 1251.

[0167] The memory unit 1205 is a storage unit 111 that stores data input by the input unit 1201, data received by the communication unit 1202, crawling results 1251 from the information gathering unit 1241, conversation logs 1252, and execution results from the AI ​​agent execution unit 1242. The memory unit 1205 stores programs that realize the functions of the input unit 1201, the communication unit 1202, the output unit 1203, and the control unit 1204. The memory unit 1205 is composed of a memory device 302.

[0168] <Figure 13 Crawling using the work support system 100> Figure 13 is a sequence diagram showing an example of crawling by the work support system 100.

[0169] (Step S1301) The AI ​​agent server 110 monitors the timing for data collection. If it is not the right time to collect data (step S1301: No), the AI ​​agent server 110 continues monitoring. If it is the right time to collect data (step S1301: Yes), the process proceeds to step S1302.

[0170] The collection timing is, for example, a predetermined time after the previous collection timing. Alternatively, if a collection instruction is received from terminal 104, the AI ​​agent server 110 may determine that it is collection timing.

[0171] (Step S1302) The AI ​​agent server 110 performs a crawl on the employee management system 103. This allows the AI ​​agent server 110 to retrieve employee information (entries from the employee information database 131) and save it to the storage 111. The AI ​​agent server 110 may also retrieve employee information only for entries that have been updated up to the current collection timing from among the entries retrieved up to the previous collection timing.

[0172] Furthermore, the AI ​​agent server 110 retrieves attendance information (entries in the attendance information DB 132) from the attendance information DB 132 and saves it to the storage 111. The AI ​​agent server 110 may also retrieve attendance information only for entries that were retrieved up to the previous collection timing, whose current collection timing falls within the attendance day 1001, and which have been updated up to the current collection timing.

[0173] Furthermore, the AI ​​agent server 110 retrieves schedule information (entries in the schedule database 133) from the schedule database 133 and saves it to the storage 111. The AI ​​agent server 110 may also retrieve schedules only for entries that were retrieved up to the previous collection timing, and that were updated on or after the current collection timing date.

[0174] (Step S1303) The AI ​​agent server 110 performs crawling on the process management system 102-i. This allows the AI ​​agent server 110 to retrieve the business processes of intermediate process Qj (entries in the business process flow DB 121) from the business flow DB 121 and save them to the storage 111. The AI ​​agent server 110 may also retrieve only the business processes of intermediate process Qj that have been updated up to the current collection timing from the entries retrieved up to the previous collection timing.

[0175] Furthermore, the AI ​​agent server 110 retrieves work order information (entries in the work order DB 122) from the work order DB 122 and saves it to the storage 111. The AI ​​agent server 110 may also retrieve work order information only for entries that have been updated up to the current collection timing from among the entries retrieved up to the previous collection timing.

[0176] Furthermore, the AI ​​agent server 110 retrieves work report information (entries in the work report DB 123) from the work report DB 123 and saves it to the storage 111. The AI ​​agent server 110 may also retrieve work report information only for entries that have been updated up to the current collection timing from among the entries retrieved up to the previous collection timing.

[0177] Furthermore, the AI ​​agent server 110 retrieves the regular work execution results (entries in the regular work execution results DB 124) from the regular work execution results DB 124 and saves them to the storage 111. The AI ​​agent server 110 may also retrieve the regular work execution results only for entries that have been updated up to the current collection timing from among the entries retrieved up to the previous collection timing.

[0178] Furthermore, the AI ​​agent server 110 retrieves non-routine work cases (entries in the non-routine work case DB 125) from the non-routine work case DB 125 and saves them to the storage 111. The AI ​​agent server 110 may also retrieve non-routine work cases only for entries that have been updated up to the current collection timing from among the entries retrieved up to the previous collection timing.

[0179] (Step S1304) The AI ​​agent server 110 performs crawling on the external systems group 105. This allows the AI ​​agent server 110 to obtain external information such as power information and weather information from the external systems group 105. The AI ​​agent server 110 may also obtain only the external information that has been updated up to the current collection timing from the external information obtained up to the previous collection timing.

[0180] Note that the data collection timings for steps S1302 to S1304 may be the same or different. Therefore, the AI ​​agent server 110 may monitor each of steps S1302 to S1304 in step S1301.

[0181] (Step S1305) The AI ​​agent server 110 identifies non-routine tasks. Specifically, for example, the AI ​​agent server 110 refers to the comment 706 of the routine task execution result to identify the combination of the case name 802, type 803, and location 804 of the relevant non-routine task case. For example, if the similarity between the comment 706 and the combination of case name 802, type 803, and location 804 is above a threshold, the AI ​​agent server 110 identifies the task described in the comment 706 of the routine task execution result as a non-routine task.

[0182] The similarity between comment 706 and case name 802 or type 803 is, for example, the cosine similarity between an embedded representation with comment 706 embedded and an embedded representation with case name 802 or type 803 embedded. Identified non-routine tasks are stored in storage 111 in association with the results of their routine tasks.

[0183] <Figure 14: Work support processing by AI agent 200> Figure 14 is a flowchart showing an example of a work support processing procedure performed by AI agent 200.

[0184] (Step S1401) AI agent 200 monitors for work support triggers. If there is no work support trigger (step S1401: No), proceed to step S1405. If there is a work support trigger (step S1401: Yes), proceed to step S1402.

[0185] The trigger for work support is, for example, the time elapsed since the previous work support trigger. Alternatively, the trigger for work support may be the data collection timing in step S1301.

[0186] (Step S1402) AI agent 200 performs the information presentation determination process. The information presentation determination process is the process of determining the information to be presented. The information to be presented includes the proposed presentation to the employee, the timing of the presentation, and the employee to whom the proposal will be presented. The proposed presentation consists of a business plan and the task plans that comprise it. The business plan and the task plans that comprise it are the business plan listed as a candidate and one or more tasks that comprise it. One business plan consists of one or more tasks. A task is a sentence (text data) that describes one of the elements that make up the business plan. Details of the information presentation determination process will be described later in Figures 15 and 16. Since a business plan consists of one or more tasks, the proposed presentation to employee W will henceforth be referred to as a "task plan".

[0187] (Step S1403) AI agent 200 performs the execution target determination process. The execution target determination process is the process of determining what to execute. The execution target is a proposed task to be executed. Details of the execution target determination process will be described later in Figure 17.

[0188] (Step S1404) The AI ​​agent 200 executes the target execution process. The target execution process is the process that performs the target of execution in step S1403. Details of the target execution process will be described later in Figure 18. This completes the work support process by the AI ​​agent 200. Steps S1405 to S1407 are exceptions for when there are instructions from the employee.

[0189] (Step S1405) AI agent 200 determines whether there are instructions from an employee via terminal 104. If there are no instructions from an employee (step S1405: No), it returns to step S1401. If there are instructions from an employee (step S1405: Yes), it proceeds to step S1406. The employee who gave the instructions is referred to as the specific employee.

[0190] (Step S1406) AI agent 200 generates task proposals based on instructions from specific employees. The generation of task proposals based on instructions from specific employees is performed by AI agent 200.

[0191] Specifically, for example, the AI ​​agent 200 refers to the crawling results 1251 held in storage 111 during the crawling shown in Figure 13 to generate task proposals based on instructions from a specific employee. The instructions from the specific employee are text messages sent by the employee from terminal 104, and may include, for example, confirmation of the next task the employee should perform and instructions for the AI ​​agent 200 to perform the task.

[0192] In other words, the AI ​​agent 200 refers to the instructions of a specific employee, employee information, attendance information, and schedule to identify the tasks or meetings that the employee should perform in the intermediate process Qj or attend as proposed tasks, and then breaks down the identified proposed tasks into proposed work items.

[0193] For example, if AI agent 200 identifies the task Qj as a task proposal, it identifies the work instruction information and work report information related to the task Qj. AI agent 200 refers to the task Qj and the related work instruction information and work report information to break down the task Qj into one or more task proposals.

[0194] (Step S1407) AI agent 200 determines whether there are any task proposals that require urgent attention among the decomposed task proposals. Specifically, for example, AI agent 200 refers to the comments 706 of the routine work execution results of the intermediate process Qj that corresponds to the business and identifies the comments 706 that correspond to the decomposed task proposals. For example, if the similarity between the decomposed task proposals and the comments 706 of the routine work execution results of the intermediate process Qj that corresponds to the business proposal is above a threshold, AI agent server 110 identifies the comments 706 of the routine work execution results of the intermediate process Qj that corresponds to the business proposal as the comments 706 that correspond to the decomposed task.

[0195] The AI ​​agent 200 then refers to the comment 706 that corresponds to the broken-down task proposal and identifies the combination of the case name 802, type 803, and location 804 of the corresponding non-routine work case. For example, if the similarity between the comment 706 and the combination of case name 802, type 803, and location 804 is above a threshold, the AI ​​agent server 110 identifies the task proposal corresponding to the comment 706 as a task that requires urgent attention.

[0196] If there are no tasks requiring urgent attention among the decomposed task proposals (Step S1407: No), proceed to step S1701 of the execution target determination process (Step S1403). If there are tasks requiring urgent attention among the decomposed task proposals (Step S1407: Yes), proceed to step S1705 of the execution target determination process (Step S1403).

[0197] <Figure 15 Information to be presented determination process (Step S1402)> Figure 15 is a flowchart showing a detailed example of the processing steps for the information presentation determination process (step S1402).

[0198] (Step S1501) AI Agent 200 calculates experience points for each employee W using its employee experience point calculation function and classifies each employee W into one of four levels, from level 1 to level 4.

[0199] (Step S1502) The AI ​​agent 200 determines whether the unselected non-routine tasks identified in step S1305 are present in storage 111. If there are no unselected non-routine tasks in storage 111 (step S1502: No), the process proceeds to step S1601. If there are unselected non-routine task cases in storage 111 (step S1502: Yes), the process proceeds to step S1503.

[0200] (Step S1503) AI agent 200 selects one unselected non-routine task. The selected non-routine task is referred to as the selected non-routine task.

[0201] (Step S1504) AI agent 200 generates task proposals for selected non-routine tasks. Specifically, for example, AI agent 200 breaks down selected non-routine tasks into one or more task proposals, treating them as business operations.

[0202] (Step S1505) AI agent 200 determines which employees should be proposed to for one or more task proposals for selected non-routine tasks. Specifically, for example, AI agent 200 refers to job title 904, career history 905, work experience 906, job definition 907, qualifications 908, and preferred language 909 to determine which employees are capable of performing each of the one or more task proposals and are therefore proposed to the employee. For example, if an employee meets the skills required for each task proposal, that employee is determined to be a proposed employee.

[0203] Furthermore, the AI ​​agent 200 may narrow down the list of potential employees to those whose experience value for non-routine tasks calculated in step S1501 is equal to or greater than the second threshold, or whose employee W classified in step S1501 is at or above a certain level. Alternatively, the AI ​​agent 200 may narrow down the list of potential employees to those 904 with a job title of a specific rank or higher.

[0204] (Step S1506) AI agent 200 determines whether a selected non-routine task is a task that requires emergency attention. Specifically, for example, AI agent 200 determines whether the selected non-routine task corresponds to the case name 802, type 803, and location 804 of the non-routine task case DB125.

[0205] For example, if there are any non-routine work cases in the non-routine work case DB 125 where the cosine similarity between the embedded representation with the selected non-routine work embedded and the embedded representation with the case name 802, type 803, and location 804 of the non-routine work case is above a threshold, the AI ​​agent 200 may determine that the selected non-routine work is a task that requires emergency attention. Alternatively, the AI ​​agent 200 may determine that the selected non-routine work is a task that requires emergency attention by comparing it with a list of emergency response cases that have been created in advance according to predetermined requirements and are stored in the storage 111.

[0206] If it is determined that the selected non-routine task does not require emergency response (Step S1506: No), the process returns to Step S1502. If it is determined that the selected non-routine task does require emergency response (Step S1506: Yes), the process proceeds to Step S1507.

[0207] (Step S1507) AI agent 200 determines whether there are any employees who can respond urgently to selected non-routine tasks that have been identified as requiring urgent attention. Specifically, for example, AI agent 200 searches for employees who are currently working or scheduled to come to work within a specified timeframe by referring to employee information DB131, attendance information DB132, and schedule DB133.

[0208] If there is an employee at the proposed location who can respond to an emergency (Step S1507: Yes), proceed to Step S1508. If there is no employee at the proposed location who can respond to an emergency (Step S1507: No), proceed to Step S1509.

[0209] (Step S1508) The AI ​​agent 200 sends a task proposal for the selected non-routine task to the terminal 104 of the designated employee who is available to respond urgently. Then, the process returns to step S1502.

[0210] (Step S1509) AI agent 200 sends a notification of the occurrence of a selected non-routine task to the information relay employee's terminal 104. An information relay employee is an employee who is not a designated employee capable of emergency response, but who is present at the workplace responsible for process Pi where the emergency response case has occurred, and who has the highest job rank 904.

[0211] The notification for the occurrence of selective non-routine work is text data such as, "Please contact your section chief or supervisor immediately." This is because in manufacturing, when there is an urgent matter, the first report is often made to the person in charge of that workplace (section chief) or the deputy person in charge (supervisor). Then, return to step S1502.

[0212] <Figure 16 Information to be presented determination process (Step S1402)> Figure 16 is a flowchart showing a detailed example of the processing procedure for the information presentation determination process (step S1402).

[0213] (Step S1601) The AI ​​agent 200 generates task proposals for each employee. The generation of task proposals is performed by the AI ​​agent 200. Specifically, for example, the AI ​​agent 200 generates task proposals by referring to the crawling results 1251 held in storage 111 in the crawling shown in Figure 13.

[0214] In other words, the AI ​​agent 200 refers to each employee's employee information, attendance information, and schedule to break down the work plan based on the intermediate Qj tasks or meetings that the employee should attend into task plans.

[0215] For example, for a combination of employee W1 and intermediate process Qj, AI agent 200 identifies employee W1's job title 904 (manager of manufacturing process P4) and job definition 907 (supervisor). Therefore, AI agent 200 identifies, for example, the entity that will execute the task proposal of intermediate process Qj, "verify routine operations in manufacturing process P4," and breaks down the task proposal of intermediate process Qj into one or more task proposals.

[0216] (Step S1602) The AI ​​agent 200 retrieves the task proposals stored in storage 111. The task proposals stored in storage 111 are the task proposals that were previously stored in storage 111 by step S11604.

[0217] (Step S1603) AI agent 200 determines whether it is the right time to present each task proposal to the employee. Specifically, for example, AI agent 200 refers to the employee's attendance information and schedule and determines that it is the right time to present the task proposal to the employee if the employee's terminal 104 is already running. AI agent 200 also determines that it is not the right time to present task proposals to employees who do not work within the allotted time or who are not working after the allotted time has elapsed.

[0218] For task proposals that are due to be presented to employees at the appropriate time, the process proceeds to the execution target determination process (step S1403). For task proposals that are not due to be presented to employees at the appropriate time, the process proceeds to step S1604.

[0219] (Step S1604) The AI ​​agent 200 stores task proposals that do not match the task to be executed in storage 111.

[0220] (Step S1605) The AI ​​agent 200 deletes unnecessary task proposals from storage 111. For example, the AI ​​agent 200 deletes task proposals whose start date and time have expired as unnecessary task proposals. After this, the process moves to step S1401 in Figure 14, where the AI ​​agent 200 waits for a work support trigger.

[0221] <Figure 17 Execution Target Determination Process (Step S1403)> Figure 17 is a flowchart showing a detailed example of the execution target determination process (step S1403). Note that for non-routine tasks, step S1701 is unnecessary because step S1508 has already been executed.

[0222] (Step S1701) The AI ​​agent 200 sends the task determined in the information presentation determination process (step S1402) to the terminal 104 of the employee to whom the task is presented. As a result, the task is displayed on the terminal 104 of the employee to whom the task is presented.

[0223] (Step S1702) AI agent 200 receives instructions from the employee's terminal 104. These instructions include accepting or modifying tasks presented by AI agent 200.

[0224] (Step S1703) The AI ​​agent 200 determines from the terminal 104 of the recipient employee whether the instruction is accepted or not. If it is accepted (step S1703: Yes), proceed to step S1705. If it is not accepted (step S1703: No), proceed to step S1704.

[0225] (Step S1704) AI agent 200 determines whether the instruction is a modification or not. If the instruction is not a modification (step S1704: No), it returns to step S1701. If the instruction is a modification (step S1704: Yes), it returns to step S1406. This generates the work plan and dusk plan according to the modification instruction (step S1406).

[0226] (Step S1705) The AI ​​agent 200 determines which task to execute. Then, it proceeds to the execution process for the selected task (step S1404).

[0227] <Figure 18 Execution target execution process (step S1404)> Figure 18 is a flowchart showing a detailed example of the processing procedure for the execution target process (step S1404).

[0228] (Step S1801) The AI ​​agent 200 selects a controlled system based on the task to be executed and requests the selected controlled system to execute the task. The controlled system is the process management system 102-i or the group of external systems 105 specified in the task to be executed. The selected controlled system is the controlled system selected based on the task to be executed. Upon receiving the request to execute the task from the selected controlled system, the selected controlled system executes the task.

[0229] For example, if the system to be selected and controlled is the process management system 102-i, and the task to be executed is the intermediate process Qj, the process management system 102-i will execute the task for the intermediate process Qj.

[0230] Furthermore, if the system to be selected and controlled is the weather information system 152, and the task to be executed is the acquisition of weather information, the weather information system 152 will execute the task to be executed, which is part of the weather information (for example, the acquisition of typhoon forecasts).

[0231] (Step S1802) The AI ​​agent 200 retrieves execution results from the system being selected and controlled. In the example above, the AI ​​agent 200 retrieves completion notifications for tasks to be executed for the intermediate process Qj, and also retrieves typhoon information.

[0232] (Step S1803) AI agent 200 generates a response for the target employee. The response is a sentence containing the execution result from the selection control system.

[0233] (Step S1804) The AI ​​agent 200 sends the answer to the employee's terminal 104.

[0234] (Step S1805) The AI ​​agent 200 determines whether or not it has received additional instructions from the terminal 104 of the employee to whom the instructions were presented. If additional instructions are received (step S1805: Yes), the process proceeds to step S1807. If no additional instructions are received or if no additional instructions are received (step S1805: No), the process proceeds to step S1806.

[0235] (Step S1806) The AI ​​agent 200 sends text data indicating that the case is closed (for example, "The case is closed. Thank you for your hard work.") to the terminal 104 of the employee to whom the case was presented. This completes the execution process (step S1404).

[0236] (Step S1807) The AI ​​agent 200 generates task proposals based on the instructions of the target employee. The generation of task proposals based on additional instructions from the target employee is also performed by the AI ​​agent 200. Specifically, for example, the AI ​​agent 200 generates task proposals based on the instructions of the target employee by referring to the crawling results 1251 held in storage 111 in the crawling shown in Figure 13. The additional instructions from the target employee are text sent by the target employee from terminal 104, for example, instructions for further tasks that the target employee wants to perform based on the execution results.

[0237] In other words, the AI ​​agent 200 refers to the instructions of a specific employee, employee information, attendance information, and schedule to identify a proposed task based on the intermediate Qj tasks or meetings that the specific employee should attend, and then breaks down the identified task proposal into task proposals.

[0238] For example, for a combination of employee W1 and intermediate process Qj, AI agent 200 identifies employee W1's job title 904 (manager of manufacturing process P4) and job definition 907 (supervisor). Therefore, AI agent 200 identifies, for example, the entity that will execute the task proposal of intermediate process Qj, "verify routine operations in manufacturing process P4," and breaks down the task proposal of intermediate process Qj into one or more task proposals.

[0239] (Step S1808) The AI ​​agent 200 determines whether there are any tasks in the decomposed tasks that require urgent attention. Specifically, for example, the AI ​​agent 200 refers to the comments 706 of the routine work execution results of the intermediate process Qj that corresponds to the business and identifies the comments 706 that correspond to the decomposed tasks. For example, if the similarity between the decomposed tasks and the comments 706 of the routine work execution results of the intermediate process Qj that corresponds to the business is above a threshold, the AI ​​agent server 110 identifies the comments 706 of the routine work execution results of the intermediate process Qj that corresponds to the business as the comments 706 that correspond to the decomposed tasks.

[0240] The AI ​​agent 200 then refers to the comment 706 that corresponds to the broken-down task proposal and identifies the combination of the case name 802, type 803, and location 804 of the corresponding non-routine work case. For example, if the similarity between the comment 706 and the combination of case name 802, type 803, and location 804 is above a threshold, the AI ​​agent server 110 identifies the task proposal corresponding to the comment 706 as a task that requires urgent attention.

[0241] If there are no tasks requiring urgent attention among the decomposed task proposals (Step S1808: No), proceed to Step S1809. If there are tasks requiring urgent attention among the decomposed task proposals (Step S1808: Yes), proceed to Step S1705 as a task.

[0242] (Step S1809) The AI ​​agent 200 stores the task proposal in storage 111. The stored task proposal is retrieved in step S1602. This completes the execution process (step S1404).

[0243] <Figure 19 Example of display screen 1 of terminal 104> Figure 19 is an explanatory diagram showing example 1 of the display screen of terminal 104. Figure 19 shows an example of the display screen displayed on terminal 104 of the employee to whom the business and tasks are presented (e.g., employee W1, whose employee ID 901 is "0001") when there are no instructions from the employee (step S1405: No) and when there is a certain work support trigger (step S1401: Yes), and the AI ​​agent 200 does not transition to step S1502: Yes even once in the information presentation determination process (step S1402).

[0244] In the example in Figure 19, we first see an example of the AI ​​agent 200 speaking to an employee and asking a question. For example, when W1, the manager of manufacturing process P4, arrives at work in the morning and activates terminal 104, the AI ​​agent 200 asks W1 to perform tasks T1 and T2, which are breakdowns of a certain task in routine work.

[0245] (T1901) Confirmation and registration of night shift work reports (T1902) Consideration of repairs to device A

[0246] Normally, employees speak to the AI ​​agent 200, but as shown in the example in Figure 19, the AI ​​agent 200 can proactively communicate with the employee about what they need when a task support trigger occurs, by having prior knowledge of the crawling results 1251 related to the employee. Figure 19 will be explained in detail below.

[0247] The display screen 1900 is displayed on the terminal 104 of the employee (in this example, W1, whose employee ID 901 is "0001"). The display screen 1900 includes an employee information display unit 1901, a history display unit 1902, and a conversation display unit 1903.

[0248] The employee information display unit 1901 displays employee information 1910, including the employee ID 901 and employee name 902.

[0249] The history display unit 1902 displays a keyword input field 1920, a search button 1921, a chat generation button 1922, and an information display field 1923. The keyword input field 1920 is a user interface that accepts and displays keyword input via input operation. The search button 1921 is a user interface that, when pressed, searches the chat history between AI agent 200 and employee W1 using the keyword entered in the keyword input field 1920. The search results are displayed in the information display field 1923.

[0250] The chat generation button 1922 is a user interface that, when pressed, allows employee W1 to start a conversation with AI agent 200. The information display area 1923 displays search results based on keywords entered in the keyword input area 1920, or the past conversation history between AI agent 200 and employee W1.

[0251] The conversation display unit 1903 includes a message display unit 1930, a message input field 1940, and a send button 1950. The message display unit 1930 displays messages from employee W1 and messages from AI agent 200. The message input field 1940 is a user interface that accepts and displays messages from employee W1 through input operations. The send button 1950 is a user interface for sending the message entered in the message input field 1940 to the AI ​​agent server 110.

[0252] In Figure 19, the display screen 1900 is shown when the terminal 104 is activated, so the message 1931 sent from the AI ​​agent 200 is displayed at the beginning of the message display unit 1930 (steps S1701, S1702).

[0253] Employee W1 sends Message 1932 in response to Message 1931 from AI Agent 200. Message 1932 states that Employee W1 approved (Step S1703: Yes) Task T1901 and gave a modification instruction to AI Agent 200 regarding Task T1902 (Step S1704: Yes). Task T1901 (confirmation and registration of work report) is determined to be an execution target.

[0254] In this case, AI Agent 200 selects the process management system 102-4 of manufacturing process P4 as the control target system and requests the process management system 102-4 of manufacturing process P4, which is the selected control target system, to execute Task T1901 (confirmation and registration of work report) (Step S1801).

[0255] The process management system 102-4 of manufacturing process P4 executes Task T1901 (confirmation and registration of work report). The process management system 102-4 of manufacturing process P4 conducts a confirmation that Employee W1 has approved, generates a new entry in Work Report DB123, and registers the work report information regarding the work report in Work Report DB123.

[0256] On the other hand, since AI Agent 200 received a request to obtain the operation history and repair history of Device A as a modification instruction for Task T1902 (consideration of repair of Device A) (Step S1704: Yes), it generates Task Plan T3 (obtaining the operation history and repair history of Device A), which is a task plan based on the modification instruction of Employee W1 (Step S1406).

[0257] AI Agent 200 determines that it is urgent (Step S1407: Yes) and decides to make Task Plan T1903 an execution target (Step S1705).

[0258] (T1903) Obtaining the operation history and repair history of Device A

[0259] In this case, the AI ​​agent 200 selects the process management system 102-4 of the manufacturing process P4 and the process management system 102-6 of the maintenance process P6 as the systems to be controlled. Of the proposed task T1903 (acquisition of operation history and repair history for equipment A), it requests the process management system 102-4 of the manufacturing process P4 to acquire the operation history for equipment A, and requests the process management system 102-6 of the maintenance process P6 to acquire the repair history for equipment A (step S1801).

[0260] The process control system 102-4 in the manufacturing process P4 retrieves the operating history of equipment A by searching its own routine work execution results DB124. The process control system 102-6 in the maintenance process P6 also retrieves the repair history of equipment A by searching its own routine work execution results DB124.

[0261] The AI ​​agent 200 obtains the operating history of device A from the process management system 102-4 in the manufacturing process P4, and the repair history of device A from the process management system 102-6 in the maintenance process P6 (step S1802).

[0262] The AI ​​agent 200 generates message 1933 using the acquired operating history and repair history of device A as a response to employee W1, the employee to whom the message was presented (step S1803), and sends it to employee W1's terminal 104 (step S1804). As a result, message 1933 is displayed on the message display unit 1930.

[0263] Employee W1 sends message 1934 in response to message 1933 from AI agent 200. Message 1934 is an additional instruction requesting consultation regarding the need for repairs to device A during the next scheduled maintenance (step S1805: Yes).

[0264] The AI ​​agent 200 generates task proposal T1904 as a proposed task and business plan based on the additional instructions (step S1807).

[0265] (T1904) Request to consult with the maintenance department

[0266] Furthermore, the AI ​​agent 200 obtains the date and time of the next scheduled maintenance from the business flow DB 121 of the process management system 102-6 of maintenance process P6 and determines its urgency (step S1808). Regarding the request to consult with the maintenance department, it determines that it is not urgent (step S1808: No) and stores the proposed task T1904 (request to consult with the maintenance department) based on the additional instruction in storage 111 (step S1809).

[0267] The AI ​​agent 200 then sends message 1935 to employee W1's terminal 104, indicating that it has received the proposed task T1904 (request to consult with the maintenance department) based on the additional instructions. As a result, message 1935 is displayed on the message display unit 1930.

[0268] As shown in the example in Figure 19, the AI ​​agent 200 can autonomously decide when and to which employee W to present what task, and present the appropriate task to the appropriate employee W at the appropriate time.

[0269] <Figure 20 Example of display screen 2 for terminal 104> Figure 20 is an explanatory diagram showing example 2 of the display screen of terminal 104. Figure 20 shows an example of the display screen displayed on terminal 104 of an employee when, without instructions from the employee (step S1405: No), and at a certain work support trigger (step S1401: Yes), the AI ​​agent 200 determines in the presentation information determination process (step S1402) that there is an unroutine task (step S1502: Yes), and the employee who can urgently respond to the task for the unroutine task is the employee with employee ID 901 "0001" (e.g., W1) (steps S1505, S1507: Yes).

[0270] In the example in Figure 20, similar to Figure 19, the AI ​​agent 200 first asks a question to the employee. For example, when the AI ​​agent 200 receives a plan for temporary maintenance of equipment C in manufacturing process P4 from the process management system 102-6 of maintenance process P6, it generates task T2001, which requests approval for the task goal G2001 from W1, the manager of manufacturing process P4.

[0271] (G2001) Calculate the impact of temporary maintenance on manufacturing process P4. (T2001) Request for approval of Goal G2001 from Section Chief W1 of Manufacturing Process P4

[0272] Normally, employees speak to the AI ​​agent 200, but as shown in the example in Figure 20, by having prior knowledge of the crawling results 1251 related to the employee, the AI ​​agent 200 can initiate a conversation with employee W about what they need at a given work support trigger. Figure 20 will be explained in detail below.

[0273] In Figure 20, unlike in Figure 19, terminal 104 is already activated, but it is assumed that message 2031 has been sent from AI agent 200 (step S1508).

[0274] Employee W1 sends message 2032 in response to message 2031 from AI agent 200. Message 2032 is the response to task T2001, stating that employee W1 has accepted goal G2001 (step S1703: Yes).

[0275] In this case, the AI ​​agent 200 selects the process management system 102-4 of manufacturing process P4 as the system to be controlled and requests the execution of goal G2001 (calculate the impact of temporary maintenance on manufacturing process P4) from the process management system 102-4 of manufacturing process P4, which is the selected system to be controlled (step S1801).

[0276] The process management system 102-4 of manufacturing process P4 executes Goal G2001 (calculate the impact on manufacturing process P4 due to ad-hoc maintenance) based on the ad-hoc maintenance plan of device C in manufacturing process P4 from the process management system 102-6 of maintenance process P6. Regarding the specific process of calculating the impact on manufacturing process P4 due to ad-hoc maintenance by referring to the ad-hoc maintenance plan of device C, that is, how the impact calculation is executed by what calculation formula and which information in the ad-hoc maintenance plan of device C is used as the argument for the variables in the calculation formula, it is assumed that it is set to be executable for the process management system 102-4 of manufacturing process P4. As a result of executing Goal G2001, AI agent 200 acquires impact 1 and impact 2 from the process management system 102-4 of manufacturing process P4 (step S1802).

[0277] AI agent 200 generates message 2033 including impact 1 and impact 2 as a response to employee W1, who is the recipient (step S1803), and transmits it to the terminal 104 of employee W1 (step S1804). Thereby, message 2033 is displayed on message display unit 1930.

[0278] Employee W1 transmits message 2034 in response to message 2033 from AI agent 200. Message 2034 indicates that employee W1 has approved the result of Goal G2001 (calculate the impact on manufacturing process P4 due to ad-hoc maintenance), and is an additional instruction indicating a request to generate a work instruction including reconfirmation of process conditions, a request to send and confirm to employee W3 (step S1805: Yes).

[0279] AI agent 200 generates task plan T2002 as a task plan based on the additional instruction (step S1807).

[0280] (T2002) Request to generate a work instruction including reconfirmation of process conditions, request to send and confirm to employee W3

[0281] The AI ​​agent 200 determines that there is no urgency (step S1808: No) and stores the proposed task T2002 based on the additional instructions (a request to generate a work order including a reconfirmation of process conditions, and a request to send it to employee W3 and have it confirmed) in storage 111 (step S1809).

[0282] The AI ​​agent 200 then sends a message 2035 to employee W1's terminal 104 indicating that it has received the proposed task T2002 (a request to generate a work order including a reconfirmation of process conditions, and a request to send it to employee W3 and have it confirmed) based on the additional instructions. As a result, message 2035 is displayed on the message display unit 1930.

[0283] As shown in the example in Figure 20, the AI ​​agent 200 can autonomously decide when and to which employee W to present a task related to non-routine work, and can immediately present an appropriate task related to urgent non-routine work to the appropriate employee W.

[0284] <Figure 21 Example of display screen 3 for terminal 104> Figure 21 is an explanatory diagram showing example 3 of the display screen of terminal 104. Figure 21 shows an example of the display screen displayed on employee W1's terminal 104 when an employee with employee ID 901 "0001" (e.g., W1) has given instructions (step S1405: Yes), the AI ​​agent 200 generates a work plan and task plan based on the employee's instructions (step S1406), and determines that the task plan is urgent (step S1407: Yes).

[0285] In the example in Figure 21, the first example is when employee W1 asks a question to AI agent 200. For example, W1, who is the manager of manufacturing process P4, instructs AI agent 200 to go to goal G2101 based on the weather information in crawling results 1251.

[0286] (G2101) Calculation of Lightning Strike Probability

[0287] AI agent 200, recognizing that goal G2101 contains terms such as "lightning strike" and "probability of lightning strike," selects the weather information system 152 as the target of the task request and generates a proposed task T2101 requesting the weather information system 152 to execute goal G2101 (step S1405: Yes, S1406). Goal G2101, which is an instruction from employee W1, is displayed as message 2131.

[0288] (T2101) Request to execute goal G2101 to weather information system 152

[0289] In this example, since goal G2101 is urgent (step S1407: Yes), AI agent 200 decides to execute goal G2101 (step S1705).

[0290] In this case, the AI ​​agent 200 selects the weather information system 152 as the system to be controlled, executes the proposed task T2101, and requests the weather information system 152, which is the selected system to be controlled, to perform goal G2101 (calculation of the probability of lightning strikes) (step S1801).

[0291] The weather information system 152 executes goal G2101 (calculation of lightning strike probability). The weather information system 152 calculates today's lightning strike probability at the site where manufacturing process P4 is carried out.

[0292] The AI ​​agent 200 obtains today's lightning strike probability from the weather information system 152 as the result of executing task proposal T2101 (a request to execute goal G2101 to the weather information system 152) (step S1802).

[0293] The AI ​​agent 200 generates a message 2132 containing today's lightning strike probability as a response to employee W1 (step S1803) and sends it to employee W1's terminal 104 (step S1804). As a result, message 2132 is displayed on the message display unit 1930.

[0294] Employee W1 sends message 2133 in response to message 2132 from AI agent 200. Message 2133 is an additional instruction indicating goal G2102, which is to derive a response plan in case of a power outage (step S1805: Yes).

[0295] (G2102) Derivation of countermeasures in the event of a power outage

[0296] Since goal G2102 includes terms such as "power outage" and "countermeasures," AI agent 200 selects the process management system 102-6 of maintenance process P6 as the target of the task request and generates a proposed task T2102 that requests the process management system 102-6 of maintenance process P6 to execute goal G2102 (step S1807).

[0297] (T2102) Execution request for goal G2102 to process control system 102-6 of maintenance process P6

[0298] AI agent 200 determines that task proposal T2102 is urgent because message 2133 contains the phrase "urgently, all at once" (step S1808: Yes), and decides to execute task proposal T2102 (step S1705).

[0299] In this case, the AI ​​agent 200 selects the process management system 102-6 of maintenance process P6, which is the recipient of the task proposal T2102 that has been determined to be executed, as the controlled system, executes the task proposal T2102 that has been determined to be executed, and requests the process management system 102-6 of maintenance process P6, which is the selected controlled system, to perform goal G2102 (derivation of countermeasures in the event of a power outage) (step S1801).

[0300] The process management system 102-6 of maintenance process P6 executes goal G2102 (deriving countermeasures in case of a power outage).

[0301] The AI ​​agent 200 obtains the following countermeasures 1 to 3 as the execution result of the proposed task T2102, which was determined to be executed from the process management system 102-6 of the maintenance process P6 (step S1802).

[0302] Countermeasure 1: Turn off the power to equipment that is waiting for a batch to be processed in preparation for a power outage. Countermeasure 2: Device C, which will be severely affected by the power outage, will have its lot loading suspended after 3 PM and will be put into an idle state. Countermeasure 3. In the event of a power outage that causes device C to stop, restart device C according to the recovery manual.

[0303] The AI ​​agent 200 generates a message 2134 containing the above countermeasures 1 to 3 as a response to employee W1 (step S1803) and sends it to employee W1's terminal 104 (step S1804). As a result, message 2134 is displayed on the message display unit 1930.

[0304] Employee W1 sent a message 2135 to AI agent 200 acknowledging countermeasures 1-3, so AI agent 200 determined that there were no further instructions (step S1805: No), and sent a message 2136 indicating completion to employee W1's terminal 104. As a result, message 2136 is displayed on the message display unit 1930.

[0305] As shown in the example in Figure 21, the AI ​​agent 200 can determine when and to which employee W to present a task related to non-routine work in response to instructions from employee W, and can immediately present an appropriate task related to urgent non-routine work to the appropriate employee W.

[0306] <Figure 22 Example of display screen 4 of terminal 104> Figure 22 is an explanatory diagram showing example 4 of the display screen of terminal 104. Figure 22 shows an example of the display screen displayed on employee W1's terminal 104 when an employee with employee ID 901 "0001" (e.g., W1) has given instructions (step S1405: Yes), the AI ​​agent 200 generates a work plan and task plan based on the employee's instructions (step S1406), and determines that the task plan is not urgent (step S1407: No).

[0307] In the example in Figure 22, similar to Figure 21, the first example is when employee W1 speaks to AI agent 200. For example, W1, the manager of manufacturing process P4, based on the scheduled maintenance of equipment C for tomorrow from the process management system 102-6 of maintenance process P6 in the crawling results 1251, sends message 2231, including goal G2201, to AI agent 200 from terminal 104 as an instruction from employee W1. Message 2231 is displayed on message display unit 1930.

[0308] (G2201) Generation of a draft work order considering the state of device C immediately after scheduled maintenance.

[0309] Then, the AI ​​agent 200 receives the instruction for goal G2201 from employee W1 (step S1405: Yes) and generates a task proposal T2201 based on that instruction (step S1406). It also retrieves employee W1's task proposal T2202, which was stored in storage 111 because it was not a work support trigger.

[0310] (T2201) Execution request for goal G2201 (T2202) Request for confirmation of today's daytime work report from employee W1

[0311] In this example, the AI ​​agent 200 determines that task proposals T2201 and T2202 are not urgent (step S1407: No) because message 2231 contains the phrases "tomorrow" and "consider the state of device C immediately after maintenance," and sends message 2232, which designates task proposals T2201 and T2202 as tasks, to employee W1's terminal 104 (step S1701). Message 2232 is displayed on the message display unit 1930.

[0312] AI agent 200 receives message 2233 from employee W1's terminal 104 (step S1702). Message 2233 indicates acceptance of proposed tasks T2201 and T2202 (step S1703: Yes) and an instruction to execute goal G2201 (proposed task T2201). Therefore, AI agent 200 decides to execute proposed task T2201 (step S1705). AI agent 200 also sends message 2234 to employee W1's terminal 104 to acknowledge the contents of message 2233.

[0313] In this case, the AI ​​agent 200 selects the process management system 102-4 of manufacturing process P4 as the system to be controlled, executes task proposal T2201 (request to execute goal G2201), and requests the process management system 102-4 of manufacturing process P4, which is the selected system to be controlled, to perform goal G2201 (generate a draft work instruction considering the state of equipment C immediately after periodic maintenance) (step S1801).

[0314] The process control system 102-4 of manufacturing process P4 refers to the work instruction DB122 and executes goal G2201 (generate a draft work instruction considering the state of equipment C immediately after periodic maintenance).

[0315] The AI ​​agent 200 obtains a draft work order as the result of executing task proposal T2201 from the process management system 102-4 of manufacturing process P4 (step S1802).

[0316] The AI ​​agent 200 generates a message 2235 containing the URL where the acquired draft work order is saved as a response to employee W1 (step S1803) and sends it to employee W1's terminal 104 (step S1804). As a result, message 2235 is displayed on the message display unit 1930.

[0317] Employee W1 sends message 2236 in response to message 2235 from AI agent 200. Message 2236 confirms the contents of Task Proposal T2202 (request for employee W1 to confirm today's daytime work report) and the execution result of Task Proposal T2201 (draft work instruction). Since there are no additional instructions in message 2236 (step S1805: No), AI agent 200 sends completion message 2237 to employee W1's terminal 104 (step S1806). As a result, message 2237 is displayed on the message display unit 1930.

[0318] As shown in the example in Figure 22, the AI ​​agent 200 can determine when and to which employee W to present what task in response to instructions from employee W, and present the appropriate task to the appropriate employee W at the appropriate time.

[0319] As explained above, this embodiment makes it possible to optimize the timing, content, and recipient of information presentation. Specifically, for example, by using all the information necessary for decision-making without any omissions, it becomes possible to perform tasks at a consistent level, adjusted to the appropriate timing in various on-site work scenarios, regardless of the employee W's skill level. Furthermore, repeated use improves the experience of both the work support system 100 and the employee W.

[0320] In the above-described embodiment, the AI ​​agent 200 and terminal 104 communicated using text data, but voice data may also be used. In this case, the AI ​​agent server 110 and terminal 104 convert the input voice data into text data for input, and convert the text data back into voice data for output. Furthermore, the AI ​​agent 200 may input not only text, but also still images and videos to the generating AI.

[0321] It should be noted that the present invention is not limited to the embodiments described above, but includes various modifications and equivalent configurations within the spirit of the attached claims. For example, the embodiments described above are described in detail to make the present invention easier to understand, and the present invention is not necessarily limited to having all of the described configurations. Furthermore, some of the configurations of one embodiment may be replaced with those of another embodiment. Furthermore, some of the configurations of one embodiment may be added to those of another embodiment. Furthermore, some of the configurations of each embodiment may be added, deleted, or replaced with other configurations.

[0322] Furthermore, each of the aforementioned configurations, functions, processing units, and processing means may be implemented in hardware, for example, by designing them as integrated circuits, or they may be implemented in software by having a processor interpret and execute programs that realize each function.

[0323] Information such as programs, tables, and files that implement each function can be stored in memory, hard disks, SSDs (Solid State Drives), or on recording media such as IC (Integrated Circuit) cards, SD cards, and DVDs (Digital Versatile Discs).

[0324] Furthermore, the control lines and information lines shown are those deemed necessary for explanation purposes and do not necessarily represent all control lines and information lines required for implementation. In reality, it can be assumed that almost all components are interconnected. [Explanation of Symbols]

[0325] 100 Work Support Systems 101 AI Agent System 102 Process Management System 103 Employee Management System 104 terminals 105 External Systems 106 Network 110 AI Agent Servers 120 Process Management Server 130 Employee Management Server 151 Power Information System 152 Weather Information System 200 AI Agents 121 Business Flow Database 122 Work Instructions Database 123 Work report DB 124. Database of Routine Task Execution Results 125 Non-routine work example database 131 Employee Information Database 132 Attendance Information Database 133 Schedule DB 134 Message DB

Claims

1. A management device having a processor for executing a program, a storage device for storing the program, and a communication interface capable of communicating with a group of target computers, which is a collection of target computers for each of the multiple users, and a group of monitored systems, which is a collection of monitored systems used by at least one of the multiple users, The aforementioned processor, A data collection process that collects information from the aforementioned group of monitored systems, Based on the collection results from the collection process, a generation process generates a task that the user should perform and a presentation timing for presenting the task. A transmission process that sends the task to the user's destination computer at the aforementioned presentation timing, A management device characterized by performing the following actions.

2. A control device according to claim 1, The aforementioned processor, The collection process, the generation process, and the transmission process are repeatedly executed. A control device characterized by the following features.

3. A control device according to claim 1, The group of monitored systems includes a first monitored system that manages the suitability of the multiple users, In the collection process, the processor collects the user's suitability from the first monitored system. In the generation process, the processor generates the task based on the user's suitability. A control device characterized by the following features.

4. A control device according to claim 3, The first monitored system maintains the suitability of the multiple users in a way that can be updated over time. In the collection process, the processor collects the user's latest suitability from the first monitored system. In the generation process, the processor generates the task based on the user's latest aptitude. A control device characterized by the following features.

5. A control device according to claim 1, The group of monitored systems includes a first monitored system that manages the current status of the multiple users, In the collection process, the processor collects the user's current status from the first monitored system. In the generation process, the processor generates the presentation timing based on the user's current status. A control device characterized by the following features.

6. A control device according to claim 5, The first monitored system maintains the status of the multiple users in a way that allows it to be updated over time. In the collection process, the processor collects the latest status of the user from the first monitored system. In the generation process, the processor generates the task based on the user's current status. A control device characterized by the following features.

7. A control device according to claim 1, The group of monitored systems includes a first monitored system that manages the schedules of the multiple users, In the collection process, the processor collects the user's schedule from the first monitored system. In the generation process, the processor generates the task and the presentation timing based on the user's schedule. A control device characterized by the following features.

8. A control device according to claim 7, The first monitored system maintains the schedules of the multiple users in a way that allows them to be updated over time. In the collection process, the processor collects the user's latest schedule from the first monitored system. In the generation process, the processor generates the task based on the user's latest schedule. A control device characterized by the following features.

9. A control device according to claim 3, The group of monitored systems includes a second monitored system that manages the tasks performed by the user, In the collection process, the processor collects the management results of the work performed by the second monitored system from the second monitored system. In the generation process, the processor generates tasks that the user can handle based on the user's aptitude and the management results. A control device characterized by the following features.

10. A control device according to claim 9, In the generation process, the processor generates the presentation timing based on the urgency of the management result. A control device characterized by the following features.

11. A control device according to claim 9, The second monitored system holds information regarding non-routine operations, In the generation process, if the management result corresponds to the non-routine operation, the processor sets the presentation timing to the timing for immediate transmission by the transmission process. A control device characterized by the following features.

12. A control device according to claim 1, In the generation process, the processor generates the task based on a plurality of related pieces of information collected by the collection process. A control device characterized by the following features.

13. A control device according to claim 1, The aforementioned processor, Based on instructions from the destination computer to which the task was sent, the process of selecting a controlled system from the group of monitored systems, controlling the process executed by the controlled system to obtain the processing result by the controlled system, and transmitting the processing result to the destination computer to which the task was sent, A management device characterized by performing the following actions.

14. A management method executed by a management device having a processor for executing a program, a storage device for storing the program, and a communication interface capable of communicating with a group of target computers which is a collection of target computers for each of a group of users and a group of monitored systems which is a collection of monitored systems used by at least one of the group of users, The aforementioned processor, A data collection process that collects information from the aforementioned group of monitored systems, Based on the collection results from the collection process, a generation process generates a task that the user should perform and a presentation timing for presenting the task. A transmission process that sends the task to the user's destination computer at the aforementioned presentation timing, A management method characterized by performing the following actions.

15. The processor of a management device having a processor for executing a program, a storage device for storing the program, and a communication interface capable of communicating with a group of target computers which is a collection of target computers for each of the multiple users, and a group of monitored systems which is a collection of monitored systems used by at least one of the multiple users, A data collection process that collects information from the aforementioned group of monitored systems, Based on the collection results from the collection process, a generation process generates a task that the user should perform and a presentation timing for presenting the task. A transmission process that sends the task to the user's destination computer at the aforementioned presentation timing, A management program characterized by causing the execution of a command.

Images