A method for executing and managing tasks using a generative AI and a predetermined task management system, a device and program directly used for the implementation, and an integrated task management and execution system using generative AI

Abstract

Based on the premise that generative AI will be used as a unified tool, rather than using various tools for various tasks, we provide a method that simplifies the handling of input and output in generative AI and enables efficient execution and management of tasks using generative AI, as well as devices, equipment, and programs that are directly used to implement the method, and an integrated task management and execution system that uses generative AI. [Solution] The method includes a first step in which the generating AI learns the predetermined knowledge required to execute and manage tasks, and a second step in which input data obtained from an input location in a predetermined task management system inferred by the generating AI is input to the generating AI, and output data from the generating AI is stored in an output location in the predetermined task management system inferred by the generating AI.

Description

[Technical Field]

[0001] The present invention relates to a method for executing and managing tasks using a generative AI and a predetermined task management system, a device and program directly used to implement the method, and an integrated task management and execution system using generative AI. [Background technology]

[0002] Even before the advent of generative AI, there were management issues in business operations, such as improving QCD (Quality, Cost, Delivery) at an organizational level, stimulating communication, and visualizing work, and to solve these issues, there were task management and project management methods, as well as information systems and tools to support them (for example, Patent Document 1).

[0003] On the other hand, since the emergence of ChatGPT, a generative AI service, in 2022, attempts have been made to use generative AI to automatically perform specific tasks and improve their efficiency (e.g., Patent Document 2).

[0004] When utilizing generative AI, various contracts and generative AI models were individually introduced within the organization (see Figure 11). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2020-003889 [Patent Document 2] Patent No. 7520336 Summary of the Invention [Problem to be solved by the invention]

[0006] However, while the incredible advances in generative AI have improved the speed and quality of generation itself, the handling of the inputs and outputs required for generation remains inefficient in many cases.

[0007] That is, as shown in Figure 12, users must find the appropriate inputs and do them manually or by copy and paste. The output also needs to be moved to other tools such as Excel by copy and paste.

[0008] The object of the present invention is to provide a method for executing and managing tasks using generative AI and a specified task management system, which simplifies the handling of input and output when using generative AI in business operations without the need for various contracts or various generative AI models, and enables tasks using generative AI to be executed and managed efficiently in business operations, as well as an apparatus and program directly used to implement the method, and an integrated task management and execution system that uses generative AI. [Means for solving the problem]

[0009] In order to achieve the above object, the method of executing and managing tasks using a generation AI and a predetermined task management system according to the present invention comprises: and, Predefined task management system a chatbot application between the generation AI and the predetermined task management system, the chatbot application being provided for dialogue with a user and for execution of tasks by the generation AI; and an adapter function unit that enables the predetermined task management system to be operated from the chatbot application. A method for executing and managing a task using the above-mentioned method, wherein the generation AI determines an input location where input data of the generation AI is input and an output location where output data of the generation AI is stored in the predetermined task management system. As a logical repository a first step of learning predetermined knowledge necessary for the execution and management of said task so that it can be reasoned about; The aforementioned Chatbot Applications and the adapter function unit of Through When performing a task, The generating AI, Input data obtained from an input location in the predetermined task management system inferred by the generation AI. of Input to the generated AI deathOutput data from the generating AI to an output location in the predetermined task management system inferred by the generating AI. of Store do The second step is The predetermined knowledge is a first knowledge that is knowledge specific to the task, a second knowledge that is knowledge for inferring the location of the input data and the output data as the logical storage destination in the predetermined task management system, and a third knowledge that is knowledge related to execution for physically acquiring the input data and physically storing the output data in the predetermined task management system. It is characterized by:

[0010] The predetermined knowledge is: The method is characterized in that the generating AI additionally learns any missing parts of the standard knowledge of the generating AI.

[0013] Another device according to the present invention is a device directly used to implement the above method, characterized in that it comprises the generation AI in which the specified knowledge has been learned, and the chatbot application that is provided between the generation AI in which the specified knowledge has been learned and the specified task management system.

[0014] Another device according to the present invention is a device directly used to implement the above method, and is characterized by having the generation AI that has learned the predetermined knowledge, the predetermined task management system, and the chatbot application that is provided between the generation AI that has learned the predetermined knowledge and the predetermined task management system.

[0015] Furthermore, the program according to the present invention is a program directly used to implement the above method, The first and second steps are executed by a computer. [Effects of the Invention]

[0017] According to the present invention, it is possible to use the generation AI as a unified tool, without using various tools for various tasks, and to simplify the handling of input and output in the generation AI, thereby enabling efficient execution and management of tasks using the generation AI. (action) In this invention, when using generative AI, the storage destination for the inputs and outputs required for generation (task execution) exists in the task management system. Task management systems were created to solve the management issues mentioned above, and their basic data units, such as tasks, activities, and tickets, almost always have locations and attributes for storing data. Even if there are no appropriate locations or attributes, they often have a function to set fields (i.e., items and attributes) that are additionally defined based on user-specific requirements, such as so-called custom fields.

[0018] Therefore, if the appropriate fields exist within the specific task in question or other related specific tasks, and there is a way to manipulate them, the input and output required to utilize the above-mentioned generating AI can be obtained from or output to that field.

[0019] Furthermore, the generation AI is given the knowledge to determine the location and location of such fields, retrieve data from the fields, and perform operations to store data in the fields. This allows the person in charge of each specific task to execute generation using the given input and save the output in the appropriate location without having to be particularly aware of the task management system. This significantly reduces the burden on the user associated with input and output, allowing the user to concentrate on so-called prompts to qualitatively refine the output results, and other tasks that lead to tangible results.

[0020] This also allows the task management system to capture in real time the progress, status, and working hours of individual tasks, as well as the impact of delays on other individual tasks, resulting in qualitative improvements that contribute to the task management system's original operational purpose of solving management issues from a more macro perspective. [Brief explanation of the drawings]

[0021] [Figure 1] (a) is a diagram showing the generation cycle of the task management and execution system, and (b) is a flowchart showing an example of the conversation between the user and the generation AI and the activities in the backstage. [Figure 2] This is a diagram showing the relationship between the input to the generation AI in the task management and execution system and the data storage in a specified task management system regarding the output from the generation AI. [Figure 3] 10A is a diagram showing a table for managing the context of task execution, and FIG. 10B is a diagram showing the context with dashed arrows. [Figure 4] FIG. 1 illustrates the relationships between multiple tasks in a project. [Figure 5] (a) shows the translation tasks in a translation project, and (b) shows a larger example. [Figure 6] This is a diagram showing input positions and output positions within a given task management system inferred by a generation AI. [Figure 7] FIG. 10 illustrates branching task generation. [Figure 8] FIG. 10 is a diagram illustrating confluence-type task creation. [Figure 9] This is an explanatory diagram of how long task flows can be achieved by accumulating small task management and execution steps, rather than aiming for massive automation of task management and execution systems all at once. [Figure 10] This is a diagram showing how an information terminal such as a PC is connected to a server, a generation AI, and a predetermined task management system via a network. [Figure 11] This is a diagram that explains the current situation in which various AI tools and AI platforms are used for various tasks. [Figure 12] FIG. 1 illustrates a conventional situation where the handling of inputs and outputs required for production remains inefficient. DETAILED DESCRIPTION OF THE INVENTION

[0022] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

[0023] (First embodiment) Before explaining an embodiment of the present invention relating to a method for executing and managing tasks using a generative AI and a predetermined task management system using Figures 1 and 2, we will first explain tasks, the relationships between tasks, and auxiliary information such as users, project descriptions, and attributes of tasks and context.

[0024] (task) A task is a job that a person or organization performs, and in task management systems, tasks are generally given the following attributes:

[0025] ID (unique number) Name Description Creator · manager Planned start date Planned end date Actual Start Date Actual Finish Date Planned man-hours Actual man-hours (Relationship between tasks) Furthermore, in many cases, there are hierarchical or pre- and post-task relationships between tasks. A hierarchical relationship is a hierarchy between tasks, with the higher-level task being called the parent task and the lower-level task being called the child task, and there is a one-to-many relationship between them. This relationship is used for groups of tasks that share something in common.

[0026] Also, unlike hierarchical relationships, ante-post relationships are horizontal relationships, and are information about the ante-post relationship or execution order of tasks, and can be managed in a table like the one shown in Figure 3(a). These relationships are shown in Figure 3(b) (the dashed arrows indicate ante-post relationships).

[0027] (Additional information) Additionally, information such as users and projects is managed as auxiliary information.

[0028] (1) User Users who are authorized to use the task management system are managed by user ID. There is a relationship between users and tasks, such as "responsible person" or "creator."

[0029] (2) Project As a group of tasks, projects are managed by project ID, and a project has multiple tasks.

[0030] Although the management units of the above information and their relationships differ depending on the task management system, in most cases a similar information structure is used. An example of such a standard is the Project Management Body of Knowledge (PMBOK) published by the Project Management Institute (PMI). This standard explains the four types of task context mentioned above.

[0031] The above relationship is shown in Figure 4.

[0032] (task and context attributes) As mentioned above, tasks are assigned attributes, but contexts are also assigned attributes that describe the nature of the relationship. Tasks and contexts are not always assigned standard attributes in the task management system, but in many cases, users can use their own custom attributes (custom fields).

[0033] In this embodiment, the standard attributes of the task management system and the above-mentioned custom fields are used as the input acquisition location and output storage location for tasks that utilize generation AI.

[0034] To minimize the burden on users when operating a chatbot application, it is desirable to be able to carry out tasks using the generation AI with as little conversation as possible. To achieve this, the generation AI needs to have the following three pieces of knowledge: (1) Business knowledge to properly execute the generation (A) General business knowledge unrelated to the creation of the work (a) Knowledge for generating information such as input interpretation and output format. (2) The knowledge required to determine the input and output locations on the task management system for generation, i.e., the tasks to be input or output and their attributes. (3) Knowledge to retrieve data from an input location on the task management system and store data in an output location on the task management system. In the following, to explain how to utilize this knowledge (2), we will show as an example a business process in which a Japanese sentence is created, translated into English, and then the English sentence is checked by a native speaker.

[0035] To execute this work with the simplest task structure, one translation project is set up as shown in Figure 5(a), and translation tasks are managed within it (translation tasks 100 and 101). Translation tasks have a status attribute, with five status values: new, Japanese input completed, English translation completed, checked, and completed.

[0036] Here, as an example of (2) above, it is possible to have the following knowledge learned: 1. When a user selects a translation job, the system retrieves the translation project and presents tasks managed in that project that have a status of "Japanese input completed." 2. If the user agrees to perform a specific task among the presented tasks, the details of that task are displayed. 3. When the user instructs the task to be executed, i.e., to translate it into Japanese, the Japanese text is retrieved from the "Japanese Text Custom Field" of the task and translated into English. 4. Once the English translation is complete, present it to the user. 5. Once the user is satisfied with the English translation, store the English text in the "English Text Custom Field." 6. If there are other tasks with the status "Japanese input completed", present them and ask whether you want to continue. If you want to continue, go to 2. If not, exit.

[0037] The knowledge in (1)(a) above is knowledge of Japanese-English translation, and the knowledge in (1)(b) is knowledge such as "translating into concise, business-oriented Japanese."

[0038] Also, as for (3) above, the syntax of the command to execute 1. of (2) is: You will be given knowledge such as get_tasks(status="Japanese input completed").

[0039] Next, as a larger-scale example, as shown in Figure 5(b), we will explain the case where the above English translation task is broken down into the following three tasks. The advantage of breaking it down like this is that it becomes possible to assign budget information such as the planned start date and planned man-hours to each of the three tasks, allowing for progress management and budget / actual management. It also becomes possible to assign individual people to each task; for example, a native checker could be assigned to someone living in the United States.

[0040] In this configuration, you can also use a custom field called Type or a standard attribute to identify the type of task. For tasks of type "native check", a text field is not provided because it is not necessary.

[0041] In this setting, the learning content of (2) above will be explained. 1. If the user selects a translation job, the system retrieves the translation project and presents tasks managed in that project that are of type "English translation" and have a status of "new" (e.g., translation task 101). 2. If the user agrees to perform a specific task among the presented tasks, the details of that task are displayed. 3. When the user instructs the execution of the task, i.e., translation into English, the task corresponding to the previous process of the task in question (e.g., translation task 100) is retrieved using the context, and the text (Japanese) is retrieved from its "Text Custom Field" and translated into English. 4. Once the English translation is complete, present it to the user. 5. Once the user is satisfied with the English translation, store the English text in the "Text Custom Field" of the task in question (e.g., Translation Task 101). 6. If there are other tasks with the type "English translation" and the status "New", present them and ask whether you want to continue. If you do, go to 2. If you do not want to continue, exit.

[0042] (Note that we have assumed here that the task management system's functionality ensures that if translation task 101 exists, the previous task is "completed" and valid Japanese text is entered in its text custom field.) As can be seen from this example, the input and output of a specific generative AI-based task are not necessarily included in the same task. For example, if the generative AI-based task is to create a proposal in response to a customer request, multiple tasks equivalent to the output of the task corresponding to that task will be generated according to the chapter structure of the proposal, and the task will be used to generate the specific content of each chapter.

[0043] Hereinafter, an embodiment of the present invention will be described with reference to FIGS. 1 and 2, relating to a method for executing and managing tasks using a generation AI and a predetermined task management system.

[0044] (A method for executing and managing tasks using generative AI and a predefined task management system) Figure 1(a) shows the generation cycle of the task management and execution system. Symbol a indicates the input action of the user speaking to the chatbot application (hereafter abbreviated as chatbot) to say hello or request task execution, while symbol b indicates the action of the chatbot issuing necessary instructions such as task execution to the generation AI. Here, the generation AI has already learned new knowledge (additional knowledge α, β, γ) other than the standard knowledge, which will be described later.

[0045] Symbol c indicates the output action from the generation AI, and symbol d indicates the response action to the user.

[0046] Furthermore, the symbol e indicates the operation in the backyard where the output data from the generation AI is stored (saved) in a predetermined task management system, which can be open source software such as REDMINE.

[0047] The symbol f indicates the operation of the generation AI obtaining input data from a specified task management system (note that the timing at which the generation AI obtains the input data naturally precedes the timing at which the generation AI stores the output data based on the input data; this will be discussed later in relation to Figure 2).

[0048] In Figure 1(a), if a conversation continues for a certain period between a user who wants to use the generation AI and the generation AI, the operations indicated by symbols a to d (abcd) are repeated. Furthermore, if input and output between the generation AI and a specific task management system continues for a certain period, the operations indicated by symbol bcef are repeated.

[0049] Figure 1(b) is a flowchart illustrating the conversation between the user and the generated AI and the activities in the backstage area.

[0050] Figure 2 shows the relationship between the input to the generation AI in the task management and execution system and the data storage in a specified task management system regarding the output from the generation AI. Note that the symbols a to f are the same as the symbols a to f shown in Figure 1.

[0051] In Figure 2, the tasks are assumed to be as follows: Task A is the preparation of a Japanese writing file, Task B is the English translation using generative AI, and Task C is the English translation being checked by a native speaker.

[0052] In code a, when a user (a chatbot application user in charge of a specific task) makes a request to the chatbot to "perform task B," code b gives an input instruction to the generation AI, and the input data (Japanese text) stored in the inferred input location of the specified task management system is output to the chatbot via codes b, c, e, and f, and then input from the chatbot to the generation AI.

[0053] Then, when there is output from the generation AI to the chatbot at code c, the chatbot asks at code d, "It's done. Do you want to save it?" If the user answers "OK" at code a, input is made from the chatbot to a specified task management system at code e, and the output data (English text) from the generation AI is stored in the inferred output location of the specified task management system.

[0054] Here, the input location (input location) where the above-mentioned input data is stored in a specified task management system and the output location (output location) where the above-mentioned output data is stored can be inferred by the generation AI, eliminating the conventional situation of manually preparing input and saving output as shown in Figure 12, thereby improving work efficiency.

[0055] Such a generating AI can infer the input location where the generating AI's input data is entered and the output location where the generating AI's output data is stored in a specified task management system because the generating AI has previously learned the specified knowledge necessary for executing and managing tasks as new knowledge such as the following.

[0056] That is, As a task execution With regard to the first knowledge, which is task-specific (task-specific) knowledge required for generation of each task, knowledge α (Figure 1(a)) that is lacking in the standard knowledge (standard knowledge) of the generation AI is additionally learned by the generation AI; with regard to the second knowledge, which is knowledge for inferring the logical storage destination of the above-mentioned input data and output data, knowledge β (Figure 1(a)) that is lacking in the standard knowledge (standard knowledge) of the generation AI is additionally learned by the generation AI; and with regard to the third knowledge, which is knowledge related to the execution for physically acquiring input data and physically storing output data, knowledge γ (Figure 1(a)) that is lacking in the standard knowledge (standard knowledge) of the generation AI is additionally learned by the generation AI through the specifications of the adapter function of the adapter function unit.

[0057] Here, the adapter function unit enables the specified task management system to be operated from the chatbot (at least input and output can be performed via the network), and in Figure 2 it is located outside both the chatbot and the specified task management system, but it may also be located inside either the chatbot or the specified task management system, or inside both the chatbot and the specified task management system.

[0058] Here, the generation AI learns knowledge α such as "Meeting minutes should be a list of around 10 items (regardless of where they are stored)," knowledge β such as "Output should be in the 'Minutes' field," and knowledge γ such as update_task(task_id, output) as a method for updating a task.

[0059] As an example of a specific method for a generative AI to learn predetermined knowledge (α, β, γ), it can use Open AI's Assistants API and have it learn using the tools provided by the API.

[0060] As shown in Figure 2, the user ID of the specified task management system is notified to the generating AI at the beginning of the session. Also, in each of tasks A to C, the "state" indicates the status shown in Figure 5, which will be described later.

[0061] Also, in Figure 2, a task management system user is a user who uses only a specified task management system without using the generation AI.

[0062] (Input and output positions in a given task management system inferred by the generative AI) Figure 6 shows the input and output positions inferred by a generation AI in a given task management system. For the parent task in the upper left of Figure 6, the generation AI infers input position a and output position b in that parent task. Then, for example, a or b in the parent task in the upper left of Figure 6 becomes the input position in another subsequent parent task (the parent task on the right side of Figure 6, which is in a chronological relationship), and its output position c is inferred. This means that the generation AI can infer input and output positions in a given task management system not only in chronological relationships but also in hierarchical relationships.

[0063] As can be seen from this example, the inputs and outputs of a particular generative AI-based task are not necessarily contained within the same task.

[0064] Figure 7 shows an example in which a generation AI generates multiple tasks that branch off from a certain task, and also generates the context between the task before the branch and these multiple tasks, while at the same time setting the attributes of the branched tasks as the output location.

[0065] Figure 8 shows an example in which a generation AI generates a single task that merges multiple tasks, and also generates the context between the tasks before and after the merge, while at the same time using the attributes of the merged task as the output location.

[0066] In this way, in this embodiment, a unified tool comprising a generation AI that has learned predetermined knowledge, a predetermined task management system, and a chatbot application provided between the generation AI that has learned the predetermined knowledge and the predetermined task management system can be used to expand not only to process 1 but also to process 2 and process 3 shown in Fig. 9. Moreover, by using an open-source predetermined task management system and using one type of generation AI, it is possible to expand not only to process 1 but also to process 2 and process 3 as a low-cost tool (see Fig. 9).

[0067] Furthermore, as the vertical axis, the AI ​​axis (automation axis), evolves through upgrades to AI (including improvements in usability), the upgraded AI can be expanded to not only process 1, but also process 2 and 3, so there is no need to aim for massive automation all at once.

[0068] (Devices used directly in carrying out the methods for performing and managing the above tasks) The devices directly used to implement the methods for performing and managing the above tasks can be any of the devices described below.

[0069] (1) A device having a chatbot application provided between a generation AI and a predetermined task management system (2) A device having a predetermined task management system and a chatbot application. (3) A generation AI that has learned the above-mentioned predetermined knowledge and a chatbot application; A device having (4) A device having a generation AI that has learned the above-mentioned predetermined knowledge, a predetermined task management system, and a chatbot application. (program) The program according to this embodiment causes a computer to execute the first and second steps described above. That is, as shown in Fig. 10, an information terminal such as a personal computer (PC) through which a user inputs, a server, a generation AI, and a predetermined task management system are connected via a network such as the Internet, enabling the execution of the program.

[0070] (Integrated task management and execution system using generative AI) The integrated task management and execution system using generative AI in this embodiment comprises a specified task management system, a chatbot that uses generative AI to interact with users and generate tasks for each task, and an adapter function unit that enables the specified task management system to be operated from the chatbot, and uses the specified task management system as a storage destination for input and output data required for generating tasks via the adapter function unit.

[0071] Then, for the first knowledge, which is task-specific knowledge required for generation of each task, the missing part α (Figure 1(a)) of the standard knowledge of the generation AI is additionally learned by the generation AI; for the second knowledge, which is knowledge for inferring the logical storage destination of the input data and the output data in the specified task management system, the missing part β (Figure 1(a)) of the standard knowledge of the generation AI is additionally learned by the generation AI; and for the third knowledge, which is knowledge regarding the execution for physically acquiring the input data and physically storing the output data in the specified task management system, the missing part γ (Figure 1(a)) of the standard knowledge of the generation AI is additionally learned by the generation AI according to the specifications of the adapter function unit.

[0072] Furthermore, by using a designated task management system, program development can be kept to a minimum, making it possible to efficiently execute and manage tasks using generative AI at low cost, and by managing the generative AI's knowledge, tasks can be flexibly and gradually separated, connected, and integrated.

[0073] As already mentioned, the adapter function unit enables a predetermined task management system to be operated from a chatbot.

[0074] According to the present embodiment described above, it is possible to solve the problems that have arisen in the past, such as the new effort of verifying the output of the generative AI and the effort of linking the data to other departments and processes, which in most cases has remained unchanged since the birth of the generative AI service.

[0075] (Variation) Although the preferred embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and changes are possible within the scope of the gist of the present invention.

[0076] For example, in addition to the above, the chatbot application may also have an efficiency improvement function that can supplement natural language conversations with operations such as pressing a button.

[0077] Furthermore, in the above-described embodiment, the generation AI and the specified task management system are described as existing separately from the PC (Figure 10), but at least one of the generation AI and the specified task management system may exist on the PC.

[0078] Furthermore, several devices have been shown as devices that can be used directly to implement the method having the first and second steps described above, but other devices can also be configured to be used directly to implement the above method using only a generation AI that has learned predetermined knowledge, or only a predetermined task management system.

[0079] Furthermore, in the above-described embodiment, the execution of a task by the generation AI was described as inputting and outputting to an existing task, but the execution of a task also includes the generation and deletion of the task itself, as well as operations that are prerequisites for task execution, such as searching.

[0080] Furthermore, the input and output method for chatbots is not limited to text, but may also be voice or other means.

[0081] Furthermore, in the above-described embodiment, a method of using tools from Open AI's Assistants API was described as a method of learning predetermined knowledge, but other methods are possible, such as developing a large-scale language model (LLM) and fine-tuning it, or simply learning using prompts, and further simplifying the prompt method using a chatbot application. [Explanation of symbols]

[0082] α: The portion of the first knowledge, which is the task-specific knowledge required for generation of each task, that is lacking in the standard knowledge (standard knowledge) of the generation AI; β: The portion of the second knowledge, which is the knowledge to infer the logical storage destination of the input data and output data of the generation AI, that is lacking in the standard knowledge (standard knowledge) of the generation AI; γ: The portion of the third knowledge, which is the knowledge related to the execution to physically obtain the input data and physically store the output data, that is lacking in the standard knowledge (standard knowledge) of the generation AI.

Claims

1. A method for executing and managing tasks using a generation AI, a predetermined task management system, a chatbot application provided between the generation AI and the predetermined task management system for dialogue with a user and for execution of tasks by the generation AI, and an adapter function unit that enables the predetermined task management system to be operated from the chatbot application, A first step of learning predetermined knowledge necessary for executing and managing the task so that the generating AI can infer, in the predetermined task management system, an input location where the input data of the generating AI is input and an output location where the output data of the generating AI is stored as logical storage locations; A second step in which, when executing the task via the chatbot application and the adapter function unit, the generation AI inputs input data obtained from an input location in the specified task management system inferred by the generation AI into the generation AI, and stores output data from the generation AI in an output location in the specified task management system inferred by the generation AI; and The predetermined knowledge is a first knowledge that is task-specific knowledge; second knowledge for inferring the logical storage locations of the input data and the output data in the predetermined task management system; and third knowledge, which is knowledge regarding execution for physically obtaining said input data and physically storing said output data in said predetermined task management system.

2. The predetermined knowledge is The method according to claim 1, wherein missing parts of the standard knowledge of the generated AI are additionally learned by the generated AI.

3. An apparatus for direct use in carrying out the method according to claim 1 or 2, comprising: The generated AI that has learned the predetermined knowledge; The chatbot application is provided between the generated AI that has learned the predetermined knowledge and the predetermined task management system; An apparatus comprising:

4. An apparatus for direct use in carrying out the method according to claim 1 or 2, comprising: The generated AI that has learned the predetermined knowledge; the predetermined task management system; The chatbot application is provided between the generated AI that has learned the predetermined knowledge and the predetermined task management system; An apparatus comprising:

5. A program directly used to implement the method according to claim 1 or 2, A program that causes a computer to execute the first and second steps.

Images