Information processing device and program

Abstract

This technology provides solutions to clarify the challenges involved in resolving operational problems faced by employees within an organization. [Solution] The customer service AI agent 4 presents the employee with one or more questions to estimate the outline of the problem in order to solve the business problem, and estimates the outline from the answers. The estimated outline is passed to a designated specialist AI agent 5A from the specialist AI agent group. Specialist AI agent 5A can present the employee with questions to estimate the problem, and the estimated problem is passed to one AI agent. That AI agent determines the tasks necessary to create a solution to the problem and the order in which they will be executed, and selects an AI agent to which to assign the determined tasks. The solution is then created by the solution creation AI agent group 3, which includes the AI ​​agent to which the problem was passed.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus and a program.

Background Art

[0002] Employees belonging to an organization such as a company may have questions or problems in performing their work. For this reason, there is a system in which employees input business questions or problems and appropriate answers are presented (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Unlike questions, business problems usually need to be solved or some action needs to be taken. Therefore, the importance is higher. The problem here can be rephrased as the gap (deviation) between the state that must be reached and the current state. Thus, solving the problem can be said to eliminate or reduce the gap.

[0005] There is one or more elements that cause this gap. There is a gap between the state that should be in the matter of interest and the current state in that element. Therefore, solving the problem becomes possible by eliminating or reducing the gap in each element. Thus, eliminating or reducing the gap in the element becomes the problem to be solved.

[0006] For example, poor product sales can sometimes arise because the product's functionality is inferior to that of competitors' products. In such cases, the challenge lies in improving the functionality, that is, at least reducing the gap. Solving that challenge will solve the problem. Therefore, in order to solve a problem, it is important to clearly define the challenge required to solve that problem.

[0007] Employees attempting to solve work-related problems may not always recognize the challenges necessary to resolve those problems. Recognizing these challenges requires a certain level of critical thinking and knowledge. Therefore, clearly defining the challenges necessary to solve a problem is considered important.

[0008] The present invention aims to provide a technology for clarifying the challenges in resolving work-related problems faced by employees belonging to an organization. [Means for solving the problem]

[0009] An information processing apparatus according to one aspect of the present invention comprises: a transmission / reception processing unit that performs processing for sending and receiving data with an employee terminal used by an employee belonging to an organization; a question presentation unit that causes the employee to send question data to be presented to the employee via the transmission / reception processing unit; a problem estimation unit that, when the employee's answer data to the question data transmitted by the question presentation unit is received by the transmission / reception processing unit, estimates a problem for solving a business problem in the organization to which the employee belongs from the answer data; and a question transmission control unit that, when the problem estimation unit does not estimate the problem, controls the question presentation unit to send the next question data. [Effects of the Invention]

[0010] This invention makes it possible to clarify the challenges for resolving work-related problems faced by employees belonging to an organization. [Brief explanation of the drawing]

[0011] [Figure 1] This figure illustrates an overview of an example of a service provided by an information processing device according to one embodiment of the present invention. [Figure 2] This diagram illustrates an example of how to build a group of AI agents that generate solutions. [Figure 3] This figure shows an example of an information processing system and its connection environment constructed by a service provider using an information processing device according to one embodiment of the present invention. [Figure 4] This figure shows an example of the hardware configuration of an AP server, which is an information processing device according to one embodiment of the present invention. [Figure 5] This figure shows an example of a functional configuration implemented on an AP server, which is an information processing device according to one embodiment of the present invention. [Modes for carrying out the invention]

[0012] Embodiments of the present invention will be described below with reference to the drawings. Figure 1 is a diagram illustrating an overview of an example of a service provided by an information processing device according to one embodiment of the present invention.

[0013] Service Provider 1 is a company that provides consulting or advisory services primarily to organizations such as companies (corporations). It provides SaaS (Software as a Service) as a service via a network. Figure 1 shows an example of a service provided via SaaS. This service is primarily intended for organizations. Hereafter, the organization will be assumed to be a company.

[0014] In this embodiment, an execution environment for artificial intelligence (AI) agents is constructed, and multiple types of AI agents are used to provide services. The information processing device according to this embodiment is the one on which this execution environment has been constructed. An AI agent is a system or program that autonomously determines and executes tasks to achieve a specific objective. Therefore, in addition to autonomy, an AI agent possesses the characteristics of acquiring information from its environment (perception), making decisions based on the acquired information (reasoning and planning), and taking appropriate actions (execution). Here, we assume the AI ​​agent is a program.

[0015] The services provided by service provider 1 are primarily targeted at companies. Therefore, the main users of the service are employees belonging to the company. The employee terminal 9 shown in Figure 1 is a terminal used by employees to access the service. This terminal 9 is, for example, an information processing device equipped with communication functions. PCs (Personal Computers), tablet devices, and smartphones can be used as employee terminals 9. Here, "employee" is a general term for anyone involved in the company's business. Therefore, even officers under the Companies Act are considered employees if they are involved in the business.

[0016] Primarily aimed at companies, this service involves clarifying the issues that need to be resolved in the company's operations and then proposing solutions to those issues. Both clarifying the issues and proposing solutions help companies to perform their operations more effectively. Services that enable this will be referred to as "consulting services" from now on.

[0017] Recognition of business problems is often easy for employees. However, employees do not always properly recognize the causes or factors that give rise to the problems. The text presenting the problem (hereinafter referred to as the "problem statement text") often cannot infer (identify) the causes or factors. For example, the problem statement text "The sales of □□ products are sluggish" can be said to appropriately represent the problem. However, even if it is possible to speculate on the candidates for the cause of the sluggish sales, it is impossible to estimate the probable cause. Solving a problem becomes possible by identifying the cause that gives rise to the problem and eliminating the cause. The task is to eliminate the cause. For this reason, in the present embodiment, starting from the problem presented by the employee, the task for solving the problem is clarified. Furthermore, a solution to the clarified task is presented (proposed).

[0018] For employees with high skills, that is, high thinking ability, deep knowledge, etc., even for problems where it is difficult to identify the task, it is possible to appropriately identify the task. However, for that, it is necessary to appropriately recognize and organize the situation regarding the problem, its background, etc., collect the necessary information, and perform the cumbersome work of summarizing them. For this reason, even employees with high skills do not necessarily find it easy to identify the task. There is a high possibility that it will take a long time.

[0019] For these reasons, in the present embodiment, the necessary information is sequentially obtained from the employee in an interactive format, and the task is clarified from the obtained information. The service provider company 1 takes the lead in the interactive format. Therefore, the employee does not have to select the necessary information, organize the selected information, etc. For this reason, the burden on the employee side is reduced, and the time resources can be used more effectively.

[0020] By clarifying the issues, it becomes possible to more appropriately and reliably create (formulate) solutions to the problems, that is, countermeasures against the causes that give rise to the problems. Therefore, in this embodiment, by clarifying the issues and creating solutions in conjunction with each other, a more convenient service is provided. Since this service does not require the intervention of experts or the like, it can be a service with reduced costs compared to the case of requesting experts or the like.

[0021] In this embodiment, as described above, a plurality of types of AI agents are used to provide a consulting service. As shown in FIG. 1, the uses of the AI agents are broadly classified into those for the dialogue AI agent 2 and the others.

[0022] The dialogue AI agent 2 is an AI agent that functions as a front end. Thereby, the input and output of information in the form of a dialogue with an employee are realized by the dialogue AI agent 2. In this embodiment, the dialogue AI agent 2 is used to clarify the issues. The clarification of the issues corresponds to the generation of a sentence that appropriately expresses the issues. That sentence will hereinafter be referred to as the "issue expression sentence".

[0023] The input of information by the employee and the output of information to the employee are performed using the employee terminal 9. The types of information to be input and output are not particularly limited. The information may be, for example, either voice information or text information. Here, in order to avoid confusion, the information to be input and output is hereinafter assumed to be text information (character information). Also, when it is not considered particularly necessary, the existence of the employee terminal 9 is ignored.

[0024] As the dialogue AI agent 2, as shown in FIG. 1, a window AI agent 4 and a group of specialized AI agents 5, which are a plurality of specialized AI agents, are assigned. AI Agent 4 at the customer service desk is the AI ​​agent that responds to the initial inquiry from an employee. For this reason, AI Agent 4 at the customer service desk is assigned a highly versatile system that can handle a variety of inquiries. "Management consultation" as indicated in Figure 1 shows an example of the content of such an inquiry. Problems classified as "management consultation" are often problems where it is not only difficult to estimate the issue, but also that a lot of information is required to estimate the issue.

[0025] AI agents tend to suffer from decreased accuracy in individual tasks as their versatility increases. Employees face a variety of business problems, each encompassing diverse domains. Therefore, it is undesirable to rely solely on the highly versatile customer service AI agent 4 to handle all these problems. For this reason, in this embodiment, a group of specialized AI agents 5 is prepared, each specializing in a different field, such as industry, and one of these specialized AI agents, 5A, is assigned to handle each task.

[0026] For example, electrical products can be classified by their use, such as household, commercial, medical, and information / communication. In terms of structure and power utilization, they can be classified into white goods, consumer electronics, and small appliances. Other classifications include power source type, legal and environmental standards, and international standards. Furthermore, areas of focus include constituent components, installed functions, power saving, and cost reduction. For these reasons, it is desirable to be able to address various issues even within the same electrical product. In this embodiment, this is achieved by providing a group of 5 specialized AI agents.

[0027] The presence of the specialized AI agent group 5 allows the customer service AI agent 4 to estimate (identify) the outline of the problem to be solved by the employee and select the appropriate specialized AI agent 5A to handle it. Therefore, after receiving an inquiry from an employee, the customer service AI agent 4 repeatedly asks questions as needed to obtain the information necessary to identify the outline of the problem. In Figure 1, these inquiries are referred to as "questions," and the employee's response to those questions is referred to as "answers." This notation will be used hereafter. Note that inquiries, questions, and answers are all data. In other words, they are inquiry data, question data, and answer data.

[0028] The specialist AI agent 5A, selected by the front desk AI agent 4, is provided with a summary (text information representing the summary) of the issue identified by the front desk AI agent 4. This summary will hereafter be referred to as the "summary statement." The expert AI agent 5A attempts to obtain the information necessary to clarify the problem from the answers by determining and presenting further questions to the employee based on the given problem summary (summary statement). These questions are presented again as needed. In this way, the expert AI agent 5A repeatedly presents questions as needed, obtains the information necessary to clarify the problem from the employee, and estimates (identifies) the problem. Problem estimation corresponds to the generation of a problem statement.

[0029] Figure 1 shows that both the customer service AI agent 4 and the specialist AI agent 5A present multiple questions and obtain answers from employees for each question. This is because the problem is assumed to require a lot of information to estimate the issue. In problems where it is not necessary to obtain further information, the customer service AI agent 4 will present only one question, and the specialist AI agent 5A will present none in most cases. Furthermore, the thicker parts of the lines extending downward from the counter AI agent 4 and the specialist AI agent 5A represent the period during which they are functioning. The same applies to the solution creation AI agent group 3.

[0030] Each of the specialized AI agents comprising the main AI agent 4 and the specialized AI agent group 5 has been trained, for example, through few-shot learning, on questioning techniques and problem structuring techniques (such as MECE (Mutually Exclusive and Collectively Exhaustive) analysis, logic trees, and issue trees) actually used by top-tier consultants. Through such training, they autonomously generate and present questions to efficiently collect necessary information and structurally organize the collected information. As a result, the main AI agent 4 can accurately estimate (identify) the overview of a problem, and each specialized AI agent can accurately estimate (identify) the problem itself.

[0031] Expert AI agent 5A selects an AI agent to whom the problem estimation results should be passed. As a result, the problem estimation results, i.e., the problem description, are passed to the AI ​​agent selected by expert AI agent 5A. The AI ​​agent to whom the problem description is passed is one of the solution-generating AI agent group 3.

[0032] Figure 2 illustrates an example of how to build a group of AI agents that generate solutions. Referring to Figure 2, we will now specifically explain an example of how to build them, along with other AI agents that exist besides the one used for conversational AI agent 2. The solution-generating AI agent group 3 is a combination of one or more AI agents other than those used for conversational AI agent 2. One or more of these AI agents are provided for each industry. Typically, as shown in Figure 2, there are multiple agents for each specialty. The task decomposition AI agent group 21, information gathering AI agent group 22, information organization AI agent group 23, document creation AI agent group 24, review AI agent group 25, and language conversion AI agent group 26 shown in Figure 2 are all AI agent groups that include one or more corresponding specialized AI agents. This is just an example, and the types, number, and combinations of AI agent groups prepared for each industry are not particularly limited. To avoid confusion, we will only consider AI agent groups 21-26 shown in Figure 2 as examples of AI agent groups.

[0033] Each task-decomposition AI agent constituting the task-decomposition AI agent group 21 autonomously determines the tasks to be executed and their execution order in order to create a solution. It also selects (determines) zero or more AI agents from the other AI agent groups 22-26 to which the determined tasks should be assigned.

[0034] The specialized AI agent 5A selects a task decomposition AI agent from a group of 21 task decomposition AI agents in the industry to which the company experiencing the problem belongs, and assigns the estimated task to that AI agent. Through this selection, it becomes possible to more appropriately determine (decompose) tasks and assign them to AI agents.

[0035] The other AI agent groups 22-26 consist of AI agents with the following functions (specializations), for example: Each AI agent constituting the information gathering AI agent group 22 is responsible for collecting necessary information from external databases (DBs) and / or internal databases. External databases include, for example, those managed by provider 6, which has a search engine AI, as shown in Figure 1. The search engine AI collects information published by each website 7 through crawling and stores it in the database. Subtasks are requested from such search engine AIs to obtain the necessary information. If the contract allows access to the company's own system 8, it may, if necessary, collect information from the database managed by the company's own system 8.

[0036] The information to be collected is determined according to the assigned task and is not particularly limited. The collected information may be used to implement RAG (Retrieval Augmented Generation). Each AI agent constituting the information organization AI agent group 23 determines the logical structure to be used for organizing the collected information, and then organizes the information using the determined logical structure. Each AI agent in the text-generating AI agent group 24 creates text that represents a solution, reflecting the results of organizing the collected information. Each AI agent in the 25-group review AI agent suite reviews the text of the proposed solutions and refines them to make them more appropriate. They may also be able to add (create) diagrams to represent the content of the text, convert it to a format desired by the employee, and perform other related tasks. Each AI agent that makes up the language conversion AI agent group 26 converts text into a language different from the original language.

[0037] Figure 2 shows that the first task decomposition AI agent 3A from the task decomposition AI agent group 21 was selected by the specialist AI agent 5A. The tasks determined by this first task decomposition AI agent 3A for creating a solution are then assigned to the second information gathering AI agent 3B from the information gathering AI agent group 22, the third information organizing AI agent 3C from the information organizing AI agent group 23, the fourth document creation AI agent 3D from the document creation AI agent group 24, and the fifth review AI agent 3E from the review AI agent group 25, respectively. The vertical arrangement of each AI agent 3A to 3E represents the execution order.

[0038] The solution-creation AI agent group 3 is constructed through selection by the specialized AI agent 5A, determination of tasks and their execution order by the first task decomposition AI agent 3A, and assignment of each task. As a result, the solution-creation AI agent group 3 consists of the first task decomposition AI agent 3A, the second information gathering AI agent 3B, the third information organization AI agent 3C, the fourth document creation AI agent 3D, and the fifth review AI agent 3E. Each AI agent 3A to 3E is executed in that order. Thus, the "information definition" shown in Figure 1 is performed by the first task decomposition AI agent 3A. Similarly, "information retrieval" is performed by the second information gathering AI agent 3B, "information organization" by the third information organization AI agent 3C, "solution creation" by the fourth document creation AI agent 3D, and "review" by the fifth review AI agent 3E.

[0039] In Figure 2, one AI agent 3B to 3E is selected from each of the AI ​​agent groups 22 to 25. However, it is possible to select multiple AI agents from one or more AI agent groups. When a problem exists in multiple domains, for example, when problems exist in finance, marketing, and competitors, it is highly likely that multiple AI agents will be selected from one or more AI agent groups. By selecting multiple AI agents from the same AI agent group, the necessary tasks can be performed for each domain. Therefore, even when there is one or more problems in multiple domains, it is possible to more reliably create appropriate solutions.

[0040] In this embodiment, multiple AI agent groups are prepared, categorized by industry and specialization. This is because, even within the same industry, problems can arise in various areas, and the response to problems in each area may differ depending on the size of the company, etc. By preparing such AI agent groups for each industry, it becomes possible to respond more appropriately to problems that occur in various areas across various industries. It also becomes possible to respond appropriately even when there is one or more issues in multiple areas. This means that solutions to problems can be created more appropriately and reliably.

[0041] As shown in Figure 1, the solutions after the review are passed to the expert AI agent 5A and presented to the employee via the expert AI agent 5A. Each solution presents a proposal for resolving a clarified issue. This allows the employee to see the problem and its solution method from the presented solutions.

[0042] In this embodiment, as described above, a solution is created and presented to the employee, starting with an inquiry made by the employee. Thus, as shown in Figure 1, the situation leading up to the creation of a solution can be divided into the following stages: F1, where the counter AI agent 4 is in charge; F2, where the specialist AI agent 5A selected by the counter AI agent 4 is in charge; and F3, where the solution creation AI agent group 3 is constructed by the AI ​​agent (task decomposition AI agent) selected by the specialist AI agent 5A, and the solution is created by the constructed solution creation AI agent group 3.

[0043] Assuming that each AI agent shown in Figure 1 or Figure 2 constitutes a system, the customer service AI agent 4 and the group of specialized AI agents 5 correspond to the question presentation unit in this embodiment. The customer service AI agent 4 corresponds to the first question presentation unit and the overview estimation unit in this embodiment, and the group of specialized AI agents 5 corresponds to the second question presentation unit and the problem estimation unit in this embodiment.

[0044] Furthermore, the solution creation AI agent group 3 corresponds to the solution creation unit in the narrow sense in this embodiment. The first task decomposition AI agent 3A corresponds to the first artificial intelligence agent in this embodiment. The other AI agents 3B to 3E correspond to the second artificial intelligence agent in this embodiment.

[0045] Further details will be explained by referring to the drawings. Figure 3 shows an example of an information processing system and its connection environment constructed by a service provider using an information processing device according to one embodiment of the present invention. Note that the configuration of the information processing system and the connection environment shown in Figure 3 are examples only and are not particularly limited. For example, the information processing system may have multiple AP servers 32. Alternatively, a load balancer may be provided, and multiple combinations of Web servers 31, AP servers 32, and DB servers 33 may be arranged. The information processing system may also be constructed using cloud services. Therefore, the location of the information processing system is not particularly limited.

[0046] The information processing system is a computer system built by service provider 1 for the purpose of providing consulting services. The above-mentioned Web server 31, AP server 32, and DB server 33 are connected to network 34. Network 34 is, for example, a LAN (Local Area Network).

[0047] The AP server 32 is an information processing device on which the execution environment for the AI ​​agent is built, and corresponds to the information processing device in this embodiment. The Web server 31 is connected to an external network 35 and enables the use of consulting services using employee terminals 9. If consulting services are provided to registered companies, the Web server 31 supports logins of employees belonging to registered companies. The DB server 33 is used for storing various data, etc.

[0048] The external network 35 is a complex network, for example, including the internet. Multiple providers 6 and multiple websites 7 are connected to this network 35. The search engine AI of provider 6 performs crawling via network 35 and collects information from each website 7.

[0049] Figure 3 shows a manufacturing company 36 as an example of a company to which the employees belong. In this manufacturing company 36, the company's own system 8 and router 37 are connected to network 38. Multiple employee terminals 9 are connected to or can be connected to network 38. Employee terminals 9 can access the web server 31 via network 38, router 37, and network 35. Network 38 is, for example, a LAN.

[0050] Figure 4 shows an example of the hardware configuration of an AP server, which is an information processing device according to one embodiment of the present invention. This hardware configuration example is just one example and is not particularly limited. For example, only one CPU (Central Processing Unit) 321 and one GPU (Graphics Processing Unit) 324 are shown, but multiple units of each may be installed. An NPU (Neural Processing Unit) may be installed instead of the GPU 324. Alternatively, neither the GPU 324 nor the NPU may be installed, and a CPU 321 equipped with an NPU core may be used.

[0051] As shown in Figure 4, the AP server 32 has a configuration in which the CPU 321, ROM (Read Only Memory) 322, RAM (Random Access Memory) 323, GPU 324, NIC (Network Interface Card) 325, auxiliary storage device 326, media drive 327, and I / FC (Interface Controller) group 328 are connected to the bus 329. The GPU 324 is connected to VRAM (Video RAM) 324A.

[0052] The NIC325 enables communication via network 34. This communication via network 34 may be either wireless or wired. The NIC325 may support multiple communication standards. Although Figure 4 shows only one NIC325, multiple NIC325s supporting different communication standards may be installed.

[0053] The auxiliary storage device 326 is a device capable of permanently storing data, such as a hard disk drive or an SSD (Solid State Drive). The media drive 327 is a device on which the recording medium 327A can be attached and detached. The media 327A is such as a CD (Compact Disc)-ROM, DVD-ROM, or DVD-RAM.

[0054] The I / FC group 328 includes various I / FCs that enable communication with various peripheral devices, including the input device 328A and the display device 328B, or with external devices. The input device 328A and the display device 328B are temporarily connected to the I / FC group 328 as needed.

[0055] The auxiliary storage device 326 stores the OS (Operating System) and various application programs that run on the OS as programs. These application programs include those developed for providing consulting services. These application programs correspond to the programs in this embodiment. Hereafter, these programs will be referred to as "consulting service applications."

[0056] This consulting service application includes parts executed by CPU321 and parts executed by GPU324. The part executed by CPU321 is, for example, the main program, while the parts executed by GPU324 are multiple subprograms. If we consider each of the AI ​​agents that make up the customer service AI agent 4, the specialized AI agent group 5, and each of the AI ​​agents that make up the industry AI agent group 20 as programs, then one subprogram corresponds to one AI agent. Therefore, the execution environment for the AI ​​agents is provided by GPU324.

[0057] ROM322 is also a device capable of permanently storing data, such as firmware and various other data. At startup, the CPU321 reads the firmware stored in ROM322 into RAM323 and executes it. Subsequently, the firmware reads the OS stored in auxiliary storage device 326 into RAM323 and executes it. Various application programs, including consulting service applications, are read into RAM323 by the OS and executed. The GPU324 can execute various subprograms stored in auxiliary storage device 326 and read into VRAM324A.

[0058] The consulting service application may be stored on media 327A and distributed. If network 35 is a complex network including the internet, it may also be distributed via network 35. When distributed via network 35, the consulting service application should be stored on a recording medium that can be directly or indirectly accessed by the information processing device distributing it. In other words, the storage medium may be directly or indirectly accessible by another information processing device that can communicate with the information processing device distributing it.

[0059] Figure 5 shows an example of a functional configuration implemented on an AP server, which is an information processing device according to one embodiment of the present invention. The functional configuration example on CPU 321 is mainly implemented by the execution of the OS or the main program of the consulting service application on CPU 121. In all of the functional configuration examples on GPU 324, the execution of each subprogram of the consulting service application on GPU 324 is implemented. Note that the functional configuration is not particularly limited. Various modifications are possible. For example, the subprograms that make up the consulting service application may be divided and executed on multiple AP servers.

[0060] As shown in Figure 5, the CPU 321 of the AP server 32 is functionally configured to include a transmission / reception processing unit 3211, a DB access processing unit 3212, a page generation instruction unit 3213, a task execution instruction unit 3214, and an external request processing unit 3215. On the other hand, the GPU324 is functionally configured with a customer service processing unit 3214, a specialized field processing unit group 3242, and multiple industry-specific processing unit groups 3243. In Figure 5, only the first industry-specific processing unit group 3243 is shown as an industry-specific processing unit group 3243.

[0061] While such a functional configuration is realized on the CPU 121 and the GPU 124, the auxiliary storage device 326 is reserved as a storage area for information (data), including a company information storage unit 3261, a common information storage unit 3262, a collected information storage unit 3263, and a solution information storage unit 3264.

[0062] The information stored in each memory unit 3261 to 3264 is, for example, as follows: The company information stored in the company information storage unit 3261 includes, for example, information about a company that has registered as a member, or information about major companies that include that company. In the latter case, the information includes, for example, a member ID (IDentification), a company registration number issued by the Legal Affairs Bureau, industry, number of employees, and address. This company information allows for the identification of not only the company to which an employee belongs, but also the size of the company and its industry.

[0063] The common information stored in the common information storage unit 3262 includes, for example, internal company information specific to the registered company, or information that has not yet been learned. This information is primarily used to implement RAG. This common information makes it possible to handle even difficult cases. Because it can be used regardless of its specifics in applicable cases, it is collectively referred to as common information. The collected information stored in the collected information storage unit 3263 is information collected from external databases or internal databases, etc. This information is collected and stored by the second information collection AI agent 3B, etc., as shown in Figure 2. As a result, it may include information from websites 7 collected via provider 6.

[0064] The solution information stored in the solution information storage unit 3264 is a compilation of the created solution and related information. For example, it includes information about the company that presented the solution (part of the company information) and information collected from employees. It is desirable that the system also include evaluation information of the company (employees) that presented the solution. Based on this evaluation information, the solution information of highly-rated solutions may be used as valuable knowledge.

[0065] The above company information and common information are stored in DB server 33. As a result, the company information storage unit 3261 and the common information storage unit 3262 store the information obtained from DB server 33. Collected information is collected as needed and temporarily stored in collected information storage unit 3263. Solution proposal information is generated on the AP server 32 side and sent to DB server 33 for storage. As a result, the solution proposal information stored in solution proposal information storage unit 3264 may include information obtained from DB server 33. For the sake of explanation, it is assumed here that the most important information that the CPU 321 or GPU 324 references or stores when executing processing is pre-stored on auxiliary storage device 326, and an example of the storage area where that information is stored is shown.

[0066] The information stored in each memory unit 3261-3264 is primarily accessed by the GPU 324. This information is read onto VRAM 324 and accessed by the GPU 324. The CPU 321 communicates with the Web server 31 via NIC 325. Data input and output between the CPU 321 and GPU 324 occurs, for example, via RAM 323. These are conveniently ignored in Figure 5, and this will also be the case in subsequent explanations.

[0067] The customer service processing unit 3241, and each target processing unit constituting each corresponding processing unit group 3242 or 3243, which are implemented on GPU324, are all implemented by GPU324 executing one of the subprograms that is an AI agent. More specifically, the customer service processing unit 3241 is implemented by GPU324 executing the subprogram that is customer service AI agent 4. The specialized field processing unit group 3242 is implemented by GPU324 executing the subprogram group corresponding to the specialized AI agent group 5. One of the multiple industry specialized processing units 3243 is implemented by GPU324 executing the subprogram group corresponding to the AA industry AI agent group 20. Hereafter, for convenience, an AI agent as a function implemented by GPU324 executing a single subprogram will be referred to as an "actual AI agent".

[0068] The components 3211 to 3215 implemented on the CPU 121 have, for example, the following functions. The transmission / reception processing unit 3211 performs processing for sending and receiving various data, including requests, with the Web server 31. Inquiries and answers from employee terminals 9 are received as requests via the Web server 31, and the transmission of various data, including questions and solutions, to employee terminals 9 is done via the Web server 31 through requests to the Web server 31. The Web server 31 generates the web page displaying the questions or solutions. The transmission / reception processing unit 3211 corresponds to the transmission / reception processing unit in this embodiment.

[0069] The DB access processing unit 3212 controls access to the DB server 33. The DB access processing unit 3212 manages the retrieval of various data from the DB server 33 and the storage of various data in the DB server 33. The sending and receiving processing unit 3211 handles the sending and receiving of requests to the DB server 33 (for example, requests generated using SQL (Structured Query Language)) and the receiving of responses from the DB server 33 in response to those requests.

[0070] The web pages sent to employee terminals 9 are generated by the web server 31. The page generation instruction unit 3213 instructs the web server 31 to generate the web page and its destination. The transmission of instructions and other information is performed via the transmission / reception processing unit 3211. The above-mentioned transmission / reception processing unit 3211, DB access processing unit 3212, page generation instruction unit 3213, and the external request processing unit 3215, which will be described later, are mainly realized by the CPU 321 executing the OS.

[0071] The task execution instruction unit 3214 manages the execution of the actual AI agent implemented on the GPU 324 and performs control according to the execution results. For this purpose, when a question or answer is received from an employee, it is passed from the transmission / reception processing unit 3211 to the task execution instruction unit 3214. Furthermore, when any of the actual AI agents generate a question or a solution, the task execution instruction unit 3214 passes that question or solution to the page generation instruction unit 3213 and instructs it to send it as a web page to the employee terminal 9. This task execution instruction unit 3214 is implemented by the CPU 321 executing the main program of the consulting service application.

[0072] When the task execution instruction unit 3214 receives a question from an employee, it passes the question to the customer service processing unit 3241 (prompt input) and instructs it to process the question. When the customer service processing unit 3241 responds with a question, the task execution instruction unit 3214 passes that question to the page generation instruction unit 3213, causing the question to be displayed on the employee terminal 9. The task execution instruction unit 3214 passes the answer to the question to the customer service processing unit 3241 and instructs it to process the question. This process of displaying the generated question on the employee terminal 9 and passing the answer to the customer service processing unit 3241 is repeated until the customer service processing unit 3241 provides a summary statement. This results in the situation F1 shown in Figure 1.

[0073] When a summary statement is provided, the task execution instruction unit 3214 passes the summary statement to a designated specialty area processing unit among the specialty area processing unit group 3242 and instructs it to process it. As a result, if a question is returned from the specialty area processing unit, the question is passed to the page generation instruction unit 3213, causing the question to be displayed on the employee terminal 9. The task execution instruction unit 3214 passes the answer to the question to the specialty area processing unit and instructs it to process it. This process of displaying the generated question on the employee terminal 9 and passing the answer to the specialty area processing unit is repeated until the issue statement is passed from the specialty area processing unit. This realizes the situation F2 shown in Figure 1.

[0074] When a problem statement is provided, the task execution instruction unit 3214 passes the problem statement to a designated AI agent from the specified industry-specific processing group 3243 and instructs it to process it. Based on this instruction, the AI ​​agent determines the tasks to be executed to create a solution and their execution order, and returns the assignment results for each task along with the determination result. The task execution instruction unit 3214 then has the AI ​​agents to which the tasks are assigned execute the determined tasks sequentially according to the determined execution order. As a result, the solution is created by the AI ​​agent that executes the task last. Situation F3 shown in Figure 1 is realized in this way.

[0075] The last AI agent executed returns a solution. The task execution instruction unit 3214 passes the solution to the page generation instruction unit 3213, which then displays the solution on the employee terminal 9. This allows the employee to review the created solution. As described above, the presentation of questions to employees by the designated AI agent within the specialized field processing group 3242 is controlled by the task execution instruction unit 3214. In other words, situation F2 is controlled by the task execution instruction unit 3214. Similarly, situation F3 is also realized through the control of the task execution instruction unit 3214. Thus, the task execution instruction unit 3214 corresponds to the question transmission control unit in this embodiment, as well as a part of the solution creation unit in this embodiment.

[0076] The second information-gathering AI agent 3B shown in Figure 2 may perform web searches or request subtasks from AI agents available on provider 6, such as search engine AIs. The external request processing unit 3215 handles such external requests. These requests are made by the task execution instruction unit 3214. The request for the request is transmitted by the transmission / reception processing unit 3211. The information received in response to the request is passed from the transmission / reception processing unit 3211 to the task execution instruction unit 3214, and then to the actual AI agent that made the request. Various types of information can be collected through external requests. Information that can be provided by credit rating agencies or corporate database companies, such as financial statements, can also be collected.

[0077] In this embodiment, the system clarifies the issue and creates solutions based on questions posed by employees, i.e., problem presentations. However, it is also possible to collect necessary information by accessing the company's system 8, and to estimate (identify) and present potential issues. Furthermore, it is possible to introduce suitable personnel to employees who are particularly keen to resolve the issue. Such personnel can be appropriately selected once the issue has been clarified.

[0078] Furthermore, in this embodiment, information is generally collected as needed, but it is also possible to anticipate potential problems, clarify the issues related to those problems, and collect information that may be necessary for creating solutions to those problems in advance. In such a case, it becomes possible to further reduce the number of questions asked of employees, and also to shorten the time required to create solutions. Including these points, various modifications are possible in this embodiment. [Explanation of Symbols]

[0079] 1. Service provider, 2. Conversational AI agent, 3. Solution creation AI agent group, 4. Customer service AI agent, 5. Specialist AI agent group, 5A Specialist AI agent, 6. Provider, 7. Website, 8. Company system, 9. Employee terminal. 20. AA Industry AI agent group, 21. Task decomposition AI agent group, 22. Information gathering AI agent group, 23. Information organization AI agent group, 24. Document creation AI agent group, 25. Review AI agent group, 26. Language conversion AI agent group.

Claims

1. A transmission / reception processing unit that performs data transmission and reception processing for employee terminals used by employees belonging to the organization, An outline estimation unit including a first conversational artificial intelligence agent capable of transmitting question data via the transmission / reception processing unit to obtain necessary information from the employee, and estimating an outline of the issues to be resolved in order to solve business problems in the organization to which the employee belongs, based on the employee's response data to the question data received by the transmission / reception processing unit, and repeatedly transmitting the question data via the transmission / reception processing unit until the outline can be estimated, A task estimation unit including a second conversational artificial intelligence agent capable of transmitting the question data based on the summary via the transmission / reception processing unit after the summary has been estimated by the summary estimation unit, estimating the task from the answer data to the question data received by the transmission / reception processing unit, and repeatedly transmitting the question data via the transmission / reception processing unit until the task can be estimated, An information processing device equipped with the following features.

2. If there are multiple problem estimation units, the summary estimation unit selects one of the multiple problem estimation units based on the summary it has estimated. After the outline estimation unit has estimated the outline, the task estimation unit selected by the outline estimation unit performs the transmission of the question data and the estimation of the task from the answer data received by the transmission of the question data. The information processing apparatus according to claim 1.

3. The system further comprises a solution creation unit that, when the problem estimation unit estimates the problem, creates a solution to solve the estimated problem. The information processing apparatus according to claim 1 or 2.

4. The solution creation unit determines, based on the problem estimated by the problem estimation unit, the tasks to be performed in order to create the solution, and the order in which to perform the tasks. The information processing apparatus according to claim 3.

5. If there are multiple artificial intelligence agents capable of performing the aforementioned task and each specializing in a different area, The solution creation unit includes a first artificial intelligence agent, which is one of the plurality of artificial intelligence agents, and which determines the task and the execution order of the task based on the problem, and selects one or more other artificial intelligence agents from the plurality of artificial intelligence agents to which the determined task is assigned, and one or more second artificial intelligence agents, which are the other artificial intelligence agents to which the task is assigned by the first artificial intelligence agent. The information processing apparatus according to claim 4.

6. In an information processing device, To perform processing for sending and receiving data between employee terminals used by employees belonging to the organization, The above-mentioned transmission and reception process causes the first conversational artificial intelligence agent to transmit question data to obtain necessary information from the employee, and estimates an outline of the issues to be resolved in order to solve the operational problems in the organization to which the employee belongs, based on the employee's response data to the question data received by the above-mentioned transmission and reception process, and repeatedly causes the first conversational artificial intelligence agent to transmit the question data by the above-mentioned transmission and reception process until the outline can be estimated. After the outline is estimated, the transmission and reception process causes the question data based on the outline to be transmitted, and the transmission and reception process estimates the problem from the answer data to the question data received, and the transmission and reception process causes the second conversational artificial intelligence agent to repeatedly transmit the question data until the problem can be estimated. A program that executes a process.

Images