Business support systems, business support methods, and programs

Abstract

To improve the accuracy of output from AI. [Solution] The search setting storage unit (101) of the business support system (1) stores search settings related to RAG (Retrieval-augmented generation) searches, which are specified by a user using the business support service that supports business operations. The search unit (102) searches for input business data, which is business data to be input to the AI ​​(Artificial Intelligence) that supports business operations, from the business database in which business data related to business operations is stored, based on the search settings. The business support unit (103) supports business operations based on the output from the AI ​​to which the input business data has been input.

Description

Technical Field

[0001] The present disclosure relates to a business support system, a business support method, and a program.

Background Art

[0002] Conventionally, a technique called RAG (Retrieval-Augmented Generation) that combines a search function and AI (Artificial Intelligence) has been known. For example, Patent Document 1 describes an answer system that executes a search for words included in a request input by a user and a search for information related to the words, inputs the execution results of these searches into a large language model, which is an example of AI, and outputs an answer from the large language model.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the technique of Patent Document 1, since all the data in the database that is the population at the time of search in RAG is subject to search, inappropriate data unrelated to the user's request may be searched and input into the AI. This is not limited to the technique of Patent Document 1 and applies to all conventional techniques related to RAG. Therefore, with the conventional techniques, the accuracy of the output from the AI could not be improved.

[0005] One object of the present disclosure is to improve the accuracy of the output from the AI. Note that the object of the present disclosure is not limited to the content described in the column of the problems to be solved by the invention.

Means for Solving the Problems

[0006] One aspect of this disclosure relates to a business support system which includes: a search setting storage unit that stores search settings for RAG (Retrieval-augmented generation) searches specified by a user of a business support service that supports business operations; a search unit that searches for input business data, which is the business data to be input to an AI (Artificial Intelligence) that supports business operations, from a business database containing business data related to the business operations based on the search settings; and a business support unit that supports business operations based on the output from the AI ​​to which the input business data has been input. [Effects of the Invention]

[0007] According to this disclosure, the accuracy of the output from the AI ​​can be improved. [Brief explanation of the drawing]

[0008] [Figure 1] This figure shows an example of the hardware configuration of a business support system. [Figure 2] This figure shows an example of a screen displayed on a user's terminal. [Figure 3] This figure shows an example of a screen displayed on a user's terminal. [Figure 4] This figure shows an example of a screen displayed on a user's terminal. [Figure 5] This figure shows an example of a screen displayed on a user's terminal. [Figure 6] This figure shows an example of the functions implemented in the business support system. [Figure 7] This figure shows an example of a search settings database. [Figure 8] This figure shows an example of AI input and output. [Figure 9] This is a diagram showing an example of a business database. [Figure 10] This figure shows an example of a process performed by a business support system. [Figure 11]This figure shows an example of the functions implemented in the modified operation support system. [Figure 12] This figure shows an example of a search settings database for variation 1. [Figure 13] This figure shows an example of the screen displayed on the user terminal in Modification Example 4. [Figure 14] This figure shows an example of the hardware configuration of modified example 4. [Figure 15] This figure shows an example of the processing flow for Modification Example 5. [Figure 16] This figure shows an example of a search settings database for variation 8. [Figure 17] This figure shows an example of AI input and output in modified example 10. [Modes for carrying out the invention]

[0009] [1. Hardware Configuration] This document describes an example of an embodiment of the business support system, business support method, and program related to this disclosure. Figure 1 is a diagram showing an example of the hardware configuration of the business support system. For example, business support system 1 includes a business support server 10, a database server 20, and a user terminal 30. Each of the business support server 10, the database server 20, and the user terminal 30 is connected to a network N such as the Internet or a LAN.

[0010] The business support server 10 is a server computer for business support services. Business support services are services that support business operations. A service called groupware is an example of a business support service. Business support services may also be services that are not specifically called groupware. In this embodiment, the case where the business support service is cloud-based is given as an example, but the business support service may be of any type and is not limited to cloud-based. Business support services may also be on-premise. The business support server 10 may be a server computer for a service other than the business support service.

[0011] For example, the business support server 10 includes a control unit 11, a storage unit 12, and a communication unit 13. The control unit 11 includes at least one processor. The storage unit 12 includes at least one of a volatile memory such as a RAM and a non-volatile memory such as a flash memory. The communication unit 13 includes at least one of a communication interface for wired communication and a communication interface for wireless communication.

[0012] The database server 20 is a server computer for business support services. The database server 20 may be a server computer for other services that cooperate with the business support services. For example, the database server 20 includes a control unit 21, a storage unit 22, and a communication unit 23. The hardware configurations of the control unit 21, the storage unit 22, and the communication unit 23 may be the same as those of the control unit 11, the storage unit 12, and the communication unit 13, respectively.

[0013] The user terminal 30 is a user's computer. The user is a person or an organization that uses the business support services. The user is a different person from the operator who operates the business support services. The operator is a business that provides the business support services to the user. For example, when an organization such as a company to which the user belongs contracts for the business support services (that is, when the organization is a tenant), the user is a person belonging to the organization. The user can also be called a customer by the operator. The user may be an individual.

[0014] For example, the user terminal 30 is a personal computer, a tablet, a smartphone, or a wearable terminal. The user terminal 30 includes a control unit 31, a storage unit 32, a communication unit 33, an operation unit 34, and a display unit 35. The hardware configurations of the control unit 31, the storage unit 32, and the communication unit 33 may be the same as those of the control unit 11, the storage unit 12, and the communication unit 13, respectively. The operation unit 34 is an input device such as a mouse or a touch panel. The display unit 35 is a liquid crystal or organic EL display.

[0015] The programs stored in the memory units 12, 22, and 32 may be supplied via the network N. The hardware configurations of the business support server 10, the database server 20, and the user terminal 30 are not limited to the example in Figure 1. For example, each of the business support server 10, the database server 20, and the user terminal 30 may include at least one of a reading unit (e.g., a memory card slot) for reading computer-readable information storage media and an input / output unit (e.g., a USB terminal) for direct connection to external devices. The programs stored on the information storage media may be supplied to each of the business support server 10, the database server 20, and the user terminal 30 via at least one of the reading unit and the input / output unit.

[0016] Furthermore, the business support system 1 may include at least one computer. The computers included in the business support system 1 are not limited to the example in Figure 1. For example, the business support system 1 may include only the business support server 10 and the database server 20. In this case, the user terminal 30 is located outside the business support system 1. The business support system 1 may include only the business support server 10. In this case, the business support server 10 and the database server 20 are located outside the business support system 1. The business support system 1 may include only the user terminal 30. In this case, the business support server 10 and the database server 20 are located outside the business support system 1.

[0017] In this embodiment, we take the example of a case where the business support system 1 can communicate with the AI ​​(Artificial Intelligence) system 2 via a network N. For example, the AI ​​system 2 includes an AI server 40. The AI ​​server 40 is the server computer for the AI ​​service. The AI ​​service is a service that makes AI available online. Details of the AI ​​will be described later. The AI ​​server 40 includes a control unit 41, a storage unit 42, and a communication unit 43. The hardware configurations of the control unit 41, storage unit 42, and communication unit 43 may be the same as those of the control unit 11, storage unit 12, and communication unit 13, respectively. The hardware configuration of the AI ​​system 2 is not limited to the example in Figure 1, as is the case with the business support system 1.

[0018] [2. Overview of the Business Support System] In this embodiment, we take the example of a case where an organization such as a company contracts for business support services as a tenant. The user uses the business support services as a member of the tenant, which is the organization such as a company to which they belong. For example, the user uses the business support services from a browser installed on the user terminal 30. The user may also use the business support services from a program dedicated to the business support services instead of a browser.

[0019] Figures 2 to 5 show examples of screens displayed on the user terminal 30. For example, when a user logs into the business support service, the user terminal 30 displays the portal screen SC1, which is the entry point to the business support service, on the display unit 35, as shown in the upper part of Figure 2. The portal screen SC1 displays announcements for all users belonging to the tenant, notifications for the user, icons for various settings, and information on business support functions available to the user.

[0020] A business support function is a function that supports business operations. A business support function can also be defined as at least one of a program that represents a series of information processing steps that support business operations, and data referenced by that program. A business support function may be similar to a publicly known business support service. For example, a business support function may be a data management function for managing data, a scheduling function for managing schedules, a communication function for supporting communication between users, an email management function for managing emails, a workflow function for implementing workflows, or other functions.

[0021] In this embodiment, a data management function will be described as an example of a business support function. Furthermore, a type of database called an "app" will be described as an example of a data management function. The term "app" is generally an abbreviation for "application," which is a type of program, but in this embodiment, an app is a type of data management function. An app is a complex business support function that has not only data management functions but also other functions such as communication functions. Various types of business-related data are registered in the app. Users may communicate with other users by registering comments in the app's records.

[0022] For example, when a user selects an app from the portal screen SC1, the user terminal 30 displays the app screen SC2, which shows the app selected by the user, on the display unit 35, as shown in the lower part of Figure 2. In the example at the bottom of Figure 2, the app screen SC2 shows an inquiry handling app that stores information related to customer inquiries. The app screen SC2 includes a list of records L20, which are the units of data registered in the app.

[0023] For example, the first row of list L20 displays the field names. Fields are sometimes called columns. The rows from the second row onward in list L20 display the values ​​of each field in each individual record. When a user selects a record from list L20, the user terminal 30 displays the details of the selected record on the application screen SC2. From the application screen SC2, which displays the record details, the user can edit or delete existing records or create new records.

[0024] In this embodiment, we take the example of a case where a user utilizes an AI assistant function as a business support function. The AI ​​assistant function is a function in which AI assists with business operations. For example, the AI ​​assistant function is realized by business support system 1 cooperating with AI system 2. The AI ​​assistant function is realized by RAG (Retrieval-Augmented Generation). There are various definitions of RAG, but the RAG in this embodiment may be RAG in various known definitions. For example, RAG may be a mechanism that searches for information and generates an answer using the search results.

[0025] In this embodiment, the user specifies search settings for RAG searches before starting to use the AI ​​assistant function. The search settings determine what kind of search to perform in RAG. For example, when the user selects icon I1, the user terminal 30 displays a list screen SC3 on the display unit 35, which shows a list of AI assistant functions for which search settings have been specified, as shown in the upper part of Figure 3. In the example in the upper part of Figure 3, there are no AI assistant functions for which search settings have been specified, so the names of the AI ​​assistant functions are not displayed on the list screen SC3. When search settings are specified for an AI assistant function, the name of that AI assistant function is displayed on the settings screen SC4.

[0026] For example, when the user selects button B30, the user terminal 30 displays a settings screen SC4 on the display unit 35, which accepts the user's specification of search settings, as shown in the lower part of Figure 3. In the example shown in the lower part of Figure 3, the search settings include multiple setting items. In this embodiment, we will mainly describe the setting item for the data source used to generate answers among the multiple setting items. Details of the other setting items will be described in the modified examples below. The search settings may include only one setting item. The number of setting items included in the search settings may be any number.

[0027] A data source is the population of data used in RAG searches. A data source can also be described as the data being searched. For example, a user can specify an app as a data source. A user may specify multiple apps as data sources. A user may specify data sources other than apps. A user may specify threads or chats in the communication function as a data source. A user may specify emails in the email management function as a data source. A user may specify data managed by other business support functions as a data source. A user may specify a combination of data managed by each of multiple business support functions, such as apps and threads, as a data source.

[0028] For example, when a user specifies search settings, the user terminal 30 displays the user's specified search settings on the settings screen SC4, as shown in the upper part of Figure 4. In the upper example of Figure 4, the user has specified the inquiry response application as the data source. The user may also specify the name of the AI ​​assistant function from the settings screen SC4. In the upper example of Figure 4, the name of the AI ​​assistant function is Customer Support Assistant. The user uses this AI assistant function to have the AI ​​assist with customer support tasks.

[0029] For example, once the user has finished specifying the search settings, the user terminal 30 displays the name of the AI ​​assistant function for which the search settings have been specified on the list screen SC3, as shown in the lower part of Figure 4. In the example at the bottom of Figure 4, "Customer Support Assistant" is displayed as the name of the AI ​​assistant function. The user can select an AI assistant function from the list screen SC3 and change the search settings for that AI assistant function. If the user wants to add a new AI assistant function, they select button B30 and specify the search settings for the new AI assistant function.

[0030] For example, a user can invoke an AI assistant function by selecting icon I2. Other users (e.g., colleagues in the same tenant) who are different from the user who specified the search settings for a particular AI assistant function may also be able to invoke that AI assistant function. If there are multiple AI assistant functions, icon I2 may be displayed for each AI assistant function, or a list of AI assistant functions may be displayed after the user selects icon I2, from which the user can select an AI assistant function. The method of invoking an AI assistant function may be by using a mention in a thread post (e.g., a mention such as "@AI"), by calling it according to a predetermined schedule, by calling it mechanically via an API (Application Programming Interface) or Webhook, or by other methods.

[0031] For example, when a user invokes the AI ​​assistant function, the user terminal 30 displays a window W indicating the AI ​​assistant function on the display unit 35, as shown in the upper part of Figure 5. In the example shown in the upper part of Figure 5, window W includes an input form F that accepts input from the user prompt, which is a prompt entered by the user. When the user enters the user prompt and selects button B, a search is performed based on the search settings specified for the AI ​​assistant function.

[0032] In the upper example of Figure 5, the customer support application is specified as the data source, so the business support server 10 performs a RAG search based on the customer support application. The business support server 10 does not search for applications other than the customer support application. The business support server 10 requests the AI ​​server 40 to generate an answer by sending the data obtained from the customer support application along with a user prompt. When the AI ​​server 40 receives the request from the business support server 10, it causes the AI ​​to generate an answer. The business support server 10 sends the output from the AI ​​to the user terminal 30. As shown in the lower part of Figure 5, the user terminal 30 displays the output from the AI ​​in window W.

[0033] As described above, the business support system 1 of this embodiment offers increased search flexibility in RAG by allowing the user to specify search settings. For example, the user can specify data sources that are effective in supporting their own work, so the business support system 1 can prevent unnecessary data sources from being included in the search. As a result, the AI ​​can generate answers for business support based on search results that use the data sources desired by the user as the population. This allows the business support system 1 to improve the accuracy of the output from the AI. The details of the business support system 1 will be described below.

[0034] [3. Functions provided by the business support system] Figure 6 shows an example of the functions implemented by the business support system 1. In Figure 6, the functions implemented by the business support server 10 and the database server 20 are shown among the functions implemented by the business support system 1.

[0035] [3-1. Functions implemented by the business support server] For example, the business support server 10 includes a data storage unit 100, a search setting storage unit 101, a search unit 102, and a business support unit 103. The data storage unit 100 is implemented by a storage unit 12. Each of the search setting storage unit 101, the search unit 102, and the business support unit 103 is implemented by a control unit 11.

[0036] [Data Storage Unit] The data storage unit 100 stores various data related to the business support service. In this embodiment, we take the example of a case where the data storage unit 100 has a storage area for each of the multiple tenants (for example, an organization, a non-organizational group, or an individual) that use the business support service. That is, we take the example of a case where the storage area of ​​the data storage unit 100 is divided for each tenant. The storage area of ​​a certain tenant stores the data of that tenant. A tenant can also be defined as a subscriber to the business support service. For example, the storage area of ​​a certain tenant stores a search settings database DB1 in which the search settings specified by the user of that tenant are stored.

[0037] Figure 7 shows an example of a search settings database DB1. For example, the search settings database DB1 stores a function ID that can identify an AI assistant function, the name of the AI ​​assistant function, a user ID that can identify the user who specified the search settings, and the search settings. Any information may be stored in the search settings database DB1. The information stored in the search settings database DB1 is not limited to the example in Figure 7. For example, the search settings database DB1 may store a history of search settings.

[0038] For example, when a user in a tenant specifies a search setting for a certain AI assistant function, the business support server 10 issues a function ID that does not overlap with other function IDs. The business support server 10 updates the search setting database DB1 stored in the tenant's storage area so that the function ID and name of the AI ​​assistant function are associated with the search setting. When a user in a tenant changes the search setting for a certain AI assistant function, the business support server 10 updates the search setting database DB1 stored in the tenant's storage area so that the search setting associated with the function ID and name of the AI ​​assistant function is changed.

[0039] While it is possible to have a separate AI for each AI assistant function, in this embodiment, a common AI is used for multiple AI assistant functions. In other words, the AI ​​assistant function is the superficial aspect from the user's perspective, and the AI ​​behind it is a single AI managed by AI System 2. Therefore, the function ID and name of the AI ​​assistant function can also be considered information that identifies the search settings.

[0040] Furthermore, the storage area of ​​the data storage unit 100 does not need to be separated for each tenant. In this case, the data of each of multiple tenants may be stored in the same storage area. The data of each of multiple tenants may be consolidated in a common database stored in the data storage unit 100. For example, the data storage unit 100 may store a search setting database DB1 common to multiple tenants. The method of managing the data of each tenant is not limited to the example of this embodiment. The method of managing the data of each tenant may be the same as that of known business support services.

[0041] Furthermore, the data stored in the data storage unit 100 is not limited to the examples described above. The data storage unit 100 can store any data. For example, the data storage unit 100 may store data necessary for displaying each screen, such as the portal screen SC1, the application screen SC2, the list screen SC3, and the settings screen SC4. The data storage unit 100 may also store a database containing information for users of individual tenants to log in to the business support service.

[0042] [Search Settings Save Section] The search settings storage unit 101 stores search settings related to RAG searches, which are specified by users of business support services that assist with business operations. The search settings are specified by users other than the operator of the business support service, not the operator. For example, the search settings may be specified by a user with administrative privileges in a tenant, or by a user without administrative privileges. Multiple users in a tenant may have administrative privileges for specifying search settings.

[0043] The search setting can be any setting for RAG's search. For example, the search setting may be information that identifies the business database DB2 itself that is searched by RAG, information that identifies the database server 20 that stores the business database DB2 that is searched by RAG, information that identifies a part of the business database DB2 that is searched by RAG, or other information. In this embodiment, we take the example where the search setting indicates an application that is searched by RAG. The search setting may also indicate a business database DB2 other than an application. The search setting may indicate multiple business database DB2s. An example of another search setting will be explained in the modification section below.

[0044] Saving search settings means recording data that indicates the search settings. In this embodiment, we take the example of the case where the search setting storage unit 101 saves the search settings in the data storage unit 100. For example, the search setting storage unit 101 saves the search settings in the search setting database DB1 of the data storage unit 100. The search setting storage unit 101 may save the search settings in another location. For example, the search setting storage unit 101 may save the search settings in a database other than the search setting database DB1, in a computer other than the business support server 10 (for example, a database server 20), or in an information storage medium.

[0045] For example, when a user in a tenant specifies a search setting from the settings screen SC4, the user terminal 30 sends the search setting to the business support server 10. When the business support server 10 receives the search setting, the search setting storage unit 101 stores the search setting in the search setting database DB1 stored in the tenant's storage area. Similarly, when a user in a tenant changes a previously specified search setting, the search setting storage unit 101 stores the changed search setting in the search setting database DB1 stored in the tenant's storage area.

[0046] [Search section] The search unit 102 searches for input business data, which is business data to be input to the AI ​​that supports the business, from the business database DB2, which stores business data related to the business, based on the search settings. The business database DB2 that the search unit 102 searches is not limited to one, but may be multiple. Input business data is business data to be input to the AI. Input business data includes at least one business data that is found in the search.

[0047] In this embodiment, we take the example where the search setting indicates the business database DB2 to be searched. Therefore, the search unit 102 searches the business database DB2 indicated by the search setting out of all business database DB2s. The search unit 102 does not search any other business database DB2s other than the business database DB2 indicated by the search setting. In other words, the search unit 102 uses the business database DB2 indicated by the search setting as the search population and excludes the other business database DB2s from the search population.

[0048] For example, the search unit 102 obtains a user prompt from the user terminal 30 and searches the business database DB2 indicated by the search settings based on the user prompt. The search unit 102 may use the user prompt as the search query directly, or it may perform morphological analysis on the user prompt to obtain the search query. The search unit 102 may also obtain keywords from the search query and use those keywords as the search query, as shown in the modified example described later.

[0049] The user prompt-based search may be performed using a method similar to that of a well-known RAG. The search unit 102 may search for business data matching the user prompt from the business database DB2 based on a well-known search engine. The actual processing of the search may be performed by the search unit 102 of the business support server 10 or by the search unit 201 of the database server 20. The search unit 102 may search for business data matching the user prompt from the business database DB2 based on a machine learning model specialized for searching. If the user does not enter a user prompt, the search unit 102 may perform a search based on a pre-prepared search query. In this case, the pre-prepared search query corresponds to the search setting.

[0050] In this embodiment, we take the example of a case where the database server 20 stores the business database DB2, so the search unit 102 requests the database server 20 to perform a search based on the business database DB2 indicated by the search settings. The request is assumed to include a search query, but the database server 20 may obtain the search query on its side. For example, the request may include a user prompt, and the database server 20 may obtain the search query from the user prompt.

[0051] For example, the database server 20 performs a search based on the search unit 201 described later and sends the result to the business support server 10. The search unit 102 retrieves the search result from the database server 20. If the business database DB2 is stored in the business support server 10, the search unit 102 can simply search the business database DB2 it stores without requesting a search from the database server 20.

[0052] [Business Support Department] The Business Support Unit 103 provides business support based on the output from the AI, which receives the input data. Input data is the data that is input to the AI. Input business data is an example of input data. Input data may include content other than what the input business data indicates. For example, input data may include not only what the input business data indicates, but also at least one of user prompts and system prompts, as described below.

[0053] The output from the AI ​​can also be described as the calculation result calculated by the AI ​​in accordance with the input data. The output from the AI ​​can also be described as the answer from the AI. In this embodiment, we take the case where the AI ​​is managed by AI system 2 as an example, so the business support unit 103 sends input data to AI system 2 and obtains the output from the AI ​​from AI system 2. If business support system 1 manages the AI, the business support unit 103 may obtain the output from the AI ​​without communicating with AI system 2.

[0054] Supporting business operations means that the business support unit 103 provides the output from the AI ​​to the user terminal 30. In this embodiment, the business support unit 103 supports business operations by displaying a window W showing the output from the AI ​​on the user terminal 30. That is, the business support unit 103 supports business operations by sending data for displaying window W to the user terminal 30. The business support unit 103 may also support business operations by displaying the output from the AI ​​on the entire screen instead of window W. The business support unit 103 may also provide the output from the AI ​​to the user terminal 30 using other means such as email instead of displaying it on the screen.

[0055] Figure 8 shows an example of AI input and output. AI is a program capable of generating electronic products based on input data, which is information input to it. AI is sometimes called generative AI. There are various definitions of AI, but the AI ​​in this embodiment may be AI in various known definitions. The AI ​​in this embodiment includes various AIs that a person skilled in the art would recognize as AI at the time of filing. For example, generative AI may be not only AI developed using machine learning methods, but also AI developed using methods other than machine learning.

[0056] For example, the AI ​​may be a large-scale language model, a machine learning model not classified as a large-scale language model (a model that has been trained using machine learning techniques, such as GANs (Generative Adversarial Networks) or VAEs (Variational Autoencoders)), or another model. In this embodiment, we take the example of a pre-trained large-scale language model corresponding to the AI. The large-scale language model may be of any type. For example, the large-scale language model may be a transformer-type model such as GPT (Generative Pre-trained Transformer) or BERT (Bidirectional Encoder Representations from Transformers), or another model not classified as a transformer (for example, a neural network).

[0057] For example, an AI includes a program that outlines a series of processes from the input of data to the output, and parameters referenced by that program. The parameters may be incorporated as part of the AI's program. The AI's parameters may be similar to publicly known parameters. For example, the AI's parameters may be weights and biases. The AI ​​may include other information besides the program and parameters. The AI ​​may be a single-modal type capable of processing only a specific format of input data, or a multimodal type capable of processing multiple formats of input data.

[0058] For example, an AI may include an input layer that processes the input data, an intermediate layer that computes an embedding representation of the input data based on parameters, and an output layer that produces an output corresponding to the embedding representation. The embedding representation is information that the AI ​​uses to recognize the input data. Embedding representations are sometimes called features or feature vectors. Embedding representations may be multidimensional vectors or other forms (e.g., arrays, matrices, single numbers, or multiple numbers). The input data may be divided into units called tokens. The AI ​​may compute an embedding representation for each individual token. The AI ​​may predict the rest of the input data based on the order of the embedding representations of the individual tokens.

[0059] In this embodiment, it is assumed that various training data have already been trained on the AI. If the AI ​​is a model capable of natural language processing, such as a large-scale language model, then it is assumed that the AI ​​has already been trained on various texts as training data. The AI's parameters have been adjusted through training on the training data. A general-purpose AI that has been pre-trained may be used as is, or an AI that has been fine-tuned for the business support system 1 may be used. The AI's training method may be the same as known machine learning methods. Similar to the definition of AI, there are various definitions of machine learning, but the machine learning in this embodiment includes various known definitions. For example, deep learning is also included in machine learning.

[0060] In this embodiment, we take the example of a case where the AI ​​is managed by AI system 2, so the input to the AI ​​can also be said to be the information that the business support server 10 sends to AI system 2. The output to the AI ​​can also be said to be the information that the business support server 10 receives from AI system 2. In this embodiment, the output to the AI ​​is information for supporting business operations. In the example in Figure 8, we take the example of a case where the estimation result is natural language text (string), but the estimation result may be in a format other than text (for example, a single number, multiple numbers, an image, or program code).

[0061] For example, the AI ​​may receive a system prompt, which is a prompt provided by the business support system 1. A prompt is an instruction to the AI. In this embodiment, we take the example of a prompt being text written in natural language, but a prompt may be text written in a language other than natural language (e.g., a programming language or a database language). A prompt may also be in a format other than text (e.g., a document format, an image format, or a tabular format).

[0062] In this embodiment, the system prompt is assumed to be stored in the data storage unit 100. The system prompt may also be stored in a computer other than the business support server 10, or in an information storage medium. The business support unit 103 can obtain the system prompt from the data storage unit 100, the other computer, or the information storage medium. User prompts may also be input to the AI. Both user prompts and system prompts may be input to the AI.

[0063] For example, a system prompt indicates the processing that the AI ​​should perform. In this embodiment, we take the example of a case where a general-purpose AI that is not specialized for a particular task is used. Therefore, the system prompt indicates the AI's task. The task is the content of the processing that the AI ​​should perform. The system prompt may also indicate the specific content that the AI ​​should output. The system prompt may also indicate what will be input as input data. The system prompt may also indicate the AI's role. In this embodiment, we take the example of a case where the system prompt is prepared by the administrator of the business support service, but the system prompt may also be prepared by the user.

[0064] In the example in Figure 8, the system prompt displays text such as, "You are an AI that assists the user's work. Based on the input data you provide, please generate information that assists the user's work." The system prompt can be any wording; it is not limited to the wording in Figure 8. The business support unit 103 inputs the system prompt, user prompt, and input data including the input business data to the AI.

[0065] In the example in Figure 8, the user prompt contains content related to a specific page in a product manual. Therefore, the search unit 102 searches for business data related to that content and obtains it as input business data. For example, among the inquiry response applications, the record containing content related to that content corresponds to the input business data. The AI ​​uses the content of the record indicated by the input business data as a reference to generate and output information to support the work (for example, content related to the specific page in the manual mentioned above).

[0066] In this embodiment, since the AI ​​system 2 manages the AI, the business support unit 103 inputs input data to the AI ​​by sending input data to the AI ​​system 2. When the AI ​​system 2 obtains input data from the business support server 10, it inputs the input data to the AI ​​it stores. If the actual data of the AI ​​is stored in the data storage unit 100, the business support unit 103 only needs to input the input data to the AI ​​stored in the data storage unit 100 without communicating with an external computer.

[0067] For example, the AI ​​calculates an embedding representation of the input data it receives, based on parameters adjusted through pre-training. The AI ​​may divide the input data into units called tokens and then calculate the embedding representation of each token. If the AI ​​includes multiple hidden layers, it may repeat the calculation of embedding representations based on each hidden layer. The AI ​​generates and outputs an estimation result corresponding to the embedding representation. When the input data is divided into tokens, the AI ​​generates and outputs an estimation result corresponding to the sequence of embedding representations of each token. The internal processing of the AI ​​is not limited to the example of this embodiment. The internal processing of the AI ​​may be the same as that of known AIs.

[0068] For example, the business support unit 103 acquires output from the AI. In this embodiment, since the AI ​​system 2 manages the AI, the AI ​​system 2 transmits output from the AI ​​to the business support server 10. The business support unit 103 acquires estimation results from the AI ​​system 2. If the actual AI data is stored in the data storage unit 100, the business support unit 103 only needs to acquire output from the AI ​​without communicating with an external computer.

[0069] Furthermore, the AI ​​may not be a general-purpose AI specialized for a specific task, but rather an AI specialized for the task of business support. In this case, it is assumed that the AI ​​has learned training data showing the relationship between training inputs and outputs. The learning of the training data may be performed by the business support server 10 or by another computer. When the training input data and the correct output shown by the training data are input to the AI, the AI ​​is trained to output the training estimation result shown by the training data.

[0070] Furthermore, the AI's learning may be a known learning method used in machine learning techniques. For example, the AI ​​may learn the training data using backpropagation or gradient descent. If the AI ​​is specialized for a particular task, it can generate appropriate estimation results even without any prompt input, so the business support unit 103 may input input data to the AI ​​without inputting any prompts. This mode in which estimation results are generated without prompt input is also within the scope of this disclosure.

[0071] [3-2. Functions implemented by the database server] For example, the database server 20 includes a data storage unit 200 and a search unit 201. The data storage unit 200 is implemented by a storage unit 22. The search unit 201 is implemented by a control unit 11.

[0072] [Data Storage Unit] The data storage unit 200 stores various data related to the business support service. In this embodiment, we take the example that the data storage unit 200, like the data storage unit 100, has separate storage areas for each of the multiple tenants. For example, the storage area for one tenant stores a business database DB2 containing business data related to that tenant's business. The business database DB2 may be separate for each business support function, or it may be common to multiple business support functions.

[0073] Figure 9 shows an example of a business database DB2. Figure 9 shows an example of an application's business database DB2. The business database DB2 may also be a database for other business support functions besides the application. For example, the business database DB2 stores individual records of the application. A record is an example of business data. The business database DB2 may also store other information. For example, the business database DB2 may store an index for searching individual records, an ID that uniquely identifies the application, the application name, application access rights, application settings, or other information.

[0074] For example, business data includes the names of application fields, the values ​​of individual fields, files uploaded in the non-field sections, comments registered in records, and other information. When a user adds a record to an application, that record is added to the application's business database DB2. When a user updates a record in an application, the contents of that record stored in the application's business database DB2 are updated.

[0075] Similarly, business data for business support functions other than the application is also stored in the business database DB2. For example, business data for the scheduling function shows the specific details of the schedule registered in the scheduling function (e.g., the date and time of the schedule, participants, location, or memo). Business data for the communication function shows the specific details of the post registered in a thread or chat (e.g., the poster, comments, the name of the thread or chat room where the post was made, or the date and time of the post). Business data for the email management function shows the content of individual emails. Similarly, for other business support functions, the business data shows the details of the business related to those other business support functions.

[0076] Furthermore, similar to the data storage unit 100, the storage area of ​​the data storage unit 200 does not need to be divided for each tenant. Also, the data stored in the data storage unit 200 is not limited to the above example. The data storage unit 200 can store any data. For example, the data storage unit 200 may store a search engine program.

[0077] [Search section] The search unit 201 searches the business database DB2 based on the search query obtained from the business support server 10. For example, the search unit 201 may obtain search settings from the business support server 10. If a data source is indicated in the search settings, the search unit 201 identifies the business database DB2 or a part thereof to be searched based on the data source indicated in the search settings. Based on a known search engine, the search unit 201 obtains business data corresponding to the search query from the identified search target as input business data. The application of search settings is a novel configuration compared to conventional technology, but the search processing after the application of search settings may be the same as the search used in known RAGs. The search unit 201 transmits the input business data that was found in the search to the business support server 10.

[0078] [4. Processes executed by the business support system] Figure 10 shows an example of processing performed by the business support system 1. The control units 11, 21, 31, and 41 execute the programs stored in the memory units 12, 22, 32, and 42, respectively, thereby executing the processing shown in Figure 10. The processing shown in Figure 10 is an example of the processing of the business support method of this embodiment. Figure 10 shows the processing related to the AI ​​assistant function among the processing performed by the business support system 1.

[0079] As shown in Figure 10, when the user terminal 30 specifies the URL of the business support service website, it performs a process with the business support server 10 to allow the user to log in to the business support service (S1). The business support server 10 performs a process with the user terminal 30 to save the search settings (S2). The process in S2 may be the same as the process described in the search setting saving unit 101. AI assistant functions associated with a function ID and name in the search settings stored in the search setting database DB1 become selectable from icon I2.

[0080] When a user specifies an AI assistant function, the business support server 10 executes a process to activate the AI ​​assistant function selected by the user in communication with the user terminal 30 (S3). In S3, the business support server 10 displays a window W on the user terminal 30 that shows the AI ​​assistant function specified by the user. The business support server 10 then executes a process to obtain a user prompt in communication with the user terminal 30 (S4). In S4, the user terminal 30 sends the user prompt entered in input form F to the business support server 10.

[0081] The business support server 10 performs a search with the database server 20 based on the search settings (S5). The process in S5 may be the same as the process described in the search unit 102. The business support server 10 performs a process with the AI ​​server 40 to obtain output from the AI ​​based on the search settings (S6). The business support server 10 performs a process to support the business based on the output from the AI ​​obtained in S6 (S7), and this process ends. The processes in S6 and S7 may be the same as the processes described in the business support unit 103. When the process in S7 is executed, the response from the AI ​​is displayed in window W, as shown in the lower part of Figure 5.

[0082] [5. Summary of Embodiments] The business support system 1 of this embodiment stores search settings related to RAG searches, which are specified by the user using the business support service. Based on the search settings, the business support system 1 searches for input data to be input to the AI ​​that supports the business from the business database DB2, which stores business data related to the business. The business support system 1 supports the business based on the output from the AI ​​to which the input data has been input. As a result, the business support system 1 can perform searches according to the search settings specified by the user, thereby improving the accuracy of the output from the AI. For example, the user can specify data sources that are effective for business support and prevent unnecessary data sources from being included in the search. As a result, the AI ​​can generate answers for business support based on search results that use the data sources desired by the user as the population.

[0083] [6. Variant] This disclosure is not limited to the embodiments described above. It may be modified as appropriate without departing from the spirit of this disclosure.

[0084] Figure 11 shows an example of the functions implemented in the modified business support system 1. As shown in Figure 11, in the modified version described later, the business support server 10 includes a user prompt acquisition unit 104, a keyword acquisition unit 105, and an access rights setting acquisition unit 106. Each of the user prompt acquisition unit 104, the keyword acquisition unit 105, and the access rights setting acquisition unit 106 is implemented by the control unit 11.

[0085] [6-1. Variation 1] For example, as briefly explained in the embodiment, a user may specify search settings for each AI assistant function. The same user may specify search settings for multiple AI assistant functions, or multiple users may each specify search settings for different AI assistant functions. In other words, the user who specifies the search settings for one AI assistant function may be different from the user who specifies the search settings for another AI assistant function. The process by which a user specifies individual search settings may be the same as in the embodiment.

[0086] Figure 12 shows an example of the search setting database DB1 in Modification Example 1. The search setting storage unit 101 in Modification Example 1 stores multiple search settings. For example, the search setting storage unit 101 stores the search settings for each of multiple AI assistant functions in the search setting database DB1. When a user specifies a search setting for a particular AI assistant function, the user terminal 30 of that user sends the specified search setting to the business support server 10. The business support server 10 receives the search setting from the user terminal 30. The search setting storage unit 101 associates the function ID of the AI ​​assistant function, etc., with the search setting and stores it in the search setting database DB1.

[0087] In the example shown in Figure 12, the search settings for the customer support assistant and the search settings for the summary creation assistant are stored in the search settings database DB1. In Modification 1, the search settings for the customer support assistant are assumed to be the same as in the embodiment. The search settings for the summary creation assistant are assumed to specify a meeting minutes app containing meeting minutes as the data source. The search settings storage unit 101 can store any number of search settings. The number of search settings stored by the search settings storage unit 101 is not limited to two. The search settings storage unit 101 may store three or more search settings.

[0088] The search unit 102 in Modification 1 searches for input business data based on a search setting selected by the user using the AI ​​or another user from among multiple search settings. Hereafter, where simply referred to as "user," it means the user who specified the search setting. Where referred to as "other user," it means a user different from the user who specified the search setting. The other user may or may not belong to the same tenant as the user. For example, the other user may not belong to the same tenant as the user, but may be a person from an organization that has a cooperative relationship with the user's tenant.

[0089] For example, the search unit 102 retrieves search settings from the search setting database DB1, specifically those associated with the function ID of an AI assistant function selected by the user or another user from among multiple AI assistant functions. Based on these retrieved search settings, the search unit 102 performs a search of the input business data. While this differs from the embodiment in that the necessary search settings are used in the search from among multiple search settings, the flow in which the search unit 102 performs a search based on the search settings may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has retrieved the input business data may also be the same as in the embodiment.

[0090] The business support system 1 in the modified example 1 stores multiple search settings. The business support system 1 searches for input business data based on the search setting selected by the user using the AI ​​or another user from among the multiple search settings. As a result, the business support system 1 can search for input business data based on a search setting appropriate for supporting the user's or other users' work, thereby improving the convenience for the user or other users. The business support system 1 can support the user's or other users' work based on the output from the AI ​​that is tailored to the user or other users.

[0091] [6-2. Variation 2] For example, in Embodiment and Modification 1, the application that serves as the data source was described as a setting item for the search settings. The setting items for the search settings are not limited to the application that serves as the data source. Modification 2 describes examples of other setting items. The search settings only need to include at least one of the setting items described in Embodiment and Modification 1 and the setting items described in Modification 2.

[0092] In Modification 2, the search setting storage unit 101 stores the search field settings for fields related to the business database DB2 that are searchable by RAG as search settings. In Modification 2, the actual data of an application is described as an example of the business database DB2. The meaning of the application fields is as described in the embodiment. If business database DB2 of other business support functions other than the application is available for RAG search, the search field settings only need to be settings for the searchable fields among the business database DB2 of the other business support functions. Fields may be identified by any information adopted in publicly known databases. For example, fields may be identified by a code or name that can identify the field.

[0093] For example, the search field setting indicates at least one field to be searched from among multiple fields. Only one field may be searched, or multiple fields may be searched. In the example of the setting screen SC4 shown at the bottom of Figure 3 and at the top of Figure 4, the user selects a field from the application specified as the data source from the "Select Field" input form. The user terminal 30 sends the search field setting indicating the field selected by the user to the business support server 10. When the business support server 10 receives the search field setting, the search setting storage unit 101 stores the search field setting in the search setting database DB1.

[0094] In the modified example 2, the search unit 102 searches the input business data based on the search field settings. The search unit 102 searches the fields indicated by the search field settings out of all the fields in the business database DB2. The search unit 102 does not search any fields other than those indicated by the search field settings out of all the fields in the business database DB2. That is, the search unit 102 uses the data of the fields indicated by the search field settings in the business database DB2 as the search population, and excludes the data of other fields from the search population. Although the search population differs from that of the embodiment, the search process itself may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched the input business data may also be the same as in the embodiment.

[0095] In the modified version 2, the business support system 1 saves the search field settings for the fields to be searched in RAG as search settings. The business support system 1 searches the input business data based on the search field settings. As a result, the business support system 1 can perform searches according to the search field settings specified by the user, thereby improving the accuracy of the output from the AI. For example, the user can specify fields that are effective for business support and prevent unnecessary fields from being included in the search. As a result, the AI ​​can generate answers for business support based on search results that use the fields desired by the user as the population.

[0096] [6-3. Modification 3] For example, the search settings are not limited to the examples in Embodiment and Modifications 1 and 2. Modification 3 describes other examples of search settings. The search settings may include some or all of the settings described in Embodiment and Modification 1, the settings described in Modification 2, and the settings described in Modification 3.

[0097] The search setting storage unit 101 in Modification 3 stores search settings related to business data, which are the content to be searched in RAG, as search settings. The content related to business data is the specific content that the business data represents. For example, if individual records in an application correspond to business data, then the content of each record (for example, the value of a field, the file registered in the record, the comment registered in the record, or other content) corresponds to the content related to business data.

[0098] The content of the business data should be appropriate to the business support function. For example, the business data for the scheduling function may be the schedule content (e.g., date and time, participants, or notes). The business data for the communication function may be the content of the post (e.g., text, attachments, poster, or date and time). Similarly, the business data for other business support functions should be the specific content managed by those other business support functions.

[0099] The search criteria are the conditions under which business data is searchable. The search criteria can be any type of content that business data can take. Users can specify what they want to search for as the search criteria, or conversely, what they do not want to search for as the search criteria. In Modification 3, the input form for the search criteria is displayed on the settings screen SC4. For example, the user specifies the search criteria from this input form. The search criteria may be text, numbers, symbols other than text and numbers, on / off checkboxes, on / off radio buttons, or other settings.

[0100] For example, if the app's records correspond to business data, the user specifies the possible values ​​for each field as search criteria. If the user wants to search only records where a checkbox field is checked, they specify that the field is checked as a search criterion. Similarly, for other business support functions such as scheduling or communication functions, the user specifies the possible content of the business data (e.g., date and time or keywords) as search criterion. The user terminal 30 sends the search criterion specified by the user to the business support server 10. When the business support server 10 receives the search criterion, the search criterion storage unit 101 stores the search criterion in the search criterion database DB1.

[0101] In the modified example 3, the search unit 102 searches the input business data based on the search content settings. The search unit 102 searches business data from the business database DB2 that meets the search content settings. The search unit 102 does not search business data from the business database DB2 that does not meet the search content settings. In other words, the search unit 102 uses business data from the business database DB2 that meets the search content settings as the search population, and excludes business data with other content from the search population.

[0102] For example, if the app records correspond to business data, the search unit 102 searches for records in the app that satisfy the search criteria (for example, records where a checkbox for a specific field is checked). Similarly, for other business support functions such as scheduling or communication functions, the search unit 102 searches for business data in the business database DB2 that satisfies the search criteria (for example, a schedule for a specific date and time, or a post containing a specific keyword). Although the search population differs from that of the embodiment, the search process itself may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched the input business data may also be the same as in the embodiment.

[0103] In the modified example 3, the business support system 1 saves the search content settings related to the content to be searched in RAG as search settings. The business support system 1 searches the input business data based on the search content settings. As a result, the business support system 1 can perform searches according to the search content settings specified by the user, thereby improving the accuracy of the output from the AI. For example, the user can target business data with content that is effective for business support, and prevent irrelevant business data from being included in the search. As a result, the AI ​​can generate answers for business support based on search results that use business data with content desired by the user as the population.

[0104] [6-4. Modification 4] For example, the search setting storage unit 101 may be able to save search settings so that business database DB2 other than a specific location within the business support service, or business database DB2 other than the business support service, can be searched. The search unit 102 may be able to search business database DB2 other than a specific location within the business support service, or business database DB2 other than the business support service.

[0105] As explained to some extent in the Embodiments, Modification 4 takes the example of a case where the AI ​​assistant function is used in a specific location within the business support service. The location within the business support service is the online location where the business support service is provided. The location within the business support service can also be the page (screen) or business support function on which the business support service is provided. If the business support service is provided on a browser, the URL of the business support service corresponds to the location within the business support service. If the business support service is provided by a dedicated application, the screen within that application corresponds to the location within the business support service. The location within the business support service may be the business support server 10, the database server 20, another computer, or a storage area on an information storage medium.

[0106] Figure 13 shows an example of the screen displayed on the user terminal 30 in the modified example 4. For example, when the user selects the location tab on the settings screen SC4, the user terminal 30 can specify the location where the AI ​​assistant function, for which the user is specifying search settings, will be used, as shown in the upper part of Figure 13. In the upper example of Figure 13, the user specifies an app as the location. The user may specify multiple apps as multiple locations. When business support services are provided on a browser, individual apps are provided with separate URLs, so when a user specifies an app, the user also specifies a URL.

[0107] Users may specify locations other than the app. For example, users may specify business support functions other than the app, such as communication functions, email management functions, or scheduling functions. Users may also specify external SaaS (Software as a Service) or intranet systems as locations other than the app. Users may specify specific locations within individual business support functions, such as threads or chat rooms in the communication functions. Users may also specify locations unrelated to specific business support functions, such as the portal screen SC1. If the user does not specify a particular location, the AI ​​assistant function may be available from any location.

[0108] In Modification 4, when a user specifies a location, the search setting storage unit 101 associates the search setting specified by the user with location identification information that can identify the location specified by the user and stores it in the search setting database DB1. The location identification information can be any information that can identify a location within the business support service in some way. For example, the location identification information may be a URL, an ID assigned to an individual screen, information that can identify a business support function such as an app, or other information. A user using a certain AI assistant function or another user can access it from the location associated with the function ID and name of that AI assistant function.

[0109] In the modified example 4, the business support unit 103 provides business support based on the output from the AI ​​at a specific location among several locations within the business support service. The specific location is where the AI ​​assistant function is used. The specific location can also be described as the location where the AI ​​assistant function is invoked. The specific location can also be described as the location where a search is used according to the search settings. The business support unit 103 does not provide business support based on the output from the AI ​​at locations other than the specific location among the several locations. In other words, the business support unit 103 provides AI-based business support only at the specific location and does not provide AI-based business support at other locations.

[0110] In variation 4, the specific location is the location where location identification information is stored in the search settings database DB1. In the lower example of Figure 13, the inquiry response application is designated as the specific location. For example, the business support unit 103 refers to the search settings database DB1 to identify the specific location specified by the user and supports the work of the user or other users using that location based on the output from the AI. In the lower example of Figure 13, since the inquiry response application is designated as the specific location, the business support unit 103 supports the work of the inquiry response application based on the output from the AI.

[0111] Furthermore, location identification information indicating specific locations where AI-based business support is available may be stored in a database other than the search setting database DB1, a computer other than the business support server 10 (for example, a database server 20), or an information storage medium. In this case, the search setting storage unit 101 stores the location identification information in the other database, another computer, or an information storage medium. The business support unit 103 may identify a specific location based on the location identification information stored in the other database, another computer, or an information storage medium, and provide business support based on the output from the AI ​​when the use of AI is requested from that specific location.

[0112] The search setting storage unit 101 in Modification 4 can save search settings so that business databases DB2 in locations other than the specified location among multiple locations can be searched. In other words, the business databases DB2 that a user can specify as a data source are not limited to the specific location specified by the user. The user can specify business databases DB2 in other locations as a data source. The user is not required to specify business databases DB2 in other locations; it is optional for the user to specify business databases DB2 in other locations. If the user specifies business databases DB2 in other locations as a data source, the search setting storage unit 101 saves a search setting indicating that other location.

[0113] In the upper and lower examples of Figure 13, the specific location is the inquiry response application. Users can specify applications other than the inquiry response application as data sources. For example, suppose the user specifies an FAQ application containing various FAQs (Frequently Asked Questions) as the data source. In this case, the search setting storage unit 101 saves a search setting that indicates the FAQ application, which is another location, as the data source. If the user specifies multiple other locations, the search setting storage unit 101 saves a search setting that indicates multiple other locations as data sources. Suppose the user specifies the inquiry response application, which is a specific location. In this case, the search setting storage unit 101 saves a search setting that indicates the inquiry response application, which is a specific location, as the data source.

[0114] In the modified example 4, the search unit 102 searches the business database DB2 in another location if the search settings are saved to search the business database DB2 in another location. For example, if another location is indicated in the search settings, the search unit 102 uses the business database DB2 in another location as the search population and performs the search. This differs from the embodiment in that the business database DB2 in another location becomes the search population, but the search process itself may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched the input business data may also be the same as in the embodiment.

[0115] The above example describes a case where a business database DB2 located elsewhere within the business support service is searched, but a business database DB2 located outside the business support service may also be searched. In Modification 4, the business support service interacts with an external service. The external service is a different business support service from the business support system 1. Modification 4 gives an example where the operator of the external service and the operator of the business support service are different from each other, but these operators may be the same.

[0116] Figure 14 shows an example of the hardware configuration of Modification 4. Modification 4 takes the case where the business support system 1 can communicate with an external system 3 via a network N. For example, the external system 3 includes an external server 50. The external server 50 is a server computer for external services. The external server 50 includes a control unit 51, a storage unit 52, and a communication unit 53. The hardware configurations of the control unit 51, storage unit 52, and communication unit 53 may be the same as those of the control unit 11, storage unit 12, and communication unit 13, respectively. The hardware configuration of the external system 3 is not limited to the example in Figure 1, as is the case with the business support system 1.

[0117] For example, a user's tenant also uses external services. The external server 50 stores a business database DB2 containing business data for the external service. Although the specific content of the external service differs from that of the business support service, the meaning of the business support functions and business data provided by the external service may be the same as that of the business support service. For example, the external server 50 provides data management functions, scheduling functions, communication functions, email management functions, workflow functions, or other functions unique to the external service. The external server 50 stores the business database DB2 for these business support functions.

[0118] The search setting storage unit 101 in Modification 4 can save search settings so that a business database DB2 located outside the business support service is searched. The method for specifying the data source may be the same as in Embodiment and Modification 4. For example, the user specifies the business database DB2 of an external service as the data source from the setting screen SC4. If there are multiple external services, the user may specify the business database DB2 of each of the multiple external services. The search setting storage unit 101 saves the search settings that indicate the business database DB2 of the external service specified by the user as the data source.

[0119] Furthermore, since the business support service and the external service are interconnected, the business support server 10 is capable of identifying what kind of business database DB2 is available in the external service. The business support server 10 may identify which business database DB2 the user's tenant is using in the external service and display the name of the external service's business database DB2 on the settings screen SC4. The user may specify the external service's business database DB2 as the data source by specifying this name.

[0120] In the modified example 4, the search unit 102 searches the business database DB2 of a location outside the business support service when the search settings are saved to search locations outside the business support service. For example, if the search settings indicate that the business database DB2 of an external service is included, the search unit 102 uses the business database DB2 of the external service as the search population and performs the search. This differs from the embodiment in that the business database DB2 of an external service becomes the search population, but the search process itself may be the same as in the embodiment.

[0121] For example, the external server 50, in the same manner as the database server 20, searches the business database DB2 it stores based on a request from the business support server 10 and obtains the input business data. The external server 50 sends the input business data to the business support server 10. The search unit 102 obtains the input business data received by the business support server 10. If the business database DB2 within the business service is specified as the data source, the search unit 102 performs a search in the same manner as in the embodiment and obtains the input business data. The processing of the business support unit 103 after the search unit 102 has retrieved the input business data may also be the same as in the embodiment.

[0122] Furthermore, the database server 20 may store a business database DB2 equivalent to the external service's business database DB2 in the data storage unit 200 by synchronizing with the external server 50. In this case, the search unit 102 may perform a search using the external service's business database DB2 as a data source by requesting a search from the database server 20 without sending a search request to the external server 50. That is, the search unit 102 may perform a search using the external service's business database DB2 as a data source within the business support system 1 without communicating with the external server 50.

[0123] In the modified example 4, the business support system 1 can save search settings so that it searches business database DB2 outside of a specific location within the business support service, or business database DB2 outside of the business support service. The business support system 1 can search business database DB2 outside of a specific location within the business support service, or business database DB2 outside of the business support service. This allows the business support system 1 to search for input business data that is effective for supporting business operations, thereby improving the accuracy of the output from the AI. In the upper and lower examples of Figure 13, the business support system 1 can perform a search on a user or other user using a customer support assistant where the inquiry response app is specified as a specific location, and also search the FAQ app, which is another location, and provide the user or other user with output from the AI ​​corresponding to the input business data of the FAQ app. For example, the business support system 1 can use not only the business database DB2 within the business support service but also the business database DB2 of external services as a data source, thereby improving the accuracy of the output from the AI.

[0124] [6-5. Variation 5] For example, in the embodiment, a case was described in which the input business data search is performed based on a user prompt, but keywords corresponding to the search query may also be obtained based on the user prompt. Modification 5 describes a case in which keywords are obtained based on a user prompt, and the input business data search is performed based on those keywords.

[0125] The business support system 1 in modified example 5 includes a user prompt acquisition unit 104 and a keyword acquisition unit 105. The user prompt acquisition unit 104 acquires user prompts entered by the user using the AI ​​or by other users. For example, as shown in the upper part of Figure 5, when a user enters a user prompt into the input form F of window W, the user terminal 30 sends the user prompt to the business support server 10. The user prompt acquisition unit 104 acquires the user prompt received by the business support server 10.

[0126] The keyword acquisition unit 105 acquires keywords for searching input business data based on user prompts. Keywords are search queries in RAG searches. The keyword acquisition unit 105 can acquire keywords from user prompts in a predetermined manner. For example, the keyword acquisition unit 105 may acquire keywords from user prompts based on AI. Modification 5 describes a case where the AI ​​for keyword acquisition and the AI ​​into which input data is input are the same, but these AIs may be different from each other. That is, there may be an AI for keyword acquisition separate from the AI ​​described in Embodiments and Modifications 1 to 5.

[0127] Figure 15 shows an example of the processing flow for Modification Example 5. For example, the keyword acquisition unit 105 inputs a keyword acquisition prompt and a user prompt to the AI. The keyword acquisition prompt is a predetermined prompt for acquiring keywords. The keyword acquisition prompt indicates that keywords should be acquired from the user prompt. If the AI ​​is a large-scale language model, the keyword acquisition prompt is written in natural language. When the user prompt and the keyword acquisition prompt are input to the AI, it calculates their embedding representations and outputs keywords corresponding to the embedding representations. The keyword acquisition unit 105 acquires the keywords output from the AI.

[0128] Note that the method of obtaining keywords is not limited to the examples above. For example, the keyword acquisition unit 105 may acquire keywords from user prompts based on an AI specialized in keyword extraction, rather than a general-purpose large-scale language model. In this case, the AI ​​is assumed to have been trained with training data showing the relationship between training user prompts and correct keywords. The keyword acquisition unit 105 inputs user prompts to the trained AI and acquires the keywords output by the trained AI. The keyword acquisition unit 105 may also acquire keywords by performing morphological analysis on the user prompts. In this case, the keyword acquisition unit 105 may acquire nouns identified by morphological analysis as keywords.

[0129] In Modification 5, the search unit 102 searches the input business data based on the search settings and keywords. For example, the search unit 102 searches the business database DB2 using the keywords obtained by the keyword acquisition unit 105 as the search query. Although this is more specific than the embodiment in that keywords are used in the search, the search process itself may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched the input business data may also be the same as in the embodiment. Note that a fixed keyword may be specified as the search setting.

[0130] In the modified example 5, the business support system 1 acquires user prompts entered by the user utilizing the AI ​​or by other users. Based on the user prompts, the business support system 1 acquires keywords for searching input business data. Based on the search settings and keywords, the business support system 1 searches the input business data. As a result, the business support system 1 can search the input business data based on appropriate keywords corresponding to the user prompts, thereby improving the accuracy of the input business data search. Consequently, the business support system 1 can improve the accuracy of the output from the AI.

[0131] [6-6. Variation 6] For example, the appropriate search settings may differ depending on the user utilizing the AI ​​or other users. For instance, if a sales representative of a tenant uses the customer support assistant, it might be more appropriate for the data source to be an app primarily used by sales representatives. If a planning representative of the same tenant uses the customer support assistant, it might be more appropriate for the data source to be an app primarily used by planning representatives. Therefore, in Modification 6, we will explain a case where the search for input business data is performed based on appropriate search settings according to the user utilizing the AI ​​or other users.

[0132] The search setting storage unit 101 in Modification 6 can store search settings according to the user using the AI ​​or other users. For example, a user specifies user information about the user using the AI ​​or other users on the settings screen SC4. User information may be user identification information that can identify individual users or other users, or user attribute identification information that can identify the attributes of individual users or other users. Attributes can also be called classifications. For example, attributes may be the user's department, position, team, other group, age, age group, years of service, gender, or other attributes.

[0133] In variation 6, we take the case where departments correspond to user information as an example. For example, a user specifies the search settings for each of multiple departments on the settings screen SC4. That is, the user specifies the search settings for each department. The search setting storage unit 101 associates the user information indicating the department specified by the user with the search settings for that department and stores them in the search setting database DB1. The user does not need to specify the search settings for all departments. The user may specify the search settings for a particular department and the search settings for other departments. Similarly, if the user information indicates elements other than departments, the search setting storage unit 101 should associate the user information indicating the other elements with the search settings for those other elements and store them in the search setting database DB1.

[0134] In Modification 6, the search unit 102 searches for input business data based on search settings corresponding to the user using the AI ​​or other users. For example, the search unit 102 obtains user information of the user using the AI ​​or other users. User information of the user using the AI ​​or other users may be stored in the data storage unit 100 in advance, or it may be obtained from the user terminal 30. In Modification 6, the department corresponds to user information, so the search unit 102 obtains user information from the user database, which stores various information about the user or other users. Similarly, if the user information indicates elements other than the department, the search unit 102 can obtain user information from the user database. The search unit 102 may also obtain user information from other databases other than the user database, other computers other than the business support server 10, or information storage media.

[0135] For example, the search unit 102 retrieves search settings associated with the user information of a user utilizing the AI ​​or another user from the search settings database DB1. Based on the retrieved search settings, the search unit 102 performs a search for input business data. For example, if the user information of a user utilizing the AI ​​or another user indicates the Sales Department, the search unit 102 performs a search for input business data based on the search settings for the Sales Department. If the user information of a user utilizing the AI ​​or another user indicates the Planning Department, the search unit 102 performs a search for input business data based on the search settings for the Planning Department. This differs from the embodiment in that search settings corresponding to the user utilizing the AI ​​or another user are used in the search, but the search process itself may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched for input business data may also be the same as in the embodiment.

[0136] The business support system 1 in modified example 6 can save search settings according to the user using the AI ​​or other users. The business support system 1 searches for input business data based on the search settings according to the user using the AI ​​or other users. As a result, the business support system 1 can search for input business data according to the user using the AI ​​or other users, thereby improving the accuracy of the output from the AI.

[0137] [6-7. Variation 7] For example, there may be two business databases, DB2, one that can be accessed by the user using the AI ​​or other users, and another that cannot be accessed by the user using the AI ​​or other users. In this case, if input business data is retrieved from the business database DB2 that cannot be accessed by the user using the AI ​​or other users, there is a possibility that the user using the AI ​​or other users may gain access to information that they should not have been able to access. Therefore, in Modification 7, we will take the example of a case where a search is performed according to access rights.

[0138] The business support system 1 in Modification 7 includes an access rights setting acquisition unit 106. The access rights setting acquisition unit 106 acquires access rights settings related to the access rights of the business database DB2. In Modification 7, it is assumed that the access rights settings are included in the business database DB2. For example, the access rights setting acquisition unit 106 acquires the access rights settings of the business database DB2 to be searched from the database server 20. The access rights settings may be stored in a database other than the business database DB2, a computer other than the database server 20 (for example, the business support server 10), or an information storage medium. In this case, the access rights setting acquisition unit 106 may acquire the access rights settings from the other database, the other computer, or the information storage medium.

[0139] In the modified example 7, the search unit 102 searches for input business data based on the search settings and the access rights settings. For example, the search unit 102 searches for input business data based on the business database DB2 that the access rights settings indicate is accessible to the user using the AI ​​or other users, among the business database DB2 of the data source indicated by the search settings. That is, the search unit 102 identifies the business database DB2 to be used as the data source based on two conditions: that it is indicated as a data source in the search settings, and that it is accessible to the user using the AI ​​or other users. The search unit 102 searches for input business data from the identified business database DB2. Although the method for identifying the business database DB2 to be used as the data source differs from the embodiment, the search process itself may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched for input business data may also be the same as in the embodiment.

[0140] Furthermore, the access rights setting may not refer to access rights for the entire business database DB2, but rather to access rights for individual business data stored in the business database DB2. For example, if an application corresponds to the business database DB2, the access rights may refer to individual records of the application, or to access rights for a part of an individual record (e.g., a field). In this case, the search unit 102 can perform the search using business data or a part thereof that is accessible to the user using the AI ​​or other users as the data source.

[0141] In the modified example 7, the business support system 1 obtains access rights settings. Based on the search settings and access rights settings, the business support system 1 searches for input business data. This allows the business support system 1 to perform searches according to the access rights of the user using the AI ​​or other users, thus preventing the user using the AI ​​or other users from obtaining information they are not supposed to have access to.

[0142] [6-8. Variation 8] For example, as briefly explained in the embodiments, the AI ​​assistant function may have multiple user interface parts associated with it. User interface parts are parts that accept user input. User interface parts can be any part, such as buttons, icons, bars, text, or other parts. User interface parts may be displayed within window W or in a location separate from window W.

[0143] For example, in the case of a customer support assistant, separate parts may be displayed, such as parts for the sales department and parts for the planning department. In this case, search settings may be saved for each individual part. For example, if the sales department part is selected, the sales department app may become the data source, and if the planning department part is selected, the planning department app may become the data source. Modification example 8 describes a case where search settings are saved for each part.

[0144] Figure 16 shows an example of the search setting database DB1 in Modification 8. The search setting storage unit 101 in Modification 8 can store search settings in association with each of several parts that receive instructions for the user or other users to use the AI. In Modification 8, the user can specify search settings for each part from the setting screen SC4. In the example of the setting screen SC4 at the bottom of Figure 3 and at the top of Figure 4, the prompt button is an example of a part. The prompt button is associated with the system prompt entered by the user when specifying the search settings. System prompts will be explained in Modification 10 below.

[0145] For example, the user terminal 30 sends the search settings for each of the multiple parts specified by the user on the settings screen SC4 to the business support server 10. The search setting storage unit 101 stores the search settings for each of the multiple parts received by the business support server 10 in the search setting database DB1. As shown in the upper part of Figure 16, the search setting storage unit 101 associates part identification information that can identify a part with the search settings specified for that part and stores them in the search setting database DB1. In the example in the upper part of Figure 16, the user specifies search settings for parts for the sales department and parts for the planning department. The part identification information can be any information such as an ID or name assigned to the part on the screen.

[0146] In the modified example 8, the search unit 102 can search input business data based on search settings associated with a part selected by the AI ​​user or another user from among multiple parts. In the lower example of Figure 16, a part for the sales department (for example, button B showing the text "For Sales Department") and a part for the planning department (for example, button B showing the text "For Planning Department") are displayed in window W. The search unit 102 obtains the search settings associated with the part identification information of the part selected by the AI ​​user or another user from among these multiple parts from the search settings database DB1. Although it differs from the embodiment in that the necessary search settings are used in the search from among the search settings of each of the multiple parts, the flow in which the search unit 102 executes the search based on the search settings may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched the input business data may also be the same as in the embodiment.

[0147] The business support system 1 of the modified example 8 can associate and store search settings with each of several parts that receive instructions for the user or other users to use the AI. The business support system 1 can search input business data based on the search settings associated with the part selected by the user or other user using the AI ​​from among the multiple parts. As a result, the business support system 1 can search input business data based on the search settings corresponding to the part selected by the user or other user, thereby improving the convenience for the user or other user. The business support system 1 can support the work of the user or other user based on the output from the AI ​​corresponding to the user or other user.

[0148] [6-9. Modification 9] For example, as explained to some extent in Modification 4, AI functions may be available from specific locations within the business support service. Even if it is the same AI assistant function, the appropriate input business data may differ depending on the location where the AI ​​assistant function is installed. Therefore, Modification 9 explains a case where search settings are saved according to the location where the AI ​​assistant function is installed.

[0149] The search setting storage unit 101 in Modification 9 can associate and save search settings with each of the multiple locations where AI is used in the business support service. The locations where AI is used are the same as those described in Modification 4. In the example in Figure 13, the location where the AI ​​assistant function is installed corresponds to the location where AI is used. The meaning of "location" is as explained in Modification 4. In Modification 9, the application is described as the location where AI is used. The user specifies multiple locations where AI is used from the settings screen SC4.

[0150] For example, suppose a user specifies two locations: an inquiry response app and an FAQ app. Suppose the user specifies that the data source for the inquiry response app is the inquiry response app in the search settings. Suppose the user specifies that the data source for the FAQ app is the FAQ app in the search settings. In this case, the user terminal 30 sends to the business support server 10 the search settings for the inquiry response app, indicating that the data source is the inquiry response app, and the search settings for the FAQ app, indicating that the data source is the FAQ app.

[0151] For example, if the user specifies a different location, the user will specify the search settings for each of the multiple locations. The user terminal 30 sends the location identification information for each of the multiple locations and the search settings specified for that location to the business support server 10. The search setting storage unit 101 stores the combination of location identification information and search settings received by the business support server 10 in the search setting database DB1. If the user specifies two locations, such as an inquiry response app and an FAQ app, the search setting storage unit 101 stores the search settings in the search setting database DB1 in association with the location identification information indicating the inquiry response app. The search setting storage unit 101 also stores the search settings in the search setting database DB1 in association with the location identification information indicating the FAQ app.

[0152] The search unit 102 in Modification 9 can search input business data based on search settings associated with locations where the user or other users utilize the AI, among multiple locations. The search unit 102 retrieves from the search setting database DB1 the search settings associated with the location selected by the user or other users utilizing the AI, from among the search settings associated with the location identification information of each of the multiple locations. Although it differs from the embodiment in that the necessary search settings are used in the search from among the search settings associated with the location identification information of each of the multiple locations, the flow in which the search unit 102 executes the search based on the search settings may be the same as in the embodiment. The processing of the business support unit 103 after the search unit 102 has searched the input business data may also be the same as in the embodiment.

[0153] The business support system 1 of the modified example 9 can associate and store search settings with each of the multiple locations where AI is used in the business support service. The business support system 1 can search input business data based on the search settings associated with the location where the user or other users use AI among the multiple locations. As a result, the business support system 1 can search input business data based on search settings according to the location where the user or other users use AI, thereby improving the convenience for the user or other users. The business support system 1 can support the work of the user or other users based on the output from AI that is tailored to the user or other users.

[0154] [6-10. Variation 10] For example, as briefly explained in the embodiment, the user may specify a system prompt from the settings screen SC4. The system prompt is not used in RAG searches and is entered into the AI ​​along with the input business data. In the embodiment, the system prompt is also shown in Figure 8, but the system prompt in Figure 8 is a prompt specified by the operator of the business support service. In the modified example 10, the user using the business support service (the customer for the operator) specifies the system prompt.

[0155] In the modified example 10, the search setting storage unit 101 stores the search setting and the system prompt entered by the user in association. In the example of the setting screen SC4 shown at the bottom of Figure 3 and at the top of Figure 4, the user enters arbitrary text into the system prompt input form. For example, the user describes in natural language the task that the AI ​​assistant function should perform while specifying the search setting. When the user specifies the system prompt and search setting on the setting screen SC4, the user terminal 30 sends the search setting and system prompt to the business support server 10. The search setting storage unit 101 stores the search setting and system prompt received by the business support server 10 in the search setting database DB1. The system prompt specified by the user may be included in the search setting.

[0156] Figure 17 shows an example of AI input and output in Modification 10. The business support unit 103 in Modification 10 supports business operations based on the input business data, system prompts, and output from the AI. As shown in Figure 17, the business support unit 103 inputs input data to the AI, including system prompts specified by the user, in addition to the input shown in Figure 8 described in the embodiment. Although the content of the input data differs from the embodiment, the AI ​​processing after the input data is received is the same as in the embodiment. The flow of business support by providing the user with the output from the AI ​​is also the same as in the embodiment.

[0157] In modified example 10, the business support system 1 associates and stores search settings with system prompts entered by the user. The business support system 1 assists with business operations based on the input business data, system prompts, and the output from the AI ​​that receives them. As a result, the business support system 1 can assist with business operations based on appropriate output corresponding to the system prompts entered by the user, thereby improving the accuracy of the output from the AI.

[0158] [6-11. Other variations] For example, you may combine two or more of the variations 1 to 10.

[0159] For example, a function described as being implemented by the business support server 10 may be implemented by the user terminal 30. In this case, the function can be implemented by a browser script or an application installed on the user terminal 30. For example, each function may be shared among multiple computers, or it may be implemented by a single computer. [Explanation of Symbols]

[0160] 1 Business support system, 2 AI system, 3 External system, 10 Business support server, 11,21,31,41,51 Control unit, 12,22,32,42,52 Storage unit, 13,23,33,43,53 Communication unit, 20 Database server, 30 User terminal, 34 Operation unit, 35 Display unit, 40 AI server, 50 External server, 100 Data storage unit, 101 Search setting storage unit, 102 Search unit, 103 Business support unit, 104 User prompt acquisition unit, 105 Keyword acquisition unit, 106 Access rights setting acquisition unit, 200 Data storage unit, 201 Search unit, B30 Button, DB1 Search setting database, DB2 Business database, L20 List, SC1 Portal screen, SC2 Application screen, SC3 List screen, SC4 Settings screen, B Button, F Input form, N Network, I1,I2 Icon, W window.

Claims

1. A search setting storage unit that stores the search settings specified by a user of a business support service that assists in business operations, relating to the search settings for RAG (Retrieval-augmented generation) A search unit searches for input business data, which is the business data to be input to the AI ​​(Artificial Intelligence) that supports the business, from a business database containing business data related to the business, based on the search settings. Based on the output from the AI ​​into which the aforementioned input business data has been input, a business support unit assists in the aforementioned business operations. Business support systems including this.

2. The search setting storage unit stores multiple search settings, The search unit searches the input business data based on the search settings selected by the user using the AI ​​or another user from among the plurality of search settings. The business support system according to claim 1.

3. The search setting storage unit stores the search field settings for the fields in the business database that are subject to search by the RAG as the search settings. The search unit searches the input business data based on the search field settings. The business support system according to claim 1 or 2.

4. The search setting storage unit stores the content relating to the business data, which is the content to be searched by the RAG, as the search setting. The search unit searches the input business data based on the search content settings. The business support system according to claim 1 or 2.

5. The search setting storage unit is capable of saving the search settings so that the business database other than a specific location within the business support service, or the business database other than the business support service, can be searched. The search unit is capable of searching the business database other than the specific location within the business support service, or the business database other than the business support service. The business support system according to claim 1 or 2.

6. The aforementioned business support system is A user prompt acquisition unit that acquires user prompts entered by the user or another user using the AI, A keyword acquisition unit that acquires keywords for searching the input business data based on the user prompt, It further includes, The search unit searches the input business data based on the search settings and the keywords. The business support system according to claim 1 or 2.

7. The search settings storage unit is capable of saving the search settings according to the user or other users who use the AI. The search unit searches the input business data based on the search settings corresponding to the user or other users using the AI. The business support system according to claim 1 or 2.

8. The aforementioned business support system further includes an access rights setting acquisition unit that acquires access rights settings related to the access rights of the business database, The search unit searches the input business data based on the search settings and the access rights settings. The business support system according to claim 1 or 2.

9. The search setting storage unit is capable of storing the search setting in association with each of the multiple parts that receive instructions from the user or another user for using the AI. The search unit is capable of searching the input business data based on the search settings associated with the part selected by the user or another user using the AI ​​from among the plurality of parts. The business support system according to claim 1 or 2.

10. The search setting storage unit is capable of associating and saving the search settings with each of the multiple locations where the AI ​​is used in the business support service. The search unit is capable of searching the input business data based on the search settings associated with the location where the user or another user utilizes the AI, among the plurality of locations. The business support system according to claim 1 or 2.

11. The search settings storage unit stores the search settings and the system prompts entered by the user in association with each other. The aforementioned business support unit supports the aforementioned business based on the input business data, the system prompt, and the output from the AI ​​that has been input. The business support system according to claim 1 or 2.

12. A search setting related to RAG (Retrieval-augmented generation) searches, which saves the search setting specified by a user using a business support service that assists with business operations. Based on the aforementioned search settings, the system searches for the input business data, which is the business data to be input to the AI ​​(Artificial Intelligence) that supports the aforementioned business, from the business database in which the business data related to the aforementioned business is stored. Based on the output from the AI ​​into which the aforementioned input business data has been input, the AI ​​supports the aforementioned business. Business support methods.

13. A search setting for RAG (Retrieval-augmented generation) searches, wherein the search setting is stored by a user of a business support service that assists with business operations. A search unit searches for input business data, which is the business data to be input to the AI ​​(Artificial Intelligence) that supports the business, from a business database containing business data related to the business, based on the search settings. Based on the output from the AI ​​into which the aforementioned input business data has been input, the business support unit supports the aforementioned business. A program that makes a computer function.

Images