Information processing device, system, and method of operating the system

Abstract

To efficiently improve the negotiation skills of sales representatives. [Solution] The information processing device includes: an acquisition unit that acquires information about conversations with a customer regarding business negotiations to encourage the customer to purchase a product; a storage unit that stores the customer's attributes extracted from the conversation information by a large-scale language model; a control unit that uses a first model that has learned the correspondence between the conversation information and the progress stages of the business negotiations, and a second model that has learned the correspondence between the combination of the attributes and the progress stages and the purchased products, to select a product to encourage a new customer to purchase based on information from a conversation with a new customer.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, a system, and an operating method of the system.

Background Art

[0002] In life insurance, medical insurance, etc., a variety of insurance products have been developed, and the contract contents of each insurance product vary widely depending on combinations of insurance premiums, insured amounts, insured contents, etc. A salesperson of an insurance company is required to listen to the needs of a customer in a business negotiation with the customer and propose an insurance product that meets the needs of the customer. Information processing systems that assist in the task of selecting an insurance product that meets the needs of a customer from a wide variety of insurance products have been variously proposed (for example, Patent Documents 1 and 2).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] In order for a salesperson to obtain a customer's contract, it is essential to obtain the customer's understanding and acceptance of the product and to propose a contract after the negotiation has matured to a certain extent, in addition to selecting the product. However, in the way of conducting a negotiation such as selecting a product that meets the needs of a customer and building a trust relationship with the customer to obtain the customer's understanding and acceptance of the product, personal negotiation skills are required. Such negotiation skills depend on the experience level of the salesperson, so there are variations among salespersons. Therefore, in order to increase the conclusion rate of the entire insurance company, it is required to efficiently improve the negotiation skills of a plurality of salespersons to a certain level or higher.

[0005] In light of the above, the following disclosure concerns an information processing device, etc., that enables the efficient improvement of sales representatives' negotiation skills. [Means for solving the problem]

[0006] The information processing device in this disclosure includes: an acquisition unit that acquires information about a conversation with a customer regarding a business negotiation to encourage the customer to purchase a product; a storage unit that stores the customer's attributes extracted from the conversation information by a large-scale language model; and a control unit that uses a first model that has learned the correspondence between the conversation information and the progress stage of the business negotiation, and a second model that has learned the correspondence between the combination of the attributes and the progress stage and the purchased product, to select a product to encourage a new customer to purchase based on information about a conversation with a new customer.

[0007] The system in this disclosure is a system comprising a terminal device and an information processing device that communicates with the terminal device, wherein the terminal device sends information of a conversation with the customer regarding a business negotiation to encourage the customer to purchase a product to the information processing device, the information processing device extracts the customer's attributes from the conversation information using a large-scale language model, and uses a first model that has learned the correspondence between the conversation information and the progress stage of the business negotiation, and a second model that has learned the correspondence between the combination of the attributes and the progress stage and the purchased product, to select a product to encourage the new customer to purchase based on the information of the conversation with the new customer.

[0008] The method of operating the system in this disclosure is a method of operating a system having a terminal device and an information processing device that communicates with the terminal device, wherein the terminal device sends information of a conversation with a customer regarding a business negotiation to encourage the customer to purchase a product to the information processing device, the information processing device extracts the customer's attributes from the conversation information using a large-scale language model, and uses a first model that has learned the correspondence between the conversation information and the progress stage of the business negotiation, and a second model that has learned the correspondence between the combination of the attributes and the progress stage and the purchased product, to select a product to encourage the new customer to purchase based on the information of the conversation with the new customer. [Effects of the Invention]

[0009] The information processing device described in this disclosure makes it possible to efficiently improve the negotiation skills of sales representatives. [Brief explanation of the drawing]

[0010] [Figure 1] This is a diagram showing an example of the configuration of an information processing system. [Figure 2] This figure shows an example of a server device configuration. [Figure 3] This is a diagram showing an example of the configuration of a terminal device. [Figure 4] This figure shows an example of the configuration of a functional module in a server device. [Figure 5] A flowchart illustrating an example of the operation procedure for server equipment, etc. [Figure 6] This figure shows an example of an output screen on a terminal device. [Figure 7] This figure shows an example of an output screen on a terminal device. [Modes for carrying out the invention]

[0011] Embodiments of the present invention will be described below.

[0012] Figure 1 shows an example configuration of one embodiment. The information processing system 1 includes a server device 10 and terminal devices 12 that are connected to the server device 10 via a network 11 so as to be able to communicate information. The server device 10 is, for example, one or a plurality of server computers that can communicate information with each other. The server device 10 is appropriately installed, for example, in one or more sales offices of an insurance company, one or more data centers, or a combination thereof. The server device 10 may also include one or more server computers that provide cloud services. The server device 10 corresponds to the "information processing device" in this embodiment. The terminal devices 12 are, for example, one or a plurality of personal computers that can communicate information with each other. The terminal devices 12 are appropriately installed, for example, in one or more sales offices and used by sales personnel. The personal computers may include portable tablet terminal devices, smartphones, etc. The network 11 is, for example, a LAN (Local Area Network) within a sales office, the Internet, an ad-hoc network, a MAN (Metropolitan Area Network), a mobile communication network, or another network, or a combination thereof.

[0013] The information processing system 1 in this embodiment supports, for example, sales representatives of insurance products in negotiations with customers. In negotiations with customers, the sales representative engages in various conversations, such as listening to the customer's insurance needs, searching for products that match those needs, explaining the products to the customer, and proposing the products to the customer and encouraging them to purchase the products, i.e., to enter into a contract, once the customer has gained an understanding and acceptance of the products.

[0014] In the information processing system 1, terminal device 12 sends information about conversations with customers regarding business negotiations aimed at encouraging customers to purchase products (hereinafter referred to as "conversation information") to server device 10, which is an information processing device. The server device 10 then extracts customer attributes (hereinafter referred to as "customer attributes") from the conversation information using a large-scale language model. The server device 10 then uses a first model (hereinafter referred to as the "progress determination model") that has learned the correspondence between conversation information and the progress stages of business negotiations, and a second model (hereinafter referred to as the "product selection model") that has learned the correspondence between combinations of customer attributes and progress stages and purchased products, to select a product (hereinafter referred to as the "proposed product") that encourages a new customer (hereinafter referred to as the "new customer") to purchase based on conversation information with that new customer.

[0015] During business negotiations, it is possible to estimate customer needs based on customer attributes. However, to ensure a smooth contract for insurance products, it is desirable to propose products only after the negotiation has reached a mature stage. By sending negotiation conversation information to the server device 10 via the terminal device 12, sales representatives can obtain information on proposed products from the server device 10 that are appropriate to the customer's attributes and the stage of the negotiation, regardless of their negotiation skills, and propose them to the customer. In other words, it becomes possible to propose products that match the customer's attributes to the customer at a stage of the negotiation where the probability of closing the deal is high. Therefore, the insurance company as a whole can efficiently improve the negotiation skills of multiple sales representatives, regardless of their individual experience levels.

[0016] Figure 2 shows an example configuration of the server device 10. The server device 10 includes a communication unit 101, a storage unit 102, and a control unit 103. If the server device 10 is composed of multiple server computers, the communication unit 101, the storage unit 102, and the control unit 103 may be appropriately distributed and arranged across multiple server computers.

[0017] The communication unit 101 includes one or more communication interfaces. The communication interface is, for example, a LAN interface. The communication unit 101 receives information used for the operation of the server device 10 and transmits information obtained by the operation of the server device 10. The server device 10 is connected to the network 11 by the communication unit 101 and communicates with the terminal device 12 via the network 11. The communication unit 101 corresponds to the "acquisition unit" that acquires conversation information from the terminal device 12 in the present embodiment.

[0018] The storage unit 102 includes, for example, one or more semiconductor memories that function as a main memory device, an auxiliary memory device, or a cache memory, one or more magnetic memories, one or more optical memories, or a combination of at least two of these. The semiconductor memory is, for example, a RAM (Random Access Memory) or a ROM (Read Only Memory). The RAM is, for example, a SRAM (Static RAM) or a DRAM (Dynamic RAM). The ROM is, for example, an EEPROM (Electrically Erasable Programmable ROM). The storage unit 102 stores a control / processing program necessary for the operation of the server device 10, various information required for the server device 10 to execute the control / processing program, and information obtained by the operation of the server device 10.

[0019] The control unit 103 includes one or more processors, one or more dedicated circuits, or a combination of these. The processor is, for example, a general-purpose processor such as a CPU (Central Processing Unit), or a dedicated processor such as a GPU (Graphics Processing Unit) specialized for specific processing. The dedicated circuit is, for example, an FPGA (Field-Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), etc. The control unit 103 executes information processing related to the operation of the server device 10 while controlling each part of the server device 10.

[0020] In this embodiment, the storage unit 102 stores insurance product information 105. The insurance product information 105 includes information such as the product name, coverage type, coverage details, and the correspondence between premiums and coverage amounts for various insurance products. The coverage type is an arbitrary classification according to the type of coverage emphasized for each insurance product, for example, a survival coverage-focused type, a death coverage-focused type, a medical coverage-focused type, etc.

[0021] Furthermore, the memory unit 102 stores the large-scale language model 109. If the server device 10 is composed of multiple server computers, the large-scale language model 109 may be stored in any server computer, for example, a server computer that provides cloud services. The large-scale language model 109 is a model that has been machine-learned using a large amount of text data to perform natural language processing, and is a language model that can perform natural language processing such as information extraction, text summarization, and question and answer in a general manner. The large-scale language model 109, for example, RAG (Retrieval-Augmented Generation), can acquire relevant information from an external knowledge base to improve the accuracy of natural language processing.

[0022] Furthermore, the memory unit 102 stores customer attribute information 106. The customer attribute information 106 has customer attribute information linked to the identification information of each customer. The control unit 103 extracts customer attributes from conversation information sent from the terminal device 12, links the customer attribute information to the identification information of each customer, and stores it in the memory unit 102 as customer attribute information 106. The conversation information includes, for example, audio data of a conversation between a salesperson and a customer acquired by the terminal device 12. The control unit 103 generates conversation text (hereinafter referred to as conversation text) from the conversation audio data by speech recognition processing. Alternatively, if speech recognition processing is performed on the terminal device 12, the conversation information may also include conversation text generated from the conversation audio data. The control unit 103 uses a large-scale language model 109 to identify the speaker from the context of the conversation text and extracts customer attributes from the text corresponding to the customer's utterance (hereinafter referred to as utterance text). The customer attributes include information such as the customer's age, gender, occupation, annual income, and family structure.

[0023] Furthermore, the memory unit 102 stores the progress determination model 107. The progress determination model 107 is a model for deriving progress stages based on conversation information, constructed by performing machine learning using training data in which the progress stages of negotiations are associated with conversation information from past negotiations between one or more sales representatives and one or more customers. Here, the conversation information may be audio data of the conversation or may include conversation text. The progress stages are represented by scores that divide the progress from the first negotiation to the negotiation at the time of product purchase into any number of stages, for example, when multiple negotiations are conducted with a customer. The progress stages are, for example, a 5-level score where the first negotiation is "1" and the negotiation at the time of product purchase is "5".

[0024] Furthermore, the memory unit 102 stores the product selection model 108. The product selection model 108 is a model for selecting product information of products that are highly likely to be purchased based on combinations of customer attributes and progress stages. It is constructed by performing machine learning using training data that associates combinations of customer attributes and progress stages when one or more customers have purchased products in the past with the purchased product information. The product information is selected from the information contained in the insurance product information 105.

[0025] Figure 3 shows an example of the configuration of the terminal device 12. The terminal device 12 includes a communication unit 121, a storage unit 122, a control unit 123, and an input / output unit 125.

[0026] The communication unit 121 includes a communication module compatible with wired or wireless LAN standards, a module compatible with mobile communication standards such as 4G (4th Generation) and 5G (5th Generation), etc. The terminal device 12 is connected to the network 11 via the communication unit 121 and communicates information with the server device 10 via the network 11.

[0027] The storage unit 122 includes one or more semiconductor memories, one or more magnetic memories, one or more optical memories, or a combination of at least two of these. The semiconductor memory is, for example, RAM or ROM. The RAM is, for example, SRAM or DRAM. The ROM is, for example, EEPROM. The storage unit 122 functions, for example, as a main memory, auxiliary memory, or cache memory. The storage unit 122 stores processing and control programs necessary for the operation of the terminal device 12, various information necessary for the control unit 123 to execute the processing and control programs, and information obtained by the operation of the control unit 123.

[0028] The control unit 123 has, for example, one or more general-purpose processors such as a CPU or MPU (Micro Processing Unit), or one or more dedicated processors specialized for a specific process. Alternatively, the control unit 123 may have one or more dedicated circuits such as FPGAs or ASICs. The control unit 123 comprehensively controls the operation of the terminal device 12 by operating according to a control and processing program, or by operating according to an operating procedure implemented as a circuit. The control unit 123 then sends and receives various information with the server device 10 via the communication unit 121 and executes the operations according to this embodiment.

[0029] The input / output unit 125 has an input interface that detects user input and sends the input information to the control unit 123. Such an input interface is any input interface, including, for example, physical keys, capacitive keys, a touchscreen integrated with a panel display, various pointing devices, or a microphone that accepts voice input. The input / output unit 120 also has an output interface that outputs information generated by the control unit 123 or acquired from the server device 10 to the user. Such an output interface is any output interface, including, for example, a connection interface to an external output device such as an external or built-in display that outputs information as images or videos, or a speaker or printer that outputs information as sound.

[0030] Figure 4 shows an example of the configuration of a functional module in the server device 10. The server device 10 has a service layer 21, a control layer 22, a skill layer 23, a batch layer 24, and a data layer 25 as functional modules executed by the control unit 103 using information stored in the storage unit 102. These functional modules may be appropriately distributed among multiple server computers when the server device 10 is composed of multiple server computers.

[0031] The service layer 21 provides a user interface that receives various information 201 from the terminal device 12 operated by the sales representative and sends various information 202 generated by information processing in the server device 10 to the terminal device 12. The various information 201 includes various requests and conversation information. The user interface includes, for example, an interactive virtual agent. The service layer 21 determines subsequent processing in response to various requests from the terminal device 12, and sends instructions 203 and 209 corresponding to the determined subsequent processing to the control layer 22 and batch layer 24, respectively, and receives responses 204 and 210 from the control layer 22 and batch layer 24, respectively. The service layer 21 sends output information to the terminal device 12 in response to responses 204 and 210 from the control layer 22 and batch layer 24, respectively.

[0032] The batch layer 24 responds to instructions 209 from the service layer 21 by receiving conversation information from the service layer 21, performing speech recognition processing on the audio data contained in the conversation information to generate conversation text, or sending instructions 207 to the large-scale language model 109 to perform natural language processing on the conversation text. Instructions 207 include, for example, prompts to create a conversation summary, meeting minutes, etc., based on the conversation text, prompts to identify customer utterances from the context of the conversation text, and prompts to extract customer attributes from customer utterances. The batch layer 24 receives the results 208 of the natural language processing performed by the large-scale language model 109 in response to instructions 207 from the large-scale language model 109 and sends them to the service layer 21 as a response 204. The batch layer 24 is tuned by terms such as insurance products contained in the insurance product information 105 in order to create summaries, meeting minutes, etc. The batch layer 24 also adds the customer attributes 213 extracted from the conversation to the customer attribute information 106 in the data layer 25. The batch layer 24 may also perform the above operations triggered by receiving information via the service layer 21 indicating that the conversation from the terminal device 12 has ended.

[0033] The control layer 22 responds to the instruction 203 from the service layer 21 by sending an instruction 205 to the skill layer 23 to perform the necessary information processing. In the skill layer 23, various information processing tasks are classified as skills accompanied by numerical vectors, and the control layer 22 can, for example, select the skill with the numerical vector that most closely matches instruction 203 by vector search and send instruction 205. Instruction 205 includes, for example, an instruction to derive the progress stage from conversation information to the progress determination model 107, and an instruction to make the product selection model 108 from the insurance product information 105 in the data layer 25 based on the combination of the progress stage and customer attributes 211 obtained from customer attribute information 106. In the skill layer 23, various information processing is performed according to instruction 203. The control layer 22 receives the results 206 or 212 of the various information processing performed in the skill layer 23 from the skill layer 23 or the data layer 25, respectively, and sends them to the service layer 21 as a response 204.

[0034] Various information processing tasks in skill layer 23 are numerically vectorized by multimodal AI (Artificial Intelligence) and subjected to vector search by control layer 22. If multiple tasks are found by the vector search, control layer 22 generates a selection of multiple tasks using a large-scale language model 109, and ambiguity can be avoided by allowing the user to select an option via service layer 21.

[0035] Figure 5 is a flowchart illustrating an example of server device operation. The procedure example in Figure 5 is an example of a procedure executed by the control unit 103 of the server device 10 in response to a request from the terminal device 12 when a sales representative conducts a business negotiation using the terminal device 12.

[0036] The procedure shown in Figure 5 is performed, for example, when a salesperson initiates a sales negotiation. When the salesperson requests a sales negotiation support service by operating the input / output unit 125 of the terminal device 12, the control unit 123 of the terminal device 12 sends the service request to the server device 10 via the communication unit 121. Then, in step S500, the control unit 103 of the server device 10 receives the service request.

[0037] In step S501, the control unit 103 of the server device 10 activates the service layer 21 for sales negotiation support in response to a service request. The service layer 21, for example, generates a virtual agent for interaction to provide a user interface and sends information to the terminal device 12 for displaying the virtual agent. The terminal device 12 then displays the virtual agent 60 on the display of the input / output unit 125, for example, as shown in the example in Figure 6, prompting the salesperson to start the sales negotiation. In the example in Figure 6, in addition to the virtual agent 60, text 61 prompting the start of the sales negotiation is displayed. The information in text 61 may also be output as audio. Once the sales negotiation has started, the terminal device 12 acquires the audio of the conversation using the microphone of the input / output unit 125.

[0038] In step S502, the control unit 103 of the server device 10 acquires conversation information. Specifically, the service layer 21 acquires conversation information sent from the terminal device 12. When a salesperson conducts a business negotiation with a customer, the terminal device 12 acquires the audio of the conversation, and conversation information, including the audio data of the conversation, is sent from the terminal device 12 to the server device 10.

[0039] In step S503, the control unit 103 of the server device 10 generates conversation text. Specifically, the batch layer 24, triggered by receiving information from the terminal device 12 indicating the end of the conversation, performs speech recognition processing on the audio data included in the conversation information to generate conversation text.

[0040] In step S504, the control unit 103 of the server device 10 performs speaker analysis. Specifically, the batch layer 24 distinguishes between the utterances of the salesperson and the customer based on the context of the conversation text and identifies the customer's utterance text.

[0041] In step S505, the control unit 103 of the server device 10 creates meeting minutes or a summary. Specifically, the batch layer 24 uses the large-scale language model 109 to create meeting minutes and a summary of the business negotiation. The information of the meeting minutes and summary is stored in the storage unit 102 and sent to the terminal device 12 as needed in response to requests from the terminal device 12.

[0042] In step S506, the control unit 103 of the server device 10 extracts customer attributes from the customer's utterance text. Specifically, the batch layer 24 extracts customer attributes using the large-scale language model 109. For example, the utterance content containing the customer's age, gender, occupation, annual income, family structure, etc., is analyzed and customer attributes are extracted. The batch layer 24 adds the extracted customer attribute information to the customer attribute information 106.

[0043] In step S507, the control unit 103 of the server device 10 derives the progress stage of the business negotiation based on the conversation information. Specifically, the control layer 22 activates the progress determination model 107 of the skill layer 23, causing the progress determination model 107 to derive the progress stage based on the conversation information. The progress determination model 107 derives, for example, the progress stage corresponding to the interaction between the salesperson and the customer contained in the conversation text. The derived progress stage information is stored in the storage unit 102.

[0044] Furthermore, during the progress of negotiations, the sales representative can, as appropriate, operate the terminal device 12 to view and search customer attribute information 106, insurance product information 105, etc., on the server device 10, present various insurance products to the customer as reference information, and explain the products. By continuing negotiations in this manner, the progress of the negotiations will advance.

[0045] In step S508, the control unit 103 of the server device 10 selects a proposed product based on a combination of customer attributes and progress stage. Specifically, the control layer 22 activates the product selection model 108 of the skill layer 23, causing the product selection model 108 to select a proposed product from the insurance product information 108 that corresponds to the combination of customer attributes and progress stage. Customer attributes include one or more of the following: age, gender, occupation, annual income, family structure, etc. If the progress stage is the final stage of negotiation, that is, the stage immediately before signing a contract, the product selection model 108 selects a proposed product that matches the customer attributes from the insurance product information 105 and outputs information about the proposed product. If the progress stage is in the middle of negotiation, that is, premature, there is a high probability that no proposed product will be selected or output.

[0046] If a proposed product is selected in step S508 (Yes in step S509), the control unit 103 of the server device 10 proceeds to step S510. On the other hand, if no proposed product is selected (No in step S509), the control unit 103 of the server device 10 skips step S510 and terminates the procedure shown in Figure 5.

[0047] In step S510, the control unit 103 of the server device 10 outputs information about the proposed product. The control layer 22 in the control unit 103 sends the information about the proposed product obtained from the skill layer 23 to the service layer 21, and the service layer 21 sends the information about the proposed product to the terminal device 12. The terminal device 12 displays the virtual agent 60 on the output display of the input / output unit 125, for example as in the example in Figure 7, and presents the information about the proposed product 70 to the salesperson. The information about the proposed product 70 includes information about the proposed product that the salesperson can propose to the customer. The information about the proposed product 70 may also be output by voice.

[0048] According to the procedure in Figure 5, even if a salesperson has relatively low experience, by sending the conversation information of the sales negotiation to the server device 10 via the terminal device 12, the progress stage of the sales negotiation can be accurately determined by the server device 10. Furthermore, based on the conversation information, a suitable product to propose that matches the customer's attributes can be appropriately selected. Then, it becomes possible to present the proposed product to the customer when the probability of securing a contract has increased. Therefore, regardless of the experience level of each salesperson, the insurance company as a whole can efficiently improve the negotiation skills of multiple salespeople. Alternatively, salespeople can be trained within the insurance company by having them conduct simulated sales negotiations using the information processing system 1. Alternatively, by obtaining and displaying minutes and summaries of past sales negotiations from the server device 10 at each progress stage via the terminal device 12, it can be used as a reference for coaching provided by the salesperson's supervisor to the salesperson.

[0049] In a modified version of this embodiment, for example, the skill layer 23 may have a task to generate personalized messages for each customer using customer attributes. This task, for example, generates prompts for the large-scale language model 109 to create thank-you notes, birthday messages, etc., and sends the generated prompts to the large-scale language model 109, making it possible to create more sophisticated personalized messages than messages using simple templates. Sales representatives can then retrieve these personalized messages using the terminal device 12 and send them to customers to follow up on the progress of the sales negotiation.

[0050] In the above-described embodiment, the processing and control program that defines the operation of the terminal device 12 may be stored in the storage unit 122 of the server device 10 or in the storage unit of another server device and downloaded via the network 11, or it may be stored in a computer-readable non-transient recording and storage medium and read by the terminal device 12 from the medium.

[0051] As described above, embodiments have been explained based on various drawings and examples, but it should be noted that those skilled in the art will find it easy to make various modifications and alterations based on this disclosure. Therefore, it should be noted that these modifications and alterations are within the scope of this disclosure. For example, the functions, etc., included in each means, each step, etc., can be rearranged in a logically consistent manner, and multiple means, steps, etc., can be combined into one or divided. [Explanation of Symbols]

[0052] 10: Server equipment 11: Network 12: Terminal device 21: Service Layer 22: Control Layer 23: Skill Level 24: Batch layer 101, 121: Communications Department 102, 122: Storage section 103, 123: Control Unit 105: Insurance Product Information 106:Customer attribute information 107: Progress Determination Model 108: Product Selection Model 109: Large-scale language models 125: Input / output section

Claims

1. An acquisition unit that acquires information about conversations with customers regarding business negotiations aimed at encouraging them to purchase products, A memory unit that stores the customer's attributes extracted from the conversation information by a large-scale language model, A control unit that uses a first model that has learned the correspondence between the information from the conversation and the progress stage of the business negotiation, and a second model that has learned the correspondence between the combination of the attribute and the progress stage and the purchased product, to select a product that encourages a new customer to purchase based on the information from the conversation with the new customer, An information processing device having

2. In claim 1, The information of the aforementioned conversation includes audio data or text of the aforementioned conversation. Information processing device.

3. In claim 1, The first model derives the progress stage based on the information from the conversation with the new customer. Information processing device.

4. In claim 3, The second model selects products to encourage purchases from the new customer based on a combination of the attributes of the new customer extracted by the large-scale language model from the conversation information with the new customer and the progress stage derived based on the conversation information with the new customer. Information processing device.

5. A system comprising a terminal device and an information processing device that communicates with the terminal device, The terminal device sends information about the conversation with the customer regarding a business negotiation to encourage the customer to purchase a product to the information processing device. The information processing device extracts the customer's attributes from the conversation information using a large-scale language model, and uses a first model that has learned the correspondence between the conversation information and the progress stage of the business negotiation, and a second model that has learned the correspondence between the combination of the attributes and the progress stage and the purchased products, to select products that encourage the new customer to purchase based on the conversation information with the new customer. system.

6. In claim 5, The information of the aforementioned conversation includes audio data or text of the aforementioned conversation. system.

7. In claim 5, The first model derives the progress stage based on the information from the conversation with the new customer. system.

8. In claim 7, The second model selects products to encourage purchases from the new customer based on a combination of the attributes of the new customer extracted by the large-scale language model from the conversation information with the new customer and the progress stage derived based on the conversation information with the new customer. system.

9. A method for operating a system having a terminal device and an information processing device that communicates with the terminal device, The terminal device sends information about the conversation with the customer regarding a business negotiation to encourage the customer to purchase a product to the information processing device. The information processing device extracts the customer's attributes from the conversation information using a large-scale language model, and uses a first model that has learned the correspondence between the conversation information and the progress stage of the business negotiation, and a second model that has learned the correspondence between the combination of the attributes and the progress stage and the purchased products, to select products that encourage the new customer to purchase based on the conversation information with the new customer. How the system works.

10. In claim 9, The information of the aforementioned conversation includes audio data or text of the aforementioned conversation. How the system works.

11. In claim 9, The first model derives the progress stage based on the information from the conversation with the new customer. How the system works.

12. In claim 11, The second model selects products to encourage purchases from the new customer based on a combination of the attributes of the new customer extracted by the large-scale language model from the conversation information with the new customer and the progress stage derived based on the conversation information with the new customer. How the system works.

Images