system

Abstract

To provide a system.SOLUTION: A system includes: first means for acquiring the past approval history and the tendency of an approver from a database storing the past information; second means for generating a first prompt sentence for outputting a draft wording that is highly likely to be approved by the approver on the basis of the acquired past approval history and the acquired tendency of the approver; third means for generating a draft wording by entering the generated first prompt text into a generative AI model; and fourth means for displaying the generated draft wording on a terminal of a business operator.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The technology of this disclosure relates to a system.

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance that responds to the user utterance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Conventionally, the in-house decision-making draft prepared by corporate sales on a daily basis has problems that the load on salespersons and SCs is large, and it is difficult for each approver to create a document draft that they can accept.

Means for Solving the Problems

[0005] To solve this problem, information is obtained from a database storing past information, and an AI that automatically generates a document draft for proposal and contract parts using machine learning is used. Furthermore, by presenting the generated document draft in an interactive manner, the load on salespersons is reduced, and a document draft that each approver can accept is created.

Brief Description of the Drawings

[0006] [Figure 1]This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Embodiment 1 of Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1 of Form Example 1. [Figure 13] This is a sequence diagram showing the processing flow of the data processing system in Example 2 of Embodiment 2. [Figure 14] This is a sequence diagram showing the processing flow of the data processing system in Application Example 2 of Form Example 2. [Figure 15] This is a sequence diagram showing the processing flow of the data processing system in Embodiment 3 of Example 3. [Figure 16] This is a sequence diagram showing the processing flow of the data processing system in Application Example 3 of Form Example 3. [Figure 17] It is a sequence diagram showing the processing flow of the data processing system in Example 1 of Form Example 1 when combined with an emotion engine. [Figure 18] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1 of Form Example 1 when combined with an emotion engine. [Figure 19] It is a sequence diagram showing the processing flow of the data processing system in Example 2 of Form Example 2 when combined with an emotion engine. [Figure 20] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 of Form Example 2 when combined with an emotion engine. [Figure 21] It is a sequence diagram showing the processing flow of the data processing system in Example 3 of Form Example 3 when combined with an emotion engine. [Figure 22] It is a sequence diagram showing the processing flow of the data processing system in Application Example 3 of Form Example 3 when combined with an emotion engine.

Embodiments for Carrying Out the Invention

[0007] Hereinafter, an example of an embodiment of the system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0008] First, the language used in the following description will be explained.

[0009] In the following embodiments, the labeled processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)), etc.

[0010] In the following embodiments, the labeled RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0011] In the following embodiments, the labeled storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0012] In the following embodiments, the labeled communication I / F (Interface) is an interface that includes a communication processor and an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark), etc.

[0013] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0014] [First Embodiment]

[0015] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0016] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0017] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0018] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0019] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0020] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0021] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0022] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0023] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0024] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0025] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0026] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[0027] "Example of form 1"

[0028] In one embodiment of the present invention, there exists a database storing past information. This database includes, for example, information on contracts and proposals made by corporate sales representatives in the past. There exists an AI that retrieves information from this database and automatically generates draft text for proposals and contracts using machine learning. This AI is trained using machine learning techniques such as deep learning and reinforcement learning. The generated draft text is presented to the sales representative through an interactive interface. This interface is implemented, for example, in the form of a chatbot or a voice assistant. "Example of Form 2"

[0029] In another embodiment of the present invention, AI generates draft documents to reduce the workload of salespeople. Specifically, the AI ​​automates some of the tasks that salespeople normally perform. For example, it significantly reduces the time and effort salespeople spend creating draft proposals and contract documents.

[0030] "Example of form 3"

[0031] In a further embodiment of the present invention, the AI ​​generates draft text that is expected to be acceptable to the approver. Specifically, the AI ​​learns past approval history and approver tendencies, and generates draft text based on that. This makes it possible to provide draft text that is highly likely to be acceptable to the approver.

[0032] The following describes the processing flow for each example of the form.

[0033] "Example of form 1"

[0034] Step 1: Retrieve the necessary information from a database containing past data. This information could include, for example, past contracts and proposals.

[0035] Step 2: Based on the acquired information, machine learning is used to automatically generate draft text for the proposal and contract sections. This AI is trained using machine learning techniques such as deep learning and reinforcement learning.

[0036] Step 3: The generated draft text is presented to the salesperson through an interactive interface. This interface can be implemented, for example, as a chatbot or voice assistant.

[0037] "Example of form 2"

[0038] Step 1: The AI ​​generates draft text to reduce the workload of salespeople. Specifically, the AI ​​automates some of the tasks that salespeople normally perform.

[0039] Step 2: For example, significantly reduce the time and effort salespeople spend creating proposals and draft contract documents.

[0040] "Example of form 3"

[0041] Step 1: The AI ​​generates draft text that is expected to satisfy the approver. Specifically, the AI ​​learns from past approval history and the approver's tendencies.

[0042] Step 2: Generate a draft document based on that. This makes it possible to provide a draft document that is more likely to be accepted by the approver.

[0043] (Example 1)

[0044] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0045] In traditional corporate sales, a major problem was the significant time and effort required to create proposals and contracts. Furthermore, it was difficult for salespeople to efficiently utilize information from past proposals and contracts, sometimes resulting in a decline in the quality of proposals. Additionally, creating documents that would satisfy approvers required experience and skill, placing a significant burden on new salespeople.

[0046] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0047] In this invention, the server includes means for retrieving information from a database storing historical information, means for training a machine learning model using the retrieved information, means for automatically generating draft text for proposals and contracts using the trained machine learning model, and means for interactively presenting the generated draft text. This reduces the burden on salespeople and enables the efficient and high-quality creation of proposals and contracts. Furthermore, by automatically generating draft text that is acceptable to approvers, the approval process can also be expedited.

[0048] A "database" is a system that stores historical information and allows you to retrieve that information as needed.

[0049] "Means of acquiring information" refers to the methods and techniques for extracting necessary information from a database.

[0050] A "machine learning model" is an algorithm or framework that learns from data and automatically performs a specific task.

[0051] "Training methods" refer to the methods and techniques used to train machine learning models using data.

[0052] "Methods for automatically generating draft documents" refer to methods and technologies that use trained machine learning models to automatically create drafts of new proposals and contracts.

[0053] "Interactive presentation methods" refer to methods and technologies for interactively displaying generated draft text to the user.

[0054] "Salesperson" refers to the person or staff member who handles corporate sales.

[0055] An "approver" refers to a person who has the authority to review and approve the contents of a proposal or contract.

[0056] Modes for carrying out the invention

[0057] This invention is a system designed to streamline the creation of proposals and contracts in corporate sales, thereby reducing the burden on salespeople. A specific embodiment of this system is described below.

[0058] Retrieve information from the database.

[0059] The server retrieves the necessary information from a database containing information on past contracts and proposals made by corporate sales representatives. This database includes contract details, proposal details, dates, and customer information. The server extracts information from the database using SQL queries. For example, executing a query such as "SELECT FROM contracts WHERE date > '2020-01-01'" retrieves contract information from 2020 onwards.

[0060] Training machine learning models

[0061] The server uses the acquired information to train a machine learning model. This model employs techniques such as deep learning and reinforcement learning. Specifically, it uses the TENSORFLOW® framework. The server cleans and normalizes the data, and then trains the model using the preprocessed data. For example, it trains a text generation model using an LSTM (Long Short-Term Memory) network.

[0062] Automatic generation of draft text

[0063] The server uses a trained LSTM model to automatically generate draft proposals and contract documents. When a user enters a prompt such as "Generate a new proposal," the server inputs this prompt into the model and retrieves the generated draft document. For example, it might generate a document stating, "This proposal provides the best solution for your needs."

[0064] Presentation to the business operator

[0065] The generated draft text is presented to the salesperson through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant. The salesperson accesses the chatbot through their device and reviews the generated draft text. For example, if the salesperson types "Show me the draft text of the proposal" into the chatbot, the chatbot will display the text "This proposal provides the best solution for your company's needs."

[0066] Specific examples and prompt statements

[0067] Suppose a user needs to create a new proposal. The user accesses the system via their device and instructs it to generate draft text based on information from similar past proposals. The server retrieves past proposal information from the database and uses a machine learning model to generate a new draft text. The generated draft text is then presented to the user via a chatbot.

[0068] Example of a prompt:

[0069] "Please generate a draft of a new proposal document based on past corporate sales proposal data."

[0070] In this way, users can efficiently create new proposals.

[0071] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0072] Step 1:

[0073] The server retrieves information on past contracts and proposals from the database. SQL queries are used as input. For example, executing the query "SELECT FROM contracts WHERE date > '2020-01-01'" will output contract information from 2020 onwards. The server stores the retrieved data in memory.

[0074] Step 2:

[0075] The server preprocesses the acquired data. Raw data retrieved from the database is used as input. Preprocessing includes data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). This results in output data in a format suitable for training machine learning models.

[0076] Step 3:

[0077] The server trains a machine learning model using preprocessed data. Preprocessed data is used as input. Specifically, it trains an LSTM (Long Short-Term Memory) network using the TensorFlow framework. The training results in a model for generating proposal and contract texts.

[0078] Step 4:

[0079] The server uses a trained LSTM model to automatically generate drafts of new proposals and contract documents. The input is a prompt sentence entered by the user. For example, the prompt "Generate a new proposal" might be entered. The server inputs this prompt into the model and outputs a generated draft document. For example, it might generate a document stating, "This proposal provides the best solution for your needs."

[0080] Step 5:

[0081] The server presents the generated draft text to the salesperson through an interactive interface. The generated draft text is used as input. The interactive interface is implemented in the form of a chatbot or voice assistant. For example, if the salesperson types "Show me the draft text of the proposal" into the chatbot, the chatbot will display the text "This proposal provides the best solution for your company's needs."

[0082] (Application Example 1)

[0083] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0084] In traditional sales activities, sales representatives must manually refer to past contract and proposal information to create new proposals and contracts, which is time-consuming and labor-intensive. Furthermore, maintaining a consistent level of quality is difficult because the quality of proposals and contracts depends on the sales representative's experience and skills. There is a growing need for a way for sales representatives to efficiently create proposals even when they are on the go.

[0085] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0086] This invention includes a server that retrieves information from a database storing historical data and automatically generates draft proposal and contract documents using machine learning; a server that interactively presents the generated draft documents; and a server that presents the generated draft documents to sales representatives via smartphone. This enables sales representatives to quickly generate high-quality proposal and contract documents based on historical data and to conduct sales activities efficiently even when on the go.

[0087] A "database containing past information" is a data storage system that stores information on past contracts and proposals made by corporate sales representatives, allowing them to search and retrieve the information as needed.

[0088] "Machine learning" is a technology in which computers learn patterns and rules from data and use that knowledge to make predictions and decisions about new data.

[0089] "Artificial intelligence that automatically generates draft text for proposals and contracts" refers to a program that uses machine learning to analyze past data and automatically generate new proposal and contract documents.

[0090] "Interactive presentation methods" refer to interfaces that interactively display generated draft text to the user, allowing the user to review and modify its content.

[0091] The "means of presenting information to sales representatives via smartphone" refers to a function that displays generated draft text on the sales representative's smartphone, enabling them to conduct sales activities efficiently even when they are out of the office.

[0092] A system for implementing this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, and means for presenting the generated draft text to sales representatives via smartphone.

[0093] System Configuration

[0094] 1. Database: The server has a database that stores information on past contracts and proposals made by corporate sales representatives. This database is built using a publicly known database management system.

[0095] 2. Artificial Intelligence: The server includes artificial intelligence that uses machine learning (e.g., OpenAI®'s GPT-3®) to automatically generate new proposals and contracts based on past data. This artificial intelligence is implemented using a publicly known programming language.

[0096] 3. Interactive Interface: The generated draft text is presented to the sales representative through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant.

[0097] 4. Smartphone Application: The generated draft text will be displayed on the sales representative's smartphone. The smartphone application runs on the ANDROID® platform, enabling sales representatives to conduct sales activities efficiently even when they are on the go.

[0098] Processing flow

[0099] 1. Data Retrieval: The server retrieves past proposals and contracts for a specific client from the database.

[0100] 2. Prompt generation: The server generates prompts to input into the artificial intelligence based on previously retrieved suggestion sentences.

[0101] 3. Document draft generation: The artificial intelligence generates new proposals and contracts based on the generated prompts.

[0102] 4. Presentation of draft text: The generated draft text is presented to the sales representative through an interactive interface.

[0103] 5. Smartphone display: The generated draft text is displayed on the sales representative's smartphone, allowing them to conduct sales activities efficiently even when they are out of the office.

[0104] Hardware and software to be used

[0105] Hardware: Servers, smartphones

[0106] software

[0107] Specific example

[0108] For example, if a past proposal from client ID 123 was "This product is high-quality and offers excellent value for money," the newly generated proposal might be "This product uses the latest technology and is highly reliable."

[0109] Example of a prompt

[0110] Previous proposal: This product is high quality and offers excellent value for money. Please generate a new proposal:

[0111] In this way, sales representatives can easily generate new proposals using their smartphones and make effective proposals to clients.

[0112] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0113] Step 1:

[0114] The server retrieves past proposals and contracts for a specific client from the database. The input is the client ID, and the output is a list of past proposals and contracts. Specifically, the server executes an SQL query to extract proposals related to the corresponding client ID from the database.

[0115] Step 2:

[0116] The server generates prompts for the artificial intelligence based on the previously retrieved proposals. The input is a list of past proposals, and the output is a prompt. Specifically, the server combines the past proposals into a single text and creates a prompt with the preface "Past proposals: ".

[0117] Step 3:

[0118] The server sends the generated prompt text to the artificial intelligence, which then generates new proposal or contract texts. The input is the prompt text, and the output is the generated proposal text. Specifically, the server calls the OpenAI GPT-3 API, passing the prompt text as input and receiving the generated text.

[0119] Step 4:

[0120] The server presents the generated proposal to the sales representative through an interactive interface. The input is the generated proposal, and the output is an interface that the sales representative can review and modify. Specifically, the server displays the generated proposal in a chatbot format, allowing the sales representative to review and modify its contents.

[0121] Step 5:

[0122] The server displays the generated proposal on the sales representative's smartphone. The input is the generated proposal, and the output is the proposal displayed on the smartphone screen. Specifically, the server sends the generated proposal to the smartphone application, allowing the sales representative to review and use the proposal even when they are away from the office.

[0123] (Example 2)

[0124] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0125] There is a need to reduce the time and effort salespeople spend creating proposals and contract drafts, and to proceed with their work more efficiently. It is also crucial to maintain a consistent quality of draft documents and to quickly generate content that approves. A system is needed to address these challenges.

[0126] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0127] In this invention, the server includes a generation AI model that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, means for generating prompt texts based on information input by the user, and means for interactively presenting the generated draft texts. This reduces the burden on sales personnel and makes it possible to efficiently generate high-quality draft texts.

[0128] A "database" is a system that stores historical information and retrieves it as needed.

[0129] "Machine learning" is a technology that allows computers to learn patterns from data and perform predictions and classifications.

[0130] A "generative AI model" is an artificial intelligence model that uses machine learning to automatically generate draft text for proposals and contracts.

[0131] A "prompt statement" is a sentence used to give instructions to an AI model based on information entered by the user.

[0132] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing for confirmation and modification.

[0133] A "salesperson" is a person responsible for creating proposals and drafting contract documents.

[0134] A "draft document" is a draft of a document related to a proposal or contract.

[0135] An "approver" is a person who has the authority to review and approve the generated draft document.

[0136] This invention is a system that reduces the burden on salespeople and efficiently generates high-quality proposals and contract drafts. This system operates through the cooperation of three parties: a server, a terminal, and a user.

[0137] First, the server maintains a database containing past information and retrieves it as needed. The database stores information about past proposals and contracts. The server uses a generative AI model (for example, OpenAI's GPT-4®) that uses machine learning to automatically generate draft text for proposals and contracts.

[0138] Users input basic information about proposals and contracts via the terminal. Specifically, they input information such as the customer name, proposal details, and contract terms (e.g., initial fees, monthly fees). For example, when a user creates a proposal for a new customer, they would input the following information into the terminal:

[0139] Customer name: XYZ Co., Ltd.

[0140] Proposal: Proposal for a new marketing strategy

[0141] Contract terms: Initial fee 1 million yen, monthly fee 100,000 yen

[0142] The server generates prompt messages based on the information entered by the user. These prompt messages include instructions for the generating AI model. For example, it generates prompt messages like the following:

[0143] "Please prepare a proposal for a new marketing strategy for XYZ Corporation. The proposal should include contract terms of an initial fee of 1 million yen and a monthly fee of 100,000 yen."

[0144] The generative AI model receives a prompt and generates draft proposals and contract documents. The goal is to produce draft documents of the same quality as those manually created by sales representatives. For example, it can generate draft documents like the following:

[0145] "XYZ Corporation"

[0146] Thank you for reviewing our proposal for a new marketing strategy. The proposal details and contract terms are outlined below.

[0147] Proposal details:

[0148] Introducing a new marketing strategy

[0149] Initial cost: 1 million yen

[0150] Monthly fee: 100,000 yen

[0151] Thank you for your consideration.

[0152] Sincerely"

[0153] The generated draft text is sent from the server to the terminal and presented to the user interactively. The user can review the displayed draft text and make revisions as needed. This reduces the workload on sales staff and enables the efficient generation of high-quality draft text.

[0154] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0155] Step 1:

[0156] Users input basic information about proposals and contracts via the terminal. Specifically, they input information such as the customer name, proposal details, and contract terms (e.g., initial fees, monthly fees). For example, when a user creates a proposal for a new customer, they would input the following information into the terminal:

[0157] Customer name: XYZ Co., Ltd.

[0158] Proposal: Proposal for a new marketing strategy

[0159] Contract terms: Initial fee 1 million yen, monthly fee 100,000 yen

[0160] Input: Customer name, proposal details, contract terms

[0161] Output: Basic information entered by the user

[0162] Step 2:

[0163] The server generates prompt messages based on the information entered by the user. These prompt messages include instructions for the AI ​​model that generates them. The server analyzes the input information and creates appropriate prompt messages. For example, it might generate a prompt message such as, "Please create a proposal for a new marketing strategy for XYZ Corporation. The proposal should include contract terms of an initial fee of 1 million yen and a monthly fee of 100,000 yen."

[0164] Input: Basic information entered by the user

[0165] Output: Generated prompt message

[0166] Step 3:

[0167] The server sends a prompt to a generative AI model (e.g., OpenAI's GPT-4) and generates a draft document. The generative AI model then creates a draft proposal or contract document based on the prompt. For example, it might generate a draft document like this:

[0168] "XYZ Corporation"

[0169] Thank you for reviewing our proposal for a new marketing strategy. The proposal details and contract terms are outlined below.

[0170] Proposal details:

[0171] Introducing a new marketing strategy

[0172] Initial cost: 1 million yen

[0173] Monthly fee: 100,000 yen

[0174] Thank you for your consideration.

[0175] Sincerely"

[0176] Input: Generated prompt message

[0177] Output: Generated draft text

[0178] Step 4:

[0179] The server sends the generated draft text to the terminal and presents it to the user interactively. The user can review the displayed draft text and make corrections as needed. The user checks the draft text on the terminal screen, verifies that there are no errors, and makes corrections as necessary.

[0180] Input: Generated draft text

[0181] Output: Draft text presented to the user

[0182] In this way, the server, terminals, and users work together, reducing the workload on sales personnel and enabling the efficient generation of high-quality draft documents.

[0183] (Application Example 2)

[0184] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0185] A problem is that salespeople spend a great deal of time and effort creating proposals and contracts. Furthermore, communication with customers and managing the progress of sales activities are also time-consuming, making efficient sales activities difficult. There is a need to solve these problems, reduce the burden on salespeople, and achieve more efficient sales activities.

[0186] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0187] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract documents using machine learning, a means for interactively presenting the generated draft documents, a means for automating communication with customers, a means for automatically generating proposals and contracts, a means for managing and analyzing customer information, and a means for managing the progress of sales activities. This significantly reduces the workload on salespeople and enables efficient sales activities.

[0188] "Past information" refers to past data and records related to sales activities.

[0189] A "database" refers to an information management system that systematically stores information and allows it to be searched and retrieved as needed.

[0190] "Machine learning" refers to the technology that allows computers to learn patterns from data and perform predictions and classifications.

[0191] "Draft text for proposals and contracts" refers to draft proposals and contracts that are presented to customers during sales activities.

[0192] "Artificial intelligence" refers to computer systems that mimic human intelligence and perform learning and reasoning.

[0193] "Interactive presentation methods" refer to methods in which users and systems exchange information in a dialogue format.

[0194] "Methods for automating communication with customers" refers to methods for automating communication with customers, such as email and chat.

[0195] "Methods for automatically generating proposals and contracts" refers to methods for automatically creating proposals and contracts necessary for sales activities.

[0196] "Means for managing and analyzing customer information" refers to methods for organizing and analyzing data related to customers.

[0197] "Means of managing the progress of sales activities" refers to methods of tracking and managing the progress of the sales process.

[0198] The system for implementing this invention is configured as follows: The server retrieves information from a database containing historical data and automatically generates draft proposal and contract documents using machine learning. The generated draft documents are presented to the user interactively. It also has the functionality to automate communication with customers and automatically generate proposals and contracts. Furthermore, it includes functions to manage and analyze customer information and manage the progress of sales activities.

[0199] Hardware and software to be used

[0200] Hardware: Servers, user terminals (smartphones, personal computers, etc.)

[0201] Software: APIs, database management systems (e.g., MySQL®)

[0202] Data processing and data calculation

[0203] 1. Retrieving information from the database: The server retrieves data related to past sales activities from the database. This data includes past proposals, contracts, and communication history with customers.

[0204] 2. Text draft generation using machine learning: Based on the acquired data, a machine learning model (e.g., OpenAI's GPT-3) is used to generate draft text for proposals and contracts.

[0205] 3. Interactive Presentation: The generated draft text is presented interactively on the user's terminal. The user can review the draft text and make modifications as needed.

[0206] 4. Automated communication with customers: The server automates email and chat communication with customers based on the generated draft messages.

[0207] 5. Automatic generation of proposals and contracts: The server automatically generates formal proposals and contracts based on the generated draft documents.

[0208] 6. Customer Information Management and Analysis: The server organizes and analyzes customer information to improve the efficiency of sales activities.

[0209] 7. Sales activity progress management: The server tracks and manages the progress of the sales process.

[0210] Specific example

[0211] For example, if the customer name is "ABC Corporation" and the product details are "latest AI-equipped smartphone," the following prompt message will be sent to the AI ​​model.

[0212] Example of a prompt

[0213] Customer Name: ABC Co., Ltd.

[0214] Product Details: Latest AI-powered smartphone

[0215] Please prepare your proposal based on the following information.

[0216] Based on this prompt, the AI ​​model generates the text of the proposal, reducing the workload on salespeople. The generated proposal is presented interactively on the user's terminal, allowing the user to review the content and make revisions as needed.

[0217] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0218] Step 1:

[0219] The server retrieves data on past sales activities from a database. Inputs include past proposals, contracts, and customer communication history stored in the database. Output is a collection of the retrieved data, which is used for subsequent processing.

[0220] Step 2:

[0221] The server uses a machine learning model (e.g., OpenAI's GPT-3) to generate draft proposal and contract documents based on the acquired data. The input is the data acquired in step 1. The output is the generated draft documents. Specifically, the data is converted into prompt sentences and sent to the generation AI model.

[0222] Step 3:

[0223] The server interactively presents the generated draft text to the user terminal. The input is the draft text generated in step 2. The output is the draft text displayed on the user terminal. Specifically, the server displays the draft text in the user interface, allowing the user to review and modify it.

[0224] Step 4:

[0225] The server automates email and chat communication with customers based on the generated draft text. The input is the draft text generated in step 2. The output is the email or chat message sent to the customer. Specifically, it converts the draft text into email or chat format and sends it automatically.

[0226] Step 5:

[0227] The server automatically generates formal proposals and contracts based on the generated draft documents. The input is the draft documents generated in step 2. The output is the formal proposal or contract document. Specifically, it formats the draft documents into a formal document format and converts them into a format that can be saved and printed.

[0228] Step 6:

[0229] The server organizes and analyzes customer information. The input is customer information stored in a database. The output is the analysis results and the organized customer information. Specifically, it organizes customer information by category and applies analytical algorithms to extract useful insights.

[0230] Step 7:

[0231] The server tracks and manages the progress of the sales process. Inputs are data and progress information related to sales activities. Outputs are progress reports. Specifically, it tracks each stage of sales activities and generates dashboards to visualize progress.

[0232] (Example 3)

[0233] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0234] Conventional draft document generation systems failed to adequately consider past approval history and approver tendencies, making it difficult to efficiently create draft documents that approved by approvers. Furthermore, draft document generation that reduced the workload on sales staff was insufficient. This resulted in delays in the approval process and decreased operational efficiency.

[0235] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes means for the user to input a prompt sentence, means for the terminal to send the prompt sentence to the server, means for the server to obtain past approval history and approver tendencies from a database, means for generating a draft text using a generation AI model, means for sending the generated draft text to the terminal, means for the terminal to display the draft text to the user, and means for the user to check the draft text and modify it as necessary. This makes it possible to efficiently generate draft text that takes into account past approval history and approver tendencies, and to provide draft text that is likely to be accepted by the approver. In addition, it is possible to reduce the burden on sales personnel and improve operational efficiency.

[0236] A "database" is a system that stores historical information and retrieves it as needed.

[0237] "Artificial intelligence" refers to a program or system that uses machine learning to automatically generate draft text.

[0238] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing the user to review and modify the text.

[0239] A "prompt message" is text that the user enters, containing instructions and conditions for generating a draft of the text.

[0240] A "terminal" is a device used by a user to input prompt messages and communicate with a server.

[0241] A "server" is a computer system that retrieves information from a database, generates draft text using a generative AI model, and sends it to a terminal.

[0242] A "generative AI model" is an artificial intelligence model that learns from past data and generates draft text.

[0243] "Approval history" refers to data that includes previously approved documents and feedback from approvers.

[0244] "Approver tendencies" refer to patterns based on the types of documents approvers have approved in the past or the types of feedback they have provided.

[0245] A "draft document" is a draft of a document generated by a generative AI model that is likely to be accepted by the approver.

[0246] A "user" is someone who enters prompt text and reviews and modifies the generated draft text.

[0247] This invention relates to a system for generating draft documents that take into account past approval history and the tendencies of approvers. Specific embodiments of this system are described below.

[0248] First, the user enters a prompt using the terminal. The prompt is text containing instructions and conditions for generating a draft document. For example, it might say, "Please create a document requesting budget approval for a new project. Based on past approval history, approvers tend to prioritize specific figures and clear objectives."

[0249] The terminal sends the entered prompt message to the server. This transmission uses communication protocols such as HTTP POST requests.

[0250] The server accesses the database to retrieve past approval history and approver trends. The database stores previously approved documents and approver feedback. For example, it might execute a query like "SELECT FROM approval_history WHERE approver_id = '12345'" using SQL.

[0251] Next, the server runs a generative AI model based on the acquired data to generate draft text. For example, OpenAI's GPT-4 is used as the generative AI model. The generative AI model has learned from past data and generates text that is likely to be accepted by the approver.

[0252] The generated draft text is sent from the server to the terminal. Communication protocols such as HTTP responses are used for transmission.

[0253] The terminal displays the received draft text to the user. The user interface includes a text area and display widgets. The user can review the displayed draft text and make revisions as needed. The revised text is used as the final approval request.

[0254] This system makes it possible to efficiently generate draft documents that take into account past approval history and approver tendencies, and to provide draft documents that are more likely to be accepted by approvers. Furthermore, it can reduce the workload on sales staff and improve operational efficiency. The flow of the specific processing in Example 3 will be explained using Figure 15.

[0255] Step 1:

[0256] The user enters a prompt message.

[0257] The user enters a prompt in the terminal's input field. For example, they might enter, "Please draft a document requesting budget approval for a new project. Based on past approval history, approvers tend to prioritize specific figures and clear objectives."

[0258] Input: Prompt message

[0259] Output: Input prompt message

[0260] Step 2:

[0261] The terminal sends a prompt message to the server.

[0262] The terminal sends the entered prompt text to the server as an HTTP POST request.

[0263] Input: Prompt message

[0264] Output: Prompt message sent to the server

[0265] Step 3:

[0266] The server retrieves past approval history and approver trends from the database.

[0267] The server accesses the database to retrieve past approval history and approver trends. For example, it might use an SQL query like "SELECT FROM approval_history WHERE approver_id = '12345'".

[0268] Input: Prompt message, database

[0269] Output: Past approval history and approver trends

[0270] Step 4:

[0271] The server generates draft text using an AI model.

[0272] The server runs a generative AI model (for example, OpenAI's GPT-4) based on the acquired data to generate draft text. The generative AI model has learned from past data and generates text that is likely to be accepted by the approver.

[0273] Input: Past approval history and approver trends

[0274] Output: Generated draft text

[0275] Step 5:

[0276] The server sends the generated text draft to the terminal.

[0277] The server sends the generated text draft to the terminal as an HTTP response.

[0278] Input: Generated text draft

[0279] Output: Text draft sent to the terminal

[0280] Step 6:

[0281] The terminal displays the text draft to the user.

[0282] The terminal displays the received text draft in a text area. The user can view the displayed text.

[0283] Input: Text draft sent to the terminal

[0284] Output: Text draft displayed to the user

[0285] Step 7:

[0286] The user checks the text draft and makes corrections if necessary.

[0287] The user checks the displayed text draft and makes corrections if necessary. For example, change "Labor cost: 5 million yen" to "Labor cost: 6 million yen" and make other necessary changes. The corrected text is used as the final approval request.

[0288] Input: Text draft displayed to the user

[0289] Output: Corrected text draft

[0290] (Application Example 3)

[0291] Next, we will explain Application Example 3 of Form Example 3. In the following explanation, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0292] Traditional electronic payment systems had the problem of requiring a great deal of time and effort to create documents that would satisfy approvers. Furthermore, because the documents were created without considering the approver's preferences, the approval process often suffered from delays. This resulted in inefficient expense reimbursement and large-scale transaction approval processes within companies, leading to a decline in overall business efficiency.

[0293] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[0294] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning; a server that interactively presents the generated draft texts; and a server that learns the approver's tendencies and generates draft texts that are likely to be accepted by the approver. This makes it possible to quickly generate draft texts that are easily accepted by the approver and streamline the electronic payment approval process.

[0295] "Past information" refers to data and history that the system has collected in the past.

[0296] A "database" is a collection of information that systematically stores past information and makes it searchable and retrievalable as needed.

[0297] "Machine learning" is a technology in which computers analyze large amounts of data and automatically learn patterns and rules.

[0298] "Draft text for proposals and contracts" refers to a draft of the text included in proposals and contracts.

[0299] "Artificial intelligence" is a system that uses technologies such as machine learning to mimic human intelligence and automatically perform specific tasks.

[0300] The "means for presenting in an interactive manner" refers to an interface for the user and the system to exchange information in an interactive format.

[0301] The "trend of the approver" is data indicating the criteria and patterns based on which a specific approver has carried out approvals in the past.

[0302] The "draft text that is highly likely to be acceptable to the approver" refers to a draft of an article that is easy to be approved, created considering the past trends of the approver.

[0303] As a form for implementing this invention, an electronic payment approval assistant system will be described as an example. This system operates with the cooperation of a server, a terminal, and a user.

[0304] First, the server has a database storing past information, which includes data on past approval histories and the trends of approvers. The server obtains information from this database and automatically generates a draft text for the proposal and contract parts using machine learning. Specifically, the server uses the APIs of Python, Pandas, and OpenAI to analyze past approval histories and learn the trends of approvers.

[0305] Next, the generated draft text is presented to the terminal in an interactive manner. The terminal is a device used by the user, such as a smartphone or a tablet, and the user can check the generated draft text through this terminal and make corrections if necessary. The interactive interface is designed so that the user can easily operate it.

[0306] Furthermore, the server learns the trends of the approvers and generates a draft text that is highly likely to be acceptable to the approvers. In this process, a draft text is generated using a generative AI model based on the features extracted from past approval histories. Examples of specific prompt texts are as follows.

[0307] Example of prompt text:

[0308] Based on the approver's tendencies, please generate the following draft text.

[0309] This business trip involves an important meeting with a new client and requires prompt approval.

[0310] Approver tendencies: {Characteristics extracted from the approver's past approval history}

[0311] This system enables the rapid generation of draft documents that are easily accepted by approvers, thereby streamlining the electronic payment approval process. For example, in the expense reimbursement process within a company, when an employee applies for travel expense reimbursement and requires supervisor approval, this system can automatically generate a draft document that is easily accepted by the supervisor, leading to quicker approval.

[0312] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[0313] Step 1:

[0314] The server retrieves approval history data from a database that stores past information.

[0315] Input: Past approval history data stored in the database

[0316] Output: Acquired approval history data

[0317] Specific operation: The server uses SQL queries to extract past approval history data from the database.

[0318] Step 2:

[0319] The server inputs the acquired approval history data into a machine learning model to learn the tendencies of the approvers.

[0320] Input: Acquired approval history data

[0321] Output: Learning results regarding approver tendencies

[0322] Specific operation: The server preprocesses the data using publicly available libraries and inputs it into a machine learning model. The model learns past approval patterns and extracts the tendencies of approvers.

[0323] Step 3:

[0324] The server inputs prompt sentences into the AI ​​model based on the learning results and generates suggested sentences.

[0325] Input: Learning results regarding approver tendencies, prompt text

[0326] Output: Generated draft text

[0327] Specific operation: The server uses the OpenAI API to input prompt text and training results into the AI ​​model, which then generates suggested text.

[0328] Step 4:

[0329] The server sends the generated draft text to the terminal and presents it interactively.

[0330] Input: Generated draft text

[0331] Output: Draft text displayed on the terminal

[0332] Specific operation: The server sends the generated draft text to the terminal via an HTTP request, and the terminal displays it in the user interface.

[0333] Step 5:

[0334] The user reviews the draft text generated through their device and makes revisions as needed.

[0335] Input: Draft text displayed on the terminal

[0336] Output: Revised draft text

[0337] Specific operation: The user reviews the draft text using the terminal interface and makes revisions in a text editor as needed.

[0338] Step 6:

[0339] The server saves the revised draft as the final approved document.

[0340] Input: Revised draft text

[0341] Output: Saved final approval document

[0342] Specific operation: The server saves the revised draft text to the database and completes the approval process.

[0343] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0344] "Example of form 1"

[0345] In one embodiment of the present invention, the AI ​​includes an emotion engine that recognizes the user's emotions. This emotion engine estimates the user's emotions from the tone of their voice, facial expressions, and word choice. Specifically, it generates positive suggestions when the user is happy, and suggestions that include apologies when the user is angry.

[0346] "Example of form 2"

[0347] Furthermore, the emotion engine adjusts the draft text based on the user's emotions. Specifically, it generates draft text that includes positive language when the user is happy and apologies when the user is angry. For example, when the user is happy, it generates text such as "This proposal will lead your business to even greater success," and when the user is angry, it generates text such as "We deeply apologize that our service did not meet your expectations."

[0348] "Example of form 3"

[0349] Furthermore, the emotion engine automatically generates draft text for proposals and contracts based on the user's emotions. Specifically, it generates positive proposals when the user is happy and proposals that include apologies when the user is angry. For example, when the user is happy, it generates a proposal such as "This proposal will lead your business to even greater success," and when the user is angry, it generates a proposal such as "We deeply apologize that our service did not meet your expectations."

[0350] The following describes the processing flow for each example of the form.

[0351] "Example of form 1"

[0352] Step 1: The AI ​​activates an emotion engine that estimates the user's emotions based on their tone of voice, facial expressions, and word choice.

[0353] Step 2: The emotion engine recognizes the user's emotions and determines the content of the suggestions based on those emotions.

[0354] Step 3: Based on the results of the emotion engine, the AI ​​generates and presents suggestions to the user. (Example 2)

[0355] Step 1: The AI ​​activates an emotion engine that recognizes the user's emotions.

[0356] Step 2: The emotion engine recognizes the user's emotions and determines the content of the draft text based on those emotions.

[0357] Step 3: Based on the results of the emotion engine, the AI ​​generates and presents text drafts to the user.

[0358] "Example of form 3"

[0359] Step 1: The AI ​​activates an emotion engine that recognizes the user's emotions.

[0360] Step 2: The emotion engine recognizes the user's emotions and determines the content of the proposal and contract based on those emotions.

[0361] Step 3: Based on the results of the emotion engine, the AI ​​generates proposal and contract sections and presents them to the user.

[0362] (Example 1)

[0363] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0364] In traditional sales activities, salespeople had to manually refer to past contract and proposal information to create new proposals. This process was time-consuming and labor-intensive, increasing the workload on salespeople. Furthermore, it was difficult to make appropriate proposals that resonated with the user's emotions, making it challenging to improve customer satisfaction.

[0365] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0366] In this invention, the server includes means for acquiring information from a database storing past information, means for training a machine learning model using the acquired information, means for automatically generating draft text for proposals and contracts using the trained machine learning model, means for presenting the generated draft text through an interactive interface, means including an emotion engine that recognizes the user's emotions, and means for adjusting the draft text based on information obtained from the emotion engine. This reduces the burden on salespeople and makes it possible to provide optimal proposals that are tailored to the user's emotions.

[0367] A "database" is an information management system that stores historical information and allows for the efficient retrieval of necessary information.

[0368] A "machine learning model" is an algorithm that learns patterns and rules based on data and uses them to make predictions and classifications on new data.

[0369] "Training" is the process of improving the performance of a machine learning model by having it learn from a large amount of data.

[0370] A "draft document" is a draft of a document that outlines the contents of a proposal or contract.

[0371] An "interactive interface" is a user interface that allows the user and the system to communicate in both directions.

[0372] An "emotion engine" is a system that estimates a user's emotions based on their tone of voice, facial expressions, and word choice.

[0373] "Adjustment" means changing the content or settings according to specific conditions or circumstances.

[0374] Modes for carrying out the invention

[0375] This invention is a system that retrieves information from a database containing historical data and automatically generates draft text for proposals and contracts using machine learning. A specific embodiment of this system is described below.

[0376] Retrieve information from the database.

[0377] The server retrieves the necessary information from a database that stores information on past contracts and proposals made by corporate sales representatives. This database includes contract details, proposal details, and customer responses. The server uses SQL queries to extract past contract information from the database. For example, it might execute a query like "SELECT FROM contracts WHERE date > '2020-01-01'".

[0378] Training machine learning models

[0379] The server uses the acquired information to train a machine learning model. This model has the ability to automatically generate draft text for proposals and contracts using techniques such as deep learning and reinforcement learning. Specifically, it uses a framework such as TensorFlow. The server builds a deep learning model using TensorFlow and trains the model using historical contract data. For example, it executes code like "model.fit(training_data, epochs=50)".

[0380] Generating a draft document

[0381] The server uses a trained machine learning model to generate draft text for the proposal and contract sections. This draft text is designed to include optimal content based on historical data. The server prompts the generation AI model with a text prompt and generates the draft text. For example, it executes code like "generate_text('Generate a new proposal text')".

[0382] The interface presents draft text.

[0383] The terminal presents the generated message draft to the salesperson through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant. The terminal displays the generated message draft in the chatbot's UI. For example, it executes code such as "chatbot.display_message(generated_text)".

[0384] User emotion recognition by an emotion engine

[0385] The device recognizes the user's emotions using an emotion engine. This emotion engine estimates emotions from the user's tone of voice, facial expressions, and word choice. The device uses OpenCV to analyze the user's facial expressions and estimate emotions. For example, it executes code like "emotion = emotion_recognition.detect_face_expression(user_image)".

[0386] Adjusting the wording to suit the emotions.

[0387] The server generates text suggestions tailored to the user's emotions based on information obtained from the emotion engine. When the user is happy, it generates positive suggestions; when the user is angry, it generates suggestions that include apologies. Based on the output of the emotion engine, the server inputs appropriate prompt sentences to the generation AI model. For example, it might execute code like "generate_text('Generate a suggestion sentence for when the user is angry')".

[0388] Specific example

[0389] Example 1: Generating a proposal

[0390] Based on the user's past contract information, the following proposal statement is generated.

[0391] Example of a prompt:

[0392] "Please generate a new proposal based on past contract information."

[0393] Example of a generated proposal:

[0394] "Customer, based on your past contract, we would like to offer you a new proposal with similar terms. We appreciate your consideration."

[0395] Example 2: Adjusting the wording to suit the emotions.

[0396] If the user is angry, the emotion engine recognizes that emotion and generates a draft message that includes an apology.

[0397] Example of a prompt:

[0398] "Please generate a suggestion message for when the user is angry."

[0399] Example of a generated proposal:

[0400] "We sincerely apologize for the inconvenience caused, customer. In this proposal, we have addressed past issues and will offer you better terms."

[0401] In this way, the system utilizes past data to automatically generate optimal suggestions tailored to the user's emotions.

[0402] The flow of the specific processing in Example 1 will be explained using Figure 17.

[0403] Step 1:

[0404] The server retrieves information from a database that stores historical data.

[0405] Input: Database query (Example: "SELECT FROM contracts WHERE date > '2020-01-01'")

[0406] Data processing: Execute SQL queries to extract past contract information.

[0407] Output: Dataset of acquired contract information

[0408] Specific operation: The server extracts historical contract information from the database using SQL queries. For example, it executes a query like "SELECT FROM contracts WHERE date > '2020-01-01'".

[0409] Step 2:

[0410] The server uses the acquired information to train a machine learning model.

[0411] Input: Header of acquired contract information

[0412] Data processing: Preprocess the dataset and convert it into a format suitable for machine learning models.

[0413] Output: Trained machine learning model

[0414] Specific operation: The server uses TensorFlow to build a deep learning model and trains the model using historical contract data. For example, it executes code like "model.fit(training_data, epochs=50)".

[0415] Step 3:

[0416] The server generates draft text for the proposal and contract sections using a trained machine learning model.

[0417] Input: A trained machine learning model, a prompt sentence (e.g., "Generate a new suggestion sentence")

[0418] Data processing: Input prompt messages into the model and generate suggested text options.

[0419] Output: Generated draft text

[0420] Specific operation: The server takes a prompt message as input to the generation AI model and generates a text proposal. For example, it executes code like "generate_text('Please generate a new proposal text')".

[0421] Step 4:

[0422] The terminal presents the generated draft text to the salesperson through an interactive interface.

[0423] Input: Generated draft text

[0424] Data processing: Convert the draft text into a format suitable for an interactive interface.

[0425] Output: Draft text to be presented to the business operator

[0426] Specific operation: The device displays the generated message draft in the chatbot's UI. For example, it executes code like "chatbot.display_message(generated_text)".

[0427] Step 5:

[0428] The device uses an emotion engine to recognize the user's emotions.

[0429] Input: User's tone of voice, facial expression, and word choice

[0430] Data processing: Analyze user emotions using an emotion engine.

[0431] Output: Estimated user sentiment

[0432] Specific operation: The device uses OpenCV to analyze the user's facial expressions and estimate their emotions. For example, it executes code like "emotion = emotion_recognition.detect_face_expression(user_image)".

[0433] Step 6:

[0434] The server generates text suggestions tailored to the user's emotions, based on information obtained from the emotion engine.

[0435] Input: Estimated user sentiment, trained machine learning model

[0436] Data processing: Generate prompt sentences that respond to emotions, input them into the model, and generate suggested sentences.

[0437] Output: Suggested text based on emotions

[0438] Specific operation: Based on the output of the emotion engine, the server inputs appropriate prompt text to the generative AI model. For example, it executes code such as "generate_text('Generate a suggestion text for when the user is angry')".

[0439] (Application Example 1)

[0440] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0441] Conventional sales support systems made it difficult for salespeople to quickly refer to past proposals and contract information when interacting with customers, and to make optimal proposals. Furthermore, it was difficult to recognize customer emotions in real time and respond accordingly. In addition, in physical stores, while store staff are required to make optimal proposals based on past sales data and proposal content, there was a lack of efficient means to do so. To solve these problems, a system is needed that recognizes customer emotions and generates proposal text that responds accordingly.

[0442] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0443] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, an emotion engine that recognizes the user's emotions, a means for generating proposal texts that correspond to the user's emotions, and a means installed in smart glasses to support customer service. This enables salespeople and store clerks to quickly make optimal proposals based on past proposals and contract information when interacting with customers, and to make proposals that correspond to the customer's emotions in real time.

[0444] "Past information" refers to data that includes information on contracts and proposals made by corporate sales representatives in the past.

[0445] A "database" is a system that stores historical information and allows you to retrieve that information as needed.

[0446] "Artificial intelligence" is a technology that uses machine learning to automatically generate draft text for proposals and contracts.

[0447] "Interactive presentation methods" refer to methods of presenting generated draft text to salespeople through chatbots, voice assistants, or similar means.

[0448] An "emotion engine" is a technology that estimates emotions from a user's tone of voice, facial expressions, and word choice.

[0449] "Means for generating suggestion text that responds to user emotions" refers to a method for generating appropriate suggestion text based on emotions estimated by an emotion engine.

[0450] "Smart glasses" are wearable devices equipped with displays and cameras that can display information and capture video.

[0451] "Means of supporting customer service" refers to a method of using smart glasses to display optimal suggestion text when store employees interact with customers.

[0452] The system for carrying out this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, an emotion engine that recognizes the user's emotions, means for generating proposal text that corresponds to the user's emotions, and means installed in smart glasses to support customer service.

[0453] The server retrieves information from a database containing historical data and automatically generates draft proposal and contract texts using machine learning. Specifically, it uses deep learning models and reinforcement learning models trained with TensorFlow. The generated draft texts are presented to sales representatives through interactive interfaces such as chatbots and voice assistants.

[0454] The device uses smart glasses to capture the customer's facial expressions and voice tone, and an emotion engine recognizes the user's emotions. The emotion engine uses an emotion recognition model based on TensorFlow to estimate emotions from the user's voice tone, facial expressions, and word choice. Based on the estimated emotions, a generative AI model is used to generate the most appropriate suggested text. The generative AI model uses a natural language generation model such as GPT-2.

[0455] When a user wears smart glasses and interacts with customers, the smart glasses' display shows optimal proposal text based on past sales data and previous suggestions. This allows the user to make suggestions tailored to the customer's emotions in real time.

[0456] For example, if a customer is smiling while looking at a product, the emotion recognition model will determine that the customer is "happy" and generate a suggestion message such as, "This product is especially popular. We are currently offering a special discount, so please give it a try." Conversely, if a customer has a confused expression, the emotion recognition model will determine that the customer is "confused" and generate a suggestion message such as, "If there is anything we can help you with, please let us know."

[0457] Examples of prompt messages include the following:

[0458] "The optimal proposal text for when the customer is happy: Past proposal data"

[0459] "The best proposal text when a customer is confused: Past proposal data"

[0460] This allows salespeople and staff to quickly provide optimal proposals based on past proposals and contract information when interacting with customers, and to make proposals that are tailored to the customer's emotions in real time.

[0461] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[0462] Step 1:

[0463] The server retrieves information from a database that stores historical data.

[0464] Input: Past proposals and contract information stored in the database

[0465] Output: Retrieved past proposals and contract information

[0466] Specific operation: The server uses SQL queries to extract the necessary information from the database and load it into memory.

[0467] Step 2:

[0468] The server uses machine learning to automatically generate draft text for proposals and contracts based on the acquired information.

[0469] Input: Past proposals and contract information

[0470] Output: Generated proposal / contract draft

[0471] Specific operation: The server inputs data into deep learning models and reinforcement learning models trained using TensorFlow, and generates optimal text suggestions.

[0472] Step 3:

[0473] The server presents the generated draft text to the salesperson through an interactive interface.

[0474] Input: Generated proposal / contract draft

[0475] Output: Draft text presented to the salesperson

[0476] Specific operation: The server sends the generated draft text to the salesperson's terminal via a chatbot or voice assistant and displays it.

[0477] Step 4:

[0478] The device uses smart glasses to capture the customer's facial expressions and voice tone.

[0479] Input: Customer's facial expressions and tone of voice

[0480] Output: Data of captured customer facial expressions and voice tone.

[0481] Specific operation: The camera and microphone built into the smart glasses on the device are used to capture the customer's facial expressions and voice in real time.

[0482] Step 5:

[0483] The device inputs the captured data into an emotion engine to recognize the user's emotions.

[0484] Input: Data of captured customer facial expressions and voice tone.

[0485] Output: Estimated user sentiment

[0486] Specific operation: The device inputs data into an emotion recognition model using TensorFlow and estimates the user's emotions.

[0487] Step 6:

[0488] The server generates the optimal suggested text using a generative AI model based on the estimated emotions.

[0489] Input: Estimated user sentiment, past suggestion data

[0490] Output: Suggestion text tailored to your emotions

[0491] Specific operation: The server inputs a prompt sentence into a natural language generation model such as GPT-2 and generates the most suitable suggested sentence.

[0492] Step 7:

[0493] The device displays the generated proposal text on the smart glasses' display.

[0494] Input: Generated proposal text

[0495] Output: Proposal text displayed on the smart glasses' screen

[0496] Specific operation: The device sends the generated proposal text to the smart glasses' display and displays it.

[0497] This allows users to make suggestions in real time that are tailored to the customer's emotions.

[0498] (Example 2)

[0499] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0500] In traditional sales activities, salespeople spent a great deal of time and effort drafting proposals and contract documents. Furthermore, it was difficult to generate appropriate wording that resonated with the user's emotions, making improving customer satisfaction a challenge.

[0501] The identification processing performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for acquiring information from a database storing past information and automatically generating draft text for the proposal and contract sections using machine learning, means for presenting the generated draft text interactively, an emotion engine for analyzing the user's emotions, and means for adjusting the draft text based on the user's emotions. This reduces the burden on salespeople and enables the generation of appropriate draft text that responds to the user's emotions.

[0502] A "database" is a system that stores historical information and retrieves it as needed.

[0503] "Artificial intelligence" is a technology that uses machine learning to automatically generate draft text for proposals and contracts.

[0504] "Interactive presentation methods" refer to methods of interactively displaying generated draft text to the user.

[0505] An "emotion engine" is a technology that analyzes a user's emotions and identifies their emotional state.

[0506] "Methods for adjusting draft text" refers to methods for appropriately modifying draft text that has been generated based on user sentiment.

[0507] Modes for carrying out the invention

[0508] This invention relates to a system that reduces the burden on business operators and generates appropriate text drafts that respond to the user's emotions. Specific embodiments of this system are described below.

[0509] System Configuration

[0510] This system consists of the following main components:

[0511] 1. Database: A system for storing historical information and retrieving it as needed.

[0512] 2. Artificial Intelligence (AI): This is a technology that uses machine learning to automatically generate draft text for proposals and contracts. Specifically, it uses a generative AI model (for example, OpenAI's GPT-4).

[0513] 3. Interactive presentation methods: These are methods of interactively displaying the generated draft text to the user. For example, using a web interface or mobile application.

[0514] 4. Emotion Engine: This is a technology for analyzing a user's emotions and identifying their emotional state. Specifically, it involves using an emotion engine.

[0515] 5. Means for adjusting the draft text: This refers to methods for appropriately modifying the draft text that was generated based on the user's emotions.

[0516] System operation

[0517] The server collects basic information such as the proposed content and contract terms entered by the user on the terminal. For example, if a user proposes a new marketing strategy, they would enter "New Marketing Strategy" as the proposed content and "Annual Contract" as the contract terms.

[0518] Next, the server generates prompt messages for the AI ​​model based on the data entered by the user. For example, it generates prompt messages like the following:

[0519] Proposal: New marketing strategy

[0520] Contract terms: Annual contract

[0521] Please generate a draft of the text.

[0522] The server sends the generated prompt text to the AI ​​model, which then generates suggested text based on the prompt text. For example, it might generate a suggested text such as, "This new marketing strategy will lead your business to even greater success."

[0523] The server then collects the user's emotional data and sends it to the emotion engine. The emotion engine analyzes the user's emotions and identifies emotional states such as happy or angry.

[0524] The server adjusts the generated text based on the analysis results of the emotion engine. For example, if the user is happy, the text is adjusted to include positive language. If the user is angry, the text is adjusted to include an apology.

[0525] Finally, the server sends the final draft text to the terminal, where the user can review and revise it as needed.

[0526] Specific example

[0527] As a concrete example, consider the following scenario:

[0528] Scenario 1: When the user is happy

[0529] Input data: Proposal content "New marketing strategy", Contract terms "Annual contract"

[0530] Output from the generating AI model: Draft text: "This new marketing strategy will lead your business to even greater success."

[0531] Emotion engine adjustment: Since the user is happy, it generates text suggestions that include positive language.

[0532] Scenario 2: When the user is angry

[0533] Input data: Proposal content "Service improvement plan", Contract terms "Monthly contract"

[0534] Output from the generated AI model: Draft text: "We sincerely apologize that our service did not meet your expectations."

[0535] Emotion engine adjustment: Since the user is angry, generate a draft message that includes an apology.

[0536] In this way, the system operates by having the server, terminal, and user work together to generate draft documents that reduce the workload on sales representatives.

[0537] The flow of the specific processing in Example 2 will be explained using Figure 19.

[0538] Step 1: Collecting user input data

[0539] The user enters basic information such as the proposal details and contract terms into the terminal. For example, if the user proposes a new marketing strategy, they would enter "New Marketing Strategy" for the proposal details and "Annual Contract" for the contract terms. The entered data is then sent to the server.

[0540] Step 2: Generating prompts for the generative AI model

[0541] The server generates prompts for the AI ​​model based on the data entered by the user. For example, it generates prompts like the following:

[0542] Proposal: New marketing strategy

[0543] Contract terms: Annual contract

[0544] Please generate a draft of the text.

[0545] This prompt message is sent to the generating AI model.

[0546] Step 3: Generating draft text using a generative AI model

[0547] The server sends the generated prompt text to a generative AI model (for example, OpenAI's GPT-4). The generative AI model generates a text draft based on the prompt text. For example, it might generate a text draft such as, "This new marketing strategy will lead your business to even greater success." The generated text draft is then returned to the server.

[0548] Step 4: User sentiment analysis using an emotion engine

[0549] The server collects user emotion data and sends it to the emotion engine. The emotion engine analyzes the user's emotions and identifies emotional states such as happiness or anger. The analysis results are then returned to the server.

[0550] Step 5: Emotion-based adjustments to the draft text

[0551] The server adjusts the generated text drafts based on the analysis results of the emotion engine. For example, if the user is happy, the drafts are adjusted to include positive language. If the user is angry, the drafts are adjusted to include apologies. The adjusted drafts are saved on the server.

[0552] Step 6: Output of the final draft

[0553] The server sends the final draft document to the terminal. The user can review the final draft document on the terminal and make revisions as needed. After the user's review, the final draft document is used as the official proposal or contract document.

[0554] (Application Example 2)

[0555] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0556] In traditional sales activities, salespeople spend a significant amount of time and effort drafting proposals and contract documents. Furthermore, they need to respond appropriately to customer emotions, which is difficult to do manually. This increases the burden on salespeople and can lead to decreased customer satisfaction. Therefore, there is a need for a system that reduces the burden on salespeople and automatically generates appropriate document drafts that respond to customer emotions.

[0557] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0558] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, a means for analyzing the user's emotions, and a means for adjusting the draft texts based on the user's emotions. This reduces the burden on salespeople and enables the automatic generation of appropriate draft texts that respond to customer emotions.

[0559] A "database" is a system that stores historical information and retrieves it as needed.

[0560] "Machine learning" is a technology that allows computers to learn patterns from data and perform predictions and classifications.

[0561] A "draft of the proposal / contract section" refers to a draft of a document used when making proposals or entering into contracts with customers during sales activities.

[0562] "Interactive presentation methods" refer to interfaces for displaying generated text drafts while interacting with the user.

[0563] "Methods for analyzing user emotions" refer to technologies for analyzing emotions from user input and behavior.

[0564] "Means for adjusting text drafts based on user emotions" refers to technologies for appropriately modifying text drafts according to the analyzed user emotions.

[0565] The system for implementing this invention consists of three main elements: a server, a terminal, and a user. The server includes a database storing historical information, a machine learning model, a sentiment analysis engine, and a generative AI model. The terminal is a device operated by the user, such as a smartphone or a personal computer. The user is a salesperson or customer support representative.

[0566] The server first retrieves historical information from the database. This information includes past proposals, contract texts, and customer feedback. Next, it uses a machine learning model to automatically generate draft proposal and contract texts from the retrieved information. These generated draft texts are then presented interactively to the user's terminal.

[0567] Users can review draft text and make revisions as needed. Furthermore, the server uses a sentiment analysis engine to analyze user emotions from their input and actions. This sentiment analysis engine can utilize existing sentiment analysis tools.

[0568] Once the user's emotions are analyzed, a generative AI model adjusts the draft text based on those emotions. For example, if the user is happy, it generates a draft text that includes positive language; if the user is angry, it generates a draft text that includes apologies. Generative AI models such as OpenAI's GPT-3 are used in this process.

[0569] As a concrete example, the prompt message when the user is satisfied with the product is as follows:

[0570] Prompt message (suggested message when the user is pleased): Thank you for your recent purchase. We hope you are satisfied with our service.

[0571] By inputting this prompt into the AI ​​model, it is possible to generate appropriate text suggestions that reflect the user's emotions.

[0572] This system reduces the workload on sales staff and enables the automatic generation of appropriate message drafts tailored to customer sentiment. This is expected to improve customer satisfaction and sales efficiency.

[0573] The flow of a specific process in Application Example 2 will be explained using Figure 20.

[0574] Step 1:

[0575] The server retrieves historical information from the database. Inputs include data such as past proposals, contract documents, and customer feedback. Outputs are the retrieved historical information. Specifically, the server executes SQL queries to extract the necessary information from the database.

[0576] Step 2:

[0577] The server uses a machine learning model to automatically generate draft text for proposals and contracts from the acquired information. The input is the historical information acquired in Step 1. The output is the generated draft text. Specifically, the server executes a machine learning algorithm to generate new draft text based on the historical information.

[0578] Step 3:

[0579] The server interactively presents the generated draft text to the terminal. The input is the draft text generated in step 2. The output is the draft text displayed on the terminal. Specifically, the server sends the draft text to the terminal in JSON format, and the terminal displays it in its user interface.

[0580] Step 4:

[0581] The user reviews the draft text via the terminal and makes revisions as needed. The input is the draft text presented in step 3. The output is the draft text revised by the user. Specifically, the user edits the draft text using the input fields on the terminal.

[0582] Step 5:

[0583] The server uses a sentiment analysis engine to analyze emotions from user input and actions. The input consists of the revised text draft from step 4 and the user's input data. The output is the user's sentiment score. Specifically, the server calls an API to analyze the user's input data and obtain the sentiment score.

[0584] Step 6:

[0585] The server uses a generative AI model to adjust the text drafts based on the user's emotions. The input is the emotion score obtained in step 5 and the modified text draft in step 4. The output is the text draft adjusted according to the emotions. Specifically, the server inputs a prompt sentence into OpenAI's GPT-3 model and generates an emotion-appropriate text draft.

[0586] Step 7:

[0587] The server presents the revised text draft to the terminal again. The input is the revised text draft from step 6. The output is the final text draft displayed on the terminal. Specifically, the server sends the revised text draft to the terminal in JSON format, and the terminal displays it in its user interface.

[0588] (Example 3)

[0589] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0590] Traditional systems could generate draft documents based on past information, but they couldn't generate documents that reflected the user's emotions. Therefore, it was difficult to automatically generate documents such as proposals and apologies that were appropriate to the user's feelings. Furthermore, the lack of concrete means to generate documents that approved by the approver resulted in an inefficient approval process.

[0591] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[0592] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, an emotion engine for detecting the user's emotions, and a means for generating draft texts based on emotion data obtained from the emotion engine. This makes it possible to automatically generate appropriate draft texts according to the user's emotions and provide draft texts that are likely to satisfy the approver.

[0593] A "database" is a system that stores historical information and retrieves it as needed.

[0594] "Artificial intelligence" is a technology that uses machine learning to analyze data and automatically generate draft text for proposals and contracts.

[0595] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing the user to review and modify the draft text.

[0596] An "emotion engine" is a technology that detects emotions from a user's facial expressions, voice, etc., and analyzes that emotional data.

[0597] "Emotional data" refers to data that indicates the user's emotional state as detected by the emotion engine.

[0598] A "draft document" is a draft of a proposal or contract document, generated by artificial intelligence.

[0599] This invention is a system that combines a database storing past information, artificial intelligence that generates text drafts using machine learning, and an emotion engine that detects user emotions. Specific embodiments of this system are described below.

[0600] System Configuration

[0601] database

[0602] The server uses a database that stores past approval history and approver trends. This database can be a relational database management system (RDBMS), such as MySQL. The database stores approval history data and approver trend data.

[0603] artificial intelligence

[0604] The server uses a generative AI model (e.g., GPT-4) to generate draft text based on information retrieved from the database. The generative AI model learns from past approval history and approver tendencies, and generates new draft text based on that.

[0605] Emotional Engine

[0606] The device uses an emotion engine to detect the user's emotions in real time. The emotion engine analyzes the user's facial expressions and voice to determine their emotions and sends the results to the server.

[0607] System operation

[0608] Data collection

[0609] The server retrieves historical approval history data and approver trend data from the database. For example, the server retrieves data by executing an SQL query such as "SELECT FROM approval_history WHERE approver_id = 'X'".

[0610] Data Learning

[0611] The server inputs the acquired data into a generating AI model to train the model. During the training process, it analyzes past approval history and approver tendencies to extract patterns.

[0612] Text draft generation

[0613] The server inputs a prompt message into the trained generative AI model, which then generates a new draft text. The prompt message includes specific instructions based on the approver's tendencies. For example, it might input a prompt message such as, "Generate a draft text that will satisfy approver X. Please include specific numbers."

[0614] Emotion detection

[0615] The device captures the user's facial expressions with its camera and sends the analysis results to the server by calling an API. For example, if the user is happy, the emotion engine detects that emotion and sends data indicating "the user is happy" to the server.

[0616] Emotion-based text drafting

[0617] The server generates appropriate text suggestions based on the received sentiment data. For example, if the user is happy, the server inputs a prompt message to the generation AI model saying, "Generate a suggestion text suggestion for when the user is happy." The generation AI model then generates a text suggestion such as, "This suggestion will lead your business to even greater success."

[0618] Specific example

[0619] Specific examples of data collection

[0620] The server executes the SQL query "SELECT FROM approval_history WHERE approver_id = 'X'" to retrieve data from the database.

[0621] Examples of text draft generation

[0622] The server inputs the prompt message, "Generate a draft document that will satisfy approver X. Please include specific numerical values," into the AI ​​model and receives the generated draft document.

[0623] Examples of generating text drafts based on emotions

[0624] The server inputs the prompt message "Generate suggested text for when the user is happy" into the AI ​​model and presents the generated text suggestions to the user.

[0625] In this way, the server, terminal, and user work together to generate the optimal draft text that reflects the emotions of the approver and the user. The flow of the specific processing in Example 3 will be explained using Figure 21.

[0626] Step 1: Data Collection

[0627] The server retrieves historical approval history data and approver trend data from the database. It takes an SQL query (e.g., "SELECT FROM approval_history WHERE approver_id = 'X'") as input and loads the retrieved data into memory as output. Specifically, the server establishes a database connection, executes the SQL query, and retrieves the results.

[0628] Step 2: Data Training

[0629] The server inputs the acquired data into a generative AI model (e.g., GPT-4) to train the model. Past approval history data and approver trend data are used as input, and a trained generative AI model is obtained as output. Specifically, the server preprocesses the data, inputs it into the generative AI model, and starts the training process.

[0630] Step 3: Draft document generation

[0631] The server inputs a prompt sentence into a fully trained generative AI model and generates a new draft text. The input is a prompt sentence (e.g., "Generate a draft text that will satisfy approver X. Please include specific numerical values."), and the output is the generated draft text. Specifically, the server inputs the prompt sentence into the generative AI model and receives the generated draft text.

[0632] Step 4: Emotion Detection

[0633] The device captures the user's facial expressions with its camera and calls an API to send the analysis results to the server. It uses the user's facial expression data as input and obtains analyzed emotion data as output. Specifically, the device captures the user's facial expressions with its camera, sends the data to the API, and receives the analysis results.

[0634] Step 5: Emotion-based drafting

[0635] The server generates appropriate text suggestions based on the received sentiment data. It uses sentiment data (e.g., "The user is happy") as input and outputs text suggestions corresponding to that sentiment. Specifically, the server analyzes the sentiment data, inputs a prompt (e.g., "Generate suggestion text suggestions for when the user is happy.") into the AI ​​model, and receives the generated text suggestions.

[0636] In this way, the server, terminal, and user work together to generate the optimal draft text that reflects the emotions of the approver or user.

[0637] (Application Example 3)

[0638] Next, we will explain Application Example 3 of Form Example 3. In the following explanation, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0639] Conventional draft document generation systems struggled to generate draft documents that considered user emotions, thus failing to improve user satisfaction. Furthermore, the lack of appropriate means to generate draft documents that approved by approvers sometimes led to delays in the approval process. Additionally, draft document generation that reduced the workload on sales representatives was insufficient.

[0640] In Application Example 3, the specific processing performed by the specific processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes artificial intelligence that retrieves information from a database storing past information and automatically generates draft text for the proposal and contract sections using machine learning, means for interactively presenting the generated draft text, an emotion engine that analyzes the user's emotions, and means for adjusting the draft text based on the emotion engine. This makes it possible to generate draft text that takes the user's emotions into consideration, thereby increasing user satisfaction. Furthermore, by generating draft text that approvers can accept, delays in the approval process can be prevented. In addition, draft text can be generated to reduce the workload of sales personnel.

[0641] "Past information" refers to data such as the user's or approver's past behavioral history, transaction history, and approval history.

[0642] A "database" refers to an information aggregation system that systematically stores past information and allows it to be searched and retrieved as needed.

[0643] "Machine learning" refers to a technology that uses computers to analyze large amounts of data, find patterns and regularities, and automatically perform specific tasks.

[0644] "Draft proposal / contract text" refers to a draft document containing the proposed content and contract details that will be presented to users and approvers.

[0645] "Artificial intelligence" refers to computer systems that use technologies such as machine learning and natural language processing to imitate and support human intellectual tasks.

[0646] "Interactive presentation methods" refer to interfaces that interactively display generated draft text to users and approvers and receive their feedback.

[0647] An "emotion engine" refers to a software module that analyzes a user's emotions and takes appropriate action based on the results.

[0648] "Means for adjusting draft text" refers to a function for appropriately modifying and adjusting the generated draft text based on the analysis results of the emotion engine.

[0649] A system for carrying out this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, an emotion engine that analyzes the user's emotions, and means for adjusting the draft text based on the emotion engine.

[0650] The server first retrieves information from the database, such as the user's and approver's past behavioral history, transaction history, and approval history. Based on this information, it uses a machine learning algorithm to generate draft text for the proposal and contract sections. The generated draft text is presented interactively to the user or approver, and feedback can be received.

[0651] Next, the emotion engine analyzes the user's emotions. The emotion engine analyzes the user's emotions from their input and actions, and adjusts the draft text based on the results. For example, if the user is feeling anxious, a reassuring message can be added to the draft text.

[0652] This system requires software such as APIs and sentiment analysis modules. Hardware requirements include servers and smartphones.

[0653] As a concrete example, consider generating a message when a user makes a large payment. If the user ID is 12345 and the payment amount is 10,000 yen, the following prompt message will be generated based on the user's past payment history and sentiment.

[0654] Example of a prompt:

[0655] User ID: 12345

[0656] Payment amount: 10,000 yen

[0657] Past payment history: Past payment history data

[0658] User's emotion: Anxiety

[0659] Please generate an appropriate message.

[0660] By inputting this prompt into the AI ​​generation model, it is possible to generate appropriate messages that take into account the user's emotions. In this way, a system that can increase user satisfaction can be realized.

[0661] The flow of the specific processing in Application Example 3 will be explained using Figure 22.

[0662] Step 1:

[0663] The server retrieves information from the database, such as the past activity history, transaction history, and approval history of users and approvers. Inputs are user IDs and approver IDs, and output is data such as past activity history, transaction history, and approval history. This data is used in subsequent processing.

[0664] Step 2:

[0665] The server generates draft text for the proposal and contract sections using a machine learning algorithm based on the acquired historical information. The input is the historical information acquired in step 1, and the output is the generated draft text. Specifically, the system inputs the historical information into a machine learning model and generates the optimal draft text.

[0666] Step 3:

[0667] The server interactively presents the generated draft text to the user or approver. The input is the draft text generated in step 2, and the output is the feedback from the user or approver. Specifically, it displays the draft text through the user interface and receives comments and revision requests from the user or approver.

[0668] Step 4:

[0669] The server analyzes user emotions using an emotion engine based on user input and behavior. Input consists of user feedback and behavioral data, while output is the user's emotional state. Specifically, it uses an emotion analysis module to analyze the user's emotions.

[0670] Step 5:

[0671] The server adjusts the draft text based on the analysis results of the emotion engine. The input is the user's emotional state obtained in step 4 and the draft text generated in step 2, and the output is the adjusted draft text. Specifically, the server modifies the content of the draft text according to the user's emotions; for example, it adds a reassuring message to a user who is feeling anxious.

[0672] Step 6:

[0673] The server presents the revised draft text to the user or approver again and receives final feedback. The input is the revised draft text from step 5, and the output is the final approval or revision request. Specifically, the draft text is displayed again through the user interface for final confirmation.

[0674] Step 7:

[0675] The server stores the final draft document in a database and uses it to generate draft documents for future proposals and contract sections as needed. The input is the final draft document, and the output is the updated database. Specifically, it stores the final draft document in the database and uses it as training data for future machine learning models.

[0676] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0677] Data generation model 58 is a form of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0678] Other examples of generative AI include Gemini® (registered trademark) (Internet search). <url: https: gemini.google.com ?hl="ja">) are some examples.

[0679] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0680] [Second Embodiment]

[0681] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0682] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0683] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0684] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0685] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0686] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0687] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0688] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0689] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0690] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0691] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0692] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[0693] "Example of form 1"

[0694] In one embodiment of the present invention, there exists a database storing past information. This database includes, for example, information on contracts and proposals made by corporate sales representatives in the past. There exists an AI that retrieves information from this database and automatically generates draft text for proposals and contracts using machine learning. This AI is trained using machine learning techniques such as deep learning and reinforcement learning. The generated draft text is presented to the sales representative through an interactive interface. This interface is implemented, for example, in the form of a chatbot or a voice assistant. "Example of Form 2"

[0695] In another embodiment of the present invention, AI generates draft documents to reduce the workload of salespeople. Specifically, the AI ​​automates some of the tasks that salespeople normally perform. For example, it significantly reduces the time and effort salespeople spend creating draft proposals and contract documents.

[0696] "Example of form 3"

[0697] In a further embodiment of the present invention, the AI ​​generates draft text that is expected to be acceptable to the approver. Specifically, the AI ​​learns past approval history and approver tendencies, and generates draft text based on that. This makes it possible to provide draft text that is highly likely to be acceptable to the approver.

[0698] The following describes the processing flow for each example of the form.

[0699] "Example of form 1"

[0700] Step 1: Retrieve the necessary information from a database containing past data. This information could include, for example, past contracts and proposals.

[0701] Step 2: Based on the acquired information, machine learning is used to automatically generate draft text for the proposal and contract sections. This AI is trained using machine learning techniques such as deep learning and reinforcement learning.

[0702] Step 3: The generated draft text is presented to the salesperson through an interactive interface. This interface can be implemented, for example, as a chatbot or voice assistant.

[0703] "Example of form 2"

[0704] Step 1: The AI ​​generates draft text to reduce the workload of salespeople. Specifically, the AI ​​automates some of the tasks that salespeople normally perform.

[0705] Step 2: For example, significantly reduce the time and effort salespeople spend creating proposals and draft contract documents.

[0706] "Example of form 3"

[0707] Step 1: The AI ​​generates draft text that is expected to satisfy the approver. Specifically, the AI ​​learns from past approval history and the approver's tendencies.

[0708] Step 2: Generate a draft document based on that. This makes it possible to provide a draft document that is more likely to be accepted by the approver.

[0709] (Example 1)

[0710] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0711] In traditional corporate sales, a major problem was the significant time and effort required to create proposals and contracts. Furthermore, it was difficult for salespeople to efficiently utilize information from past proposals and contracts, sometimes resulting in a decline in the quality of proposals. Additionally, creating documents that would satisfy approvers required experience and skill, placing a significant burden on new salespeople.

[0712] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0713] In this invention, the server includes means for retrieving information from a database storing historical information, means for training a machine learning model using the retrieved information, means for automatically generating draft text for proposals and contracts using the trained machine learning model, and means for interactively presenting the generated draft text. This reduces the burden on salespeople and enables the efficient and high-quality creation of proposals and contracts. Furthermore, by automatically generating draft text that is acceptable to approvers, the approval process can also be expedited.

[0714] A "database" is a system that stores historical information and allows you to retrieve that information as needed.

[0715] "Means of acquiring information" refers to the methods and techniques for extracting necessary information from a database.

[0716] A "machine learning model" is an algorithm or framework that learns from data and automatically performs a specific task.

[0717] "Training methods" refer to the methods and techniques used to train machine learning models using data.

[0718] "Methods for automatically generating draft documents" refer to methods and technologies that use trained machine learning models to automatically create drafts of new proposals and contracts.

[0719] "Interactive presentation methods" refer to methods and technologies for interactively displaying generated draft text to the user.

[0720] "Salesperson" refers to the person or staff member who handles corporate sales.

[0721] An "approver" refers to a person who has the authority to review and approve the contents of a proposal or contract.

[0722] Modes for carrying out the invention

[0723] This invention is a system designed to streamline the creation of proposals and contracts in corporate sales, thereby reducing the burden on salespeople. A specific embodiment of this system is described below.

[0724] Retrieve information from the database.

[0725] The server retrieves the necessary information from a database containing information on past contracts and proposals made by corporate sales representatives. This database includes contract details, proposal details, dates, and customer information. The server extracts information from the database using SQL queries. For example, executing a query such as "SELECT FROM contracts WHERE date > '2020-01-01'" retrieves contract information from 2020 onwards.

[0726] Training machine learning models

[0727] The server uses the acquired information to train a machine learning model. This model employs techniques such as deep learning and reinforcement learning. Specifically, it uses the TensorFlow framework. The server cleans and normalizes the data, and then trains the model using the preprocessed data. For example, it trains a text generation model using an LSTM (Long Short-Term Memory) network.

[0728] Automatic generation of draft text

[0729] The server uses a trained LSTM model to automatically generate draft proposals and contract documents. When a user enters a prompt such as "Generate a new proposal," the server inputs this prompt into the model and retrieves the generated draft document. For example, it might generate a document stating, "This proposal provides the best solution for your needs."

[0730] Presentation to the business operator

[0731] The generated draft text is presented to the salesperson through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant. The salesperson accesses the chatbot through their device and reviews the generated draft text. For example, if the salesperson types "Show me the draft text of the proposal" into the chatbot, the chatbot will display the text "This proposal provides the best solution for your company's needs."

[0732] Specific examples and prompt statements

[0733] Suppose a user needs to create a new proposal. The user accesses the system via their device and instructs it to generate draft text based on information from similar past proposals. The server retrieves past proposal information from the database and uses a machine learning model to generate a new draft text. The generated draft text is then presented to the user via a chatbot.

[0734] Example of a prompt:

[0735] "Please generate a draft of a new proposal document based on past corporate sales proposal data."

[0736] In this way, users can efficiently create new proposals.

[0737] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0738] Step 1:

[0739] The server retrieves information on past contracts and proposals from the database. SQL queries are used as input. For example, executing the query "SELECT FROM contracts WHERE date > '2020-01-01'" will output contract information from 2020 onwards. The server stores the retrieved data in memory.

[0740] Step 2:

[0741] The server preprocesses the acquired data. Raw data retrieved from the database is used as input. Preprocessing includes data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). This results in output data in a format suitable for training machine learning models.

[0742] Step 3:

[0743] The server trains a machine learning model using preprocessed data. Preprocessed data is used as input. Specifically, it trains an LSTM (Long Short-Term Memory) network using the TensorFlow framework. The training results in a model for generating proposal and contract texts.

[0744] Step 4:

[0745] The server uses a trained LSTM model to automatically generate drafts of new proposals and contract documents. The input is a prompt sentence entered by the user. For example, the prompt "Generate a new proposal" might be entered. The server inputs this prompt into the model and outputs a generated draft document. For example, it might generate a document stating, "This proposal provides the best solution for your needs."

[0746] Step 5:

[0747] The server presents the generated draft text to the salesperson through an interactive interface. The generated draft text is used as input. The interactive interface is implemented in the form of a chatbot or voice assistant. For example, if the salesperson types "Show me the draft text of the proposal" into the chatbot, the chatbot will display the text "This proposal provides the best solution for your company's needs."

[0748] (Application Example 1)

[0749] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0750] In traditional sales activities, sales representatives must manually refer to past contract and proposal information to create new proposals and contracts, which is time-consuming and labor-intensive. Furthermore, maintaining a consistent level of quality is difficult because the quality of proposals and contracts depends on the sales representative's experience and skills. There is a growing need for a way for sales representatives to efficiently create proposals even when they are on the go.

[0751] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0752] This invention includes a server that retrieves information from a database storing historical data and automatically generates draft proposal and contract documents using machine learning; a server that interactively presents the generated draft documents; and a server that presents the generated draft documents to sales representatives via smartphone. This enables sales representatives to quickly generate high-quality proposal and contract documents based on historical data and to conduct sales activities efficiently even when on the go.

[0753] A "database containing past information" is a data storage system that stores information on past contracts and proposals made by corporate sales representatives, allowing them to search and retrieve the information as needed.

[0754] "Machine learning" is a technology in which computers learn patterns and rules from data and use that knowledge to make predictions and decisions about new data.

[0755] "Artificial intelligence that automatically generates draft text for proposals and contracts" refers to a program that uses machine learning to analyze past data and automatically generate new proposal and contract documents.

[0756] "Interactive presentation methods" refer to interfaces that interactively display generated draft text to the user, allowing the user to review and modify its content.

[0757] The "means of presenting information to sales representatives via smartphone" refers to a function that displays generated draft text on the sales representative's smartphone, enabling them to conduct sales activities efficiently even when they are out of the office.

[0758] A system for implementing this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, and means for presenting the generated draft text to sales representatives via smartphone.

[0759] System Configuration

[0760] 1. Database: The server has a database that stores information on past contracts and proposals made by corporate sales representatives. This database is built using a publicly known database management system.

[0761] 2. Artificial Intelligence: The server includes artificial intelligence that uses machine learning (e.g., OpenAI's GPT-3) to automatically generate new proposals and contracts based on past data. This artificial intelligence is implemented using a publicly known programming language.

[0762] 3. Interactive Interface: The generated draft text is presented to the sales representative through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant.

[0763] 4. Smartphone Application: The generated draft text is displayed on the sales representative's smartphone. The smartphone application runs on the Android platform, enabling sales representatives to conduct sales activities efficiently even when they are on the go.

[0764] Processing flow

[0765] 1. Data Retrieval: The server retrieves past proposals and contracts for a specific client from the database.

[0766] 2. Prompt generation: The server generates prompts to input into the artificial intelligence based on previously retrieved suggestion sentences.

[0767] 3. Document draft generation: The artificial intelligence generates new proposals and contracts based on the generated prompts.

[0768] 4. Presentation of draft text: The generated draft text is presented to the sales representative through an interactive interface.

[0769] 5. Smartphone display: The generated draft text is displayed on the sales representative's smartphone, allowing them to conduct sales activities efficiently even when they are out of the office.

[0770] Hardware and software to be used

[0771] Hardware: Servers, smartphones

[0772] software

[0773] Specific example

[0774] For example, if a past proposal from client ID 123 was "This product is high-quality and offers excellent value for money," the newly generated proposal might be "This product uses the latest technology and is highly reliable."

[0775] Example of a prompt

[0776] Previous proposal: This product is high quality and offers excellent value for money. Please generate a new proposal:

[0777] In this way, sales representatives can easily generate new proposals using their smartphones and make effective proposals to clients.

[0778] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0779] Step 1:

[0780] The server retrieves past proposals and contracts for a specific client from the database. The input is the client ID, and the output is a list of past proposals and contracts. Specifically, the server executes an SQL query to extract proposals related to the corresponding client ID from the database.

[0781] Step 2:

[0782] The server generates prompts for the artificial intelligence based on the previously retrieved proposals. The input is a list of past proposals, and the output is a prompt. Specifically, the server combines the past proposals into a single text and creates a prompt with the preface "Past proposals: ".

[0783] Step 3:

[0784] The server sends the generated prompt text to the artificial intelligence, which then generates new proposal or contract texts. The input is the prompt text, and the output is the generated proposal text. Specifically, the server calls the OpenAI GPT-3 API, passing the prompt text as input and receiving the generated text.

[0785] Step 4:

[0786] The server presents the generated proposal to the sales representative through an interactive interface. The input is the generated proposal, and the output is an interface that the sales representative can review and modify. Specifically, the server displays the generated proposal in a chatbot format, allowing the sales representative to review and modify its contents.

[0787] Step 5:

[0788] The server displays the generated proposal on the sales representative's smartphone. The input is the generated proposal, and the output is the proposal displayed on the smartphone screen. Specifically, the server sends the generated proposal to the smartphone application, allowing the sales representative to review and use the proposal even when they are away from the office.

[0789] (Example 2)

[0790] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0791] There is a need to reduce the time and effort salespeople spend creating proposals and contract drafts, and to proceed with their work more efficiently. It is also crucial to maintain a consistent quality of draft documents and to quickly generate content that approves. A system is needed to address these challenges.

[0792] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0793] In this invention, the server includes a generation AI model that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, means for generating prompt texts based on information input by the user, and means for interactively presenting the generated draft texts. This reduces the burden on sales personnel and makes it possible to efficiently generate high-quality draft texts.

[0794] A "database" is a system that stores historical information and retrieves it as needed.

[0795] "Machine learning" is a technology that allows computers to learn patterns from data and perform predictions and classifications.

[0796] A "generative AI model" is an artificial intelligence model that uses machine learning to automatically generate draft text for proposals and contracts.

[0797] A "prompt statement" is a sentence used to give instructions to an AI model based on information entered by the user.

[0798] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing for confirmation and modification.

[0799] A "salesperson" is a person responsible for creating proposals and drafting contract documents.

[0800] A "draft document" is a draft of a document related to a proposal or contract.

[0801] An "approver" is a person who has the authority to review and approve the generated draft document.

[0802] This invention is a system that reduces the burden on salespeople and efficiently generates high-quality proposals and contract drafts. This system operates through the cooperation of three parties: a server, a terminal, and a user.

[0803] First, the server maintains a database containing past information and retrieves it as needed. The database stores information about past proposals and contracts. The server uses a generative AI model (for example, OpenAI's GPT-4) that uses machine learning to automatically generate draft text for proposals and contracts.

[0804] Users input basic information about proposals and contracts via the terminal. Specifically, they input information such as the customer name, proposal details, and contract terms (e.g., initial fees, monthly fees). For example, when a user creates a proposal for a new customer, they would input the following information into the terminal:

[0805] Customer name: XYZ Co., Ltd.

[0806] Proposal: Proposal for a new marketing strategy

[0807] Contract terms: Initial fee 1 million yen, monthly fee 100,000 yen

[0808] The server generates prompt messages based on the information entered by the user. These prompt messages include instructions for the generating AI model. For example, it generates prompt messages like the following:

[0809] "Please prepare a proposal for a new marketing strategy for XYZ Corporation. The proposal should include contract terms of an initial fee of 1 million yen and a monthly fee of 100,000 yen."

[0810] The generative AI model receives a prompt and generates draft proposals and contract documents. The goal is to produce draft documents of the same quality as those manually created by sales representatives. For example, it can generate draft documents like the following:

[0811] "XYZ Corporation"

[0812] Thank you for reviewing our proposal for a new marketing strategy. The proposal details and contract terms are outlined below.

[0813] Proposal details:

[0814] Introducing a new marketing strategy

[0815] Initial cost: 1 million yen

[0816] Monthly fee: 100,000 yen

[0817] Thank you for your consideration.

[0818] Sincerely"

[0819] The generated draft text is sent from the server to the terminal and presented to the user interactively. The user can review the displayed draft text and make revisions as needed. This reduces the workload on sales staff and enables the efficient generation of high-quality draft text.

[0820] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0821] Step 1:

[0822] Users input basic information about proposals and contracts via the terminal. Specifically, they input information such as the customer name, proposal details, and contract terms (e.g., initial fees, monthly fees). For example, when a user creates a proposal for a new customer, they would input the following information into the terminal:

[0823] Customer name: XYZ Co., Ltd.

[0824] Proposal: Proposal for a new marketing strategy

[0825] Contract terms: Initial fee 1 million yen, monthly fee 100,000 yen

[0826] Input: Customer name, proposal details, contract terms

[0827] Output: Basic information entered by the user

[0828] Step 2:

[0829] The server generates prompt messages based on the information entered by the user. These prompt messages include instructions for the AI ​​model that generates them. The server analyzes the input information and creates appropriate prompt messages. For example, it might generate a prompt message such as, "Please create a proposal for a new marketing strategy for XYZ Corporation. The proposal should include contract terms of an initial fee of 1 million yen and a monthly fee of 100,000 yen."

[0830] Input: Basic information entered by the user

[0831] Output: Generated prompt message

[0832] Step 3:

[0833] The server sends a prompt to a generative AI model (e.g., OpenAI's GPT-4) and generates a draft document. The generative AI model then creates a draft proposal or contract document based on the prompt. For example, it might generate a draft document like this:

[0834] "XYZ Corporation"

[0835] Thank you for reviewing our proposal for a new marketing strategy. The proposal details and contract terms are outlined below.

[0836] Proposal details:

[0837] Introducing a new marketing strategy

[0838] Initial cost: 1 million yen

[0839] Monthly fee: 100,000 yen

[0840] Thank you for your consideration.

[0841] Sincerely"

[0842] Input: Generated prompt message

[0843] Output: Generated draft text

[0844] Step 4:

[0845] The server sends the generated draft text to the terminal and presents it to the user interactively. The user can review the displayed draft text and make corrections as needed. The user checks the draft text on the terminal screen, verifies that there are no errors, and makes corrections as necessary.

[0846] Input: Generated draft text

[0847] Output: Draft text presented to the user

[0848] In this way, the server, terminals, and users work together, reducing the workload on sales personnel and enabling the efficient generation of high-quality draft documents.

[0849] (Application Example 2)

[0850] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[0851] A problem is that salespeople spend a great deal of time and effort creating proposals and contracts. Furthermore, communication with customers and managing the progress of sales activities are also time-consuming, making efficient sales activities difficult. There is a need to solve these problems, reduce the burden on salespeople, and achieve more efficient sales activities.

[0852] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0853] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract documents using machine learning, a means for interactively presenting the generated draft documents, a means for automating communication with customers, a means for automatically generating proposals and contracts, a means for managing and analyzing customer information, and a means for managing the progress of sales activities. This significantly reduces the workload on salespeople and enables efficient sales activities.

[0854] "Past information" refers to past data and records related to sales activities.

[0855] A "database" refers to an information management system that systematically stores information and allows it to be searched and retrieved as needed.

[0856] "Machine learning" refers to the technology that allows computers to learn patterns from data and perform predictions and classifications.

[0857] "Draft text for proposals and contracts" refers to draft proposals and contracts that are presented to customers during sales activities.

[0858] "Artificial intelligence" refers to computer systems that mimic human intelligence and perform learning and reasoning.

[0859] "Interactive presentation methods" refer to methods in which users and systems exchange information in a dialogue format.

[0860] "Methods for automating communication with customers" refers to methods for automating communication with customers, such as email and chat.

[0861] "Methods for automatically generating proposals and contracts" refers to methods for automatically creating proposals and contracts necessary for sales activities.

[0862] "Means for managing and analyzing customer information" refers to methods for organizing and analyzing data related to customers.

[0863] "Means of managing the progress of sales activities" refers to methods of tracking and managing the progress of the sales process.

[0864] The system for implementing this invention is configured as follows: The server retrieves information from a database containing historical data and automatically generates draft proposal and contract documents using machine learning. The generated draft documents are presented to the user interactively. It also has the functionality to automate communication with customers and automatically generate proposals and contracts. Furthermore, it includes functions to manage and analyze customer information and manage the progress of sales activities.

[0865] Hardware and software to be used

[0866] Hardware: Servers, user terminals (smartphones, personal computers, etc.)

[0867] Software: APIs, database management systems (e.g., MySQL)

[0868] Data processing and data calculation

[0869] 1. Retrieving information from the database: The server retrieves data related to past sales activities from the database. This data includes past proposals, contracts, and communication history with customers.

[0870] 2. Text draft generation using machine learning: Based on the acquired data, a machine learning model (e.g., OpenAI's GPT-3) is used to generate draft text for proposals and contracts.

[0871] 3. Interactive Presentation: The generated draft text is presented interactively on the user's terminal. The user can review the draft text and make modifications as needed.

[0872] 4. Automated communication with customers: The server automates email and chat communication with customers based on the generated draft messages.

[0873] 5. Automatic generation of proposals and contracts: The server automatically generates formal proposals and contracts based on the generated draft documents.

[0874] 6. Customer Information Management and Analysis: The server organizes and analyzes customer information to improve the efficiency of sales activities.

[0875] 7. Sales activity progress management: The server tracks and manages the progress of the sales process.

[0876] Specific example

[0877] For example, if the customer name is "ABC Corporation" and the product details are "latest AI-equipped smartphone," the following prompt message will be sent to the AI ​​model.

[0878] Example of a prompt

[0879] Customer Name: ABC Co., Ltd.

[0880] Product Details: Latest AI-powered smartphone

[0881] Please prepare your proposal based on the following information.

[0882] Based on this prompt, the AI ​​model generates the text of the proposal, reducing the workload on salespeople. The generated proposal is presented interactively on the user's terminal, allowing the user to review the content and make revisions as needed.

[0883] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0884] Step 1:

[0885] The server retrieves data on past sales activities from a database. Inputs include past proposals, contracts, and customer communication history stored in the database. Output is a collection of the retrieved data, which is used for subsequent processing.

[0886] Step 2:

[0887] The server uses a machine learning model (e.g., OpenAI's GPT-3) to generate draft proposal and contract documents based on the acquired data. The input is the data acquired in step 1. The output is the generated draft documents. Specifically, the data is converted into prompt sentences and sent to the generation AI model.

[0888] Step 3:

[0889] The server interactively presents the generated draft text to the user terminal. The input is the draft text generated in step 2. The output is the draft text displayed on the user terminal. Specifically, the server displays the draft text in the user interface, allowing the user to review and modify it.

[0890] Step 4:

[0891] The server automates email and chat communication with customers based on the generated draft text. The input is the draft text generated in step 2. The output is the email or chat message sent to the customer. Specifically, it converts the draft text into email or chat format and sends it automatically.

[0892] Step 5:

[0893] The server automatically generates formal proposals and contracts based on the generated draft documents. The input is the draft documents generated in step 2. The output is the formal proposal or contract document. Specifically, it formats the draft documents into a formal document format and converts them into a format that can be saved and printed.

[0894] Step 6:

[0895] The server organizes and analyzes customer information. The input is customer information stored in a database. The output is the analysis results and the organized customer information. Specifically, it organizes customer information by category and applies analytical algorithms to extract useful insights.

[0896] Step 7:

[0897] The server tracks and manages the progress of the sales process. Inputs are data and progress information related to sales activities. Outputs are progress reports. Specifically, it tracks each stage of sales activities and generates dashboards to visualize progress.

[0898] (Example 3)

[0899] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0900] Conventional draft document generation systems failed to adequately consider past approval history and approver tendencies, making it difficult to efficiently create draft documents that approved by approvers. Furthermore, draft document generation that reduced the workload on sales staff was insufficient. This resulted in delays in the approval process and decreased operational efficiency.

[0901] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes means for the user to input a prompt sentence, means for the terminal to send the prompt sentence to the server, means for the server to obtain past approval history and approver tendencies from a database, means for generating a draft text using a generation AI model, means for sending the generated draft text to the terminal, means for the terminal to display the draft text to the user, and means for the user to check the draft text and modify it as necessary. This makes it possible to efficiently generate draft text that takes into account past approval history and approver tendencies, and to provide draft text that is likely to be accepted by the approver. In addition, it is possible to reduce the burden on sales personnel and improve operational efficiency.

[0902] A "database" is a system that stores historical information and retrieves it as needed.

[0903] "Artificial intelligence" refers to a program or system that uses machine learning to automatically generate draft text.

[0904] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing the user to review and modify the text.

[0905] A "prompt message" is text that the user enters, containing instructions and conditions for generating a draft of the text.

[0906] A "terminal" is a device used by a user to input prompt messages and communicate with a server.

[0907] A "server" is a computer system that retrieves information from a database, generates draft text using a generative AI model, and sends it to a terminal.

[0908] A "generative AI model" is an artificial intelligence model that learns from past data and generates draft text.

[0909] "Approval history" refers to data that includes previously approved documents and feedback from approvers.

[0910] "Approver tendencies" refer to patterns based on the types of documents approvers have approved in the past or the types of feedback they have provided.

[0911] A "draft document" is a draft of a document generated by a generative AI model that is likely to be accepted by the approver.

[0912] A "user" is someone who enters prompt text and reviews and modifies the generated draft text.

[0913] This invention relates to a system for generating draft documents that take into account past approval history and the tendencies of approvers. Specific embodiments of this system are described below.

[0914] First, the user enters a prompt using the terminal. The prompt is text containing instructions and conditions for generating a draft document. For example, it might say, "Please create a document requesting budget approval for a new project. Based on past approval history, approvers tend to prioritize specific figures and clear objectives."

[0915] The terminal sends the entered prompt message to the server. This transmission uses communication protocols such as HTTP POST requests.

[0916] The server accesses the database to retrieve past approval history and approver trends. The database stores previously approved documents and approver feedback. For example, it might execute a query like "SELECT FROM approval_history WHERE approver_id = '12345'" using SQL.

[0917] Next, the server runs a generative AI model based on the acquired data to generate draft text. For example, OpenAI's GPT-4 is used as the generative AI model. The generative AI model has learned from past data and generates text that is likely to be accepted by the approver.

[0918] The generated draft text is sent from the server to the terminal. Communication protocols such as HTTP responses are used for transmission.

[0919] The terminal displays the received draft text to the user. The user interface includes a text area and display widgets. The user can review the displayed draft text and make revisions as needed. The revised text is used as the final approval request.

[0920] This system makes it possible to efficiently generate draft documents that take into account past approval history and approver tendencies, and to provide draft documents that are more likely to be accepted by approvers. Furthermore, it can reduce the workload on sales staff and improve operational efficiency. The flow of the specific processing in Example 3 will be explained using Figure 15.

[0921] Step 1:

[0922] The user enters a prompt message.

[0923] The user enters a prompt in the terminal's input field. For example, they might enter, "Please draft a document requesting budget approval for a new project. Based on past approval history, approvers tend to prioritize specific figures and clear objectives."

[0924] Input: Prompt message

[0925] Output: Input prompt message

[0926] Step 2:

[0927] The terminal sends a prompt message to the server.

[0928] The terminal sends the entered prompt text to the server as an HTTP POST request.

[0929] Input: Prompt message

[0930] Output: Prompt message sent to the server

[0931] Step 3:

[0932] The server retrieves past approval history and approver trends from the database.

[0933] The server accesses the database to retrieve past approval history and approver trends. For example, it might use an SQL query like "SELECT FROM approval_history WHERE approver_id = '12345'".

[0934] Input: Prompt message, database

[0935] Output: Past approval history and approver trends

[0936] Step 4:

[0937] The server generates draft text using an AI model.

[0938] The server runs a generative AI model (for example, OpenAI's GPT-4) based on the acquired data to generate draft text. The generative AI model has learned from past data and generates text that is likely to be accepted by the approver.

[0939] Input: Past approval history and approver trends

[0940] Output: Generated draft text

[0941] Step 5:

[0942] The server sends the generated draft text to the terminal.

[0943] The server sends the generated draft text to the terminal as an HTTP response.

[0944] Input: Generated draft text

[0945] Output: Draft text sent to the terminal

[0946] Step 6:

[0947] The device displays a draft text to the user.

[0948] The device displays the received draft text in a text area. The user can then review the displayed text.

[0949] Input: Draft text sent to the terminal

[0950] Output: Draft text displayed to the user

[0951] Step 7:

[0952] The user reviews the draft text and makes revisions as needed.

[0953] The user reviews the displayed draft text and makes any necessary revisions. For example, they might change "Personnel costs: 5 million yen" to "Personnel costs: 6 million yen." The revised text will then be used as the final approval request.

[0954] Input: Draft text displayed to the user

[0955] Output: Revised draft text

[0956] (Application Example 3)

[0957] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[0958] Traditional electronic payment systems had the problem of requiring a great deal of time and effort to create documents that would satisfy approvers. Furthermore, because the documents were created without considering the approver's preferences, the approval process often suffered from delays. This resulted in inefficient expense reimbursement and large-scale transaction approval processes within companies, leading to a decline in overall business efficiency.

[0959] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[0960] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning; a server that interactively presents the generated draft texts; and a server that learns the approver's tendencies and generates draft texts that are likely to be accepted by the approver. This makes it possible to quickly generate draft texts that are easily accepted by the approver and streamline the electronic payment approval process.

[0961] "Past information" refers to data and history that the system has collected in the past.

[0962] A "database" is a collection of information that systematically stores past information and makes it searchable and retrievalable as needed.

[0963] "Machine learning" is a technology in which computers analyze large amounts of data and automatically learn patterns and rules.

[0964] "Draft text for proposals and contracts" refers to a draft of the text included in proposals and contracts.

[0965] "Artificial intelligence" is a system that uses technologies such as machine learning to mimic human intelligence and automatically perform specific tasks.

[0966] "Interactive presentation methods" refer to interfaces that allow users and systems to exchange information in a conversational format.

[0967] "Approver tendencies" refer to data that shows what criteria or patterns a particular approver has used to make approvals in the past.

[0968] A "draft document that is likely to satisfy the approver" refers to a draft document that is likely to be approved, created considering the approver's past tendencies.

[0969] As an example of how to implement this invention, an electronic payment approval assistant system will be described. This system operates in cooperation with three parties: a server, a terminal, and a user.

[0970] First, the server maintains a database containing historical information, including past approval history and data on approver tendencies. The server retrieves information from this database and uses machine learning to automatically generate draft text for proposals and contracts. Specifically, the server uses Python, Pandas, and OpenAI APIs to analyze past approval history and learn approver tendencies.

[0971] Next, the generated draft text is presented interactively on the device. This device is a user-operated device such as a smartphone or tablet, through which the user can review the generated draft text and make modifications as needed. The interactive interface is designed for easy user operation.

[0972] Furthermore, the server learns the approver's tendencies and generates draft text that is more likely to satisfy the approver. In this process, a generative AI model is used to generate draft text based on features extracted from past approval history. Examples of specific prompt texts are as follows:

[0973] Example of a prompt:

[0974] Based on the approver's tendencies, please generate the following draft text.

[0975] This business trip involves an important meeting with a new client and requires prompt approval.

[0976] Approver tendencies: {Characteristics extracted from the approver's past approval history}

[0977] This system enables the rapid generation of draft documents that are easily accepted by approvers, thereby streamlining the electronic payment approval process. For example, in the expense reimbursement process within a company, when an employee applies for travel expense reimbursement and requires supervisor approval, this system can automatically generate a draft document that is easily accepted by the supervisor, leading to quicker approval.

[0978] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[0979] Step 1:

[0980] The server retrieves approval history data from a database that stores past information.

[0981] Input: Past approval history data stored in the database

[0982] Output: Acquired approval history data

[0983] Specific operation: The server uses SQL queries to extract past approval history data from the database.

[0984] Step 2:

[0985] The server inputs the acquired approval history data into a machine learning model to learn the tendencies of the approvers.

[0986] Input: Acquired approval history data

[0987] Output: Learning results regarding approver tendencies

[0988] Specific operation: The server preprocesses the data using publicly available libraries and inputs it into a machine learning model. The model learns past approval patterns and extracts the tendencies of approvers.

[0989] Step 3:

[0990] The server inputs prompt sentences into the AI ​​model based on the learning results and generates suggested sentences.

[0991] Input: Learning results regarding approver tendencies, prompt text

[0992] Output: Generated draft text

[0993] Specific operation: The server uses the OpenAI API to input prompt text and training results into the AI ​​model, which then generates suggested text.

[0994] Step 4:

[0995] The server sends the generated draft text to the terminal and presents it interactively.

[0996] Input: Generated draft text

[0997] Output: Draft text displayed on the terminal

[0998] Specific operation: The server sends the generated draft text to the terminal via an HTTP request, and the terminal displays it in the user interface.

[0999] Step 5:

[1000] The user reviews the draft text generated through their device and makes revisions as needed.

[1001] Input: Draft text displayed on the terminal

[1002] Output: Revised draft text

[1003] Specific operation: The user reviews the draft text using the terminal interface and makes revisions in a text editor as needed.

[1004] Step 6:

[1005] The server saves the revised draft as the final approved document.

[1006] Input: Revised draft text

[1007] Output: Saved final approval document

[1008] Specific operation: The server saves the revised draft text to the database and completes the approval process.

[1009] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1010] "Example of form 1"

[1011] In one embodiment of the present invention, the AI ​​includes an emotion engine that recognizes the user's emotions. This emotion engine estimates the user's emotions from the tone of their voice, facial expressions, and word choice. Specifically, it generates positive suggestions when the user is happy, and suggestions that include apologies when the user is angry.

[1012] "Example of form 2"

[1013] Furthermore, the emotion engine adjusts the draft text based on the user's emotions. Specifically, it generates draft text that includes positive language when the user is happy and apologies when the user is angry. For example, when the user is happy, it generates text such as "This proposal will lead your business to even greater success," and when the user is angry, it generates text such as "We deeply apologize that our service did not meet your expectations."

[1014] "Example of form 3"

[1015] Furthermore, the emotion engine automatically generates draft text for proposals and contracts based on the user's emotions. Specifically, it generates positive proposals when the user is happy and proposals that include apologies when the user is angry. For example, when the user is happy, it generates a proposal such as "This proposal will lead your business to even greater success," and when the user is angry, it generates a proposal such as "We deeply apologize that our service did not meet your expectations."

[1016] The following describes the processing flow for each example of the form.

[1017] "Example of form 1"

[1018] Step 1: The AI ​​activates an emotion engine that estimates the user's emotions based on their tone of voice, facial expressions, and word choice.

[1019] Step 2: The emotion engine recognizes the user's emotions and determines the content of the suggestions based on those emotions.

[1020] Step 3: Based on the results of the emotion engine, the AI ​​generates and presents suggestions to the user. (Example 2)

[1021] Step 1: The AI ​​activates an emotion engine that recognizes the user's emotions.

[1022] Step 2: The emotion engine recognizes the user's emotions and determines the content of the draft text based on those emotions.

[1023] Step 3: Based on the results of the emotion engine, the AI ​​generates and presents text drafts to the user.

[1024] "Example of form 3"

[1025] Step 1: The AI ​​activates an emotion engine that recognizes the user's emotions.

[1026] Step 2: The emotion engine recognizes the user's emotions and determines the content of the proposal and contract based on those emotions.

[1027] Step 3: Based on the results of the emotion engine, the AI ​​generates proposal and contract sections and presents them to the user.

[1028] (Example 1)

[1029] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1030] In traditional sales activities, salespeople had to manually refer to past contract and proposal information to create new proposals. This process was time-consuming and labor-intensive, increasing the workload on salespeople. Furthermore, it was difficult to make appropriate proposals that resonated with the user's emotions, making it challenging to improve customer satisfaction.

[1031] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1032] In this invention, the server includes means for acquiring information from a database storing past information, means for training a machine learning model using the acquired information, means for automatically generating draft text for proposals and contracts using the trained machine learning model, means for presenting the generated draft text through an interactive interface, means including an emotion engine that recognizes the user's emotions, and means for adjusting the draft text based on information obtained from the emotion engine. This reduces the burden on salespeople and makes it possible to provide optimal proposals that are tailored to the user's emotions.

[1033] A "database" is an information management system that stores historical information and allows for the efficient retrieval of necessary information.

[1034] A "machine learning model" is an algorithm that learns patterns and rules based on data and uses them to make predictions and classifications on new data.

[1035] "Training" is the process of improving the performance of a machine learning model by having it learn from a large amount of data.

[1036] A "draft document" is a draft of a document that outlines the contents of a proposal or contract.

[1037] An "interactive interface" is a user interface that allows the user and the system to communicate in both directions.

[1038] An "emotion engine" is a system that estimates a user's emotions based on their tone of voice, facial expressions, and word choice.

[1039] "Adjustment" means changing the content or settings according to specific conditions or circumstances.

[1040] Modes for carrying out the invention

[1041] This invention is a system that retrieves information from a database containing historical data and automatically generates draft text for proposals and contracts using machine learning. A specific embodiment of this system is described below.

[1042] Retrieve information from the database.

[1043] The server retrieves the necessary information from a database that stores information on past contracts and proposals made by corporate sales representatives. This database includes contract details, proposal details, and customer responses. The server uses SQL queries to extract past contract information from the database. For example, it might execute a query like "SELECT FROM contracts WHERE date > '2020-01-01'".

[1044] Training machine learning models

[1045] The server uses the acquired information to train a machine learning model. This model has the ability to automatically generate draft text for proposals and contracts using techniques such as deep learning and reinforcement learning. Specifically, it uses a framework such as TensorFlow. The server builds a deep learning model using TensorFlow and trains the model using historical contract data. For example, it executes code like "model.fit(training_data, epochs=50)".

[1046] Generating a draft document

[1047] The server uses a trained machine learning model to generate draft text for the proposal and contract sections. This draft text is designed to include optimal content based on historical data. The server prompts the generation AI model with a text prompt and generates the draft text. For example, it executes code like "generate_text('Generate a new proposal text')".

[1048] The interface presents draft text.

[1049] The terminal presents the generated message draft to the salesperson through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant. The terminal displays the generated message draft in the chatbot's UI. For example, it executes code such as "chatbot.display_message(generated_text)".

[1050] User emotion recognition by an emotion engine

[1051] The device recognizes the user's emotions using an emotion engine. This emotion engine estimates emotions from the user's tone of voice, facial expressions, and word choice. The device uses OpenCV to analyze the user's facial expressions and estimate emotions. For example, it executes code like "emotion = emotion_recognition.detect_face_expression(user_image)".

[1052] Adjusting the wording to suit the emotions.

[1053] The server generates text suggestions tailored to the user's emotions based on information obtained from the emotion engine. When the user is happy, it generates positive suggestions; when the user is angry, it generates suggestions that include apologies. Based on the output of the emotion engine, the server inputs appropriate prompt sentences to the generation AI model. For example, it might execute code like "generate_text('Generate a suggestion sentence for when the user is angry')".

[1054] Specific example

[1055] Example 1: Generating a proposal

[1056] Based on the user's past contract information, the following proposal statement is generated.

[1057] Example of a prompt:

[1058] "Please generate a new proposal based on past contract information."

[1059] Example of a generated proposal:

[1060] "Customer, based on your past contract, we would like to offer you a new proposal with similar terms. We appreciate your consideration."

[1061] Example 2: Adjusting the wording to suit the emotions.

[1062] If the user is angry, the emotion engine recognizes that emotion and generates a draft message that includes an apology.

[1063] Example of a prompt:

[1064] "Please generate a suggestion message for when the user is angry."

[1065] Example of a generated proposal:

[1066] "We sincerely apologize for the inconvenience caused, customer. In this proposal, we have addressed past issues and will offer you better terms."

[1067] In this way, the system utilizes past data to automatically generate optimal suggestions tailored to the user's emotions.

[1068] The flow of the specific processing in Example 1 will be explained using Figure 17.

[1069] Step 1:

[1070] The server retrieves information from a database that stores historical data.

[1071] Input: Database query (Example: "SELECT FROM contracts WHERE date > '2020-01-01'")

[1072] Data processing: Execute SQL queries to extract past contract information.

[1073] Output: Dataset of acquired contract information

[1074] Specific operation: The server extracts historical contract information from the database using SQL queries. For example, it executes a query like "SELECT FROM contracts WHERE date > '2020-01-01'".

[1075] Step 2:

[1076] The server uses the acquired information to train a machine learning model.

[1077] Input: Header of acquired contract information

[1078] Data processing: Preprocess the dataset and convert it into a format suitable for machine learning models.

[1079] Output: Trained machine learning model

[1080] Specific operation: The server uses TensorFlow to build a deep learning model and trains the model using historical contract data. For example, it executes code like "model.fit(training_data, epochs=50)".

[1081] Step 3:

[1082] The server generates draft text for the proposal and contract sections using a trained machine learning model.

[1083] Input: A trained machine learning model, a prompt sentence (e.g., "Generate a new suggestion sentence")

[1084] Data processing: Input prompt messages into the model and generate suggested text options.

[1085] Output: Generated draft text

[1086] Specific operation: The server takes a prompt message as input to the generation AI model and generates a text proposal. For example, it executes code like "generate_text('Please generate a new proposal text')".

[1087] Step 4:

[1088] The terminal presents the generated draft text to the salesperson through an interactive interface.

[1089] Input: Generated draft text

[1090] Data processing: Convert the draft text into a format suitable for an interactive interface.

[1091] Output: Draft text to be presented to the business operator

[1092] Specific operation: The device displays the generated message draft in the chatbot's UI. For example, it executes code like "chatbot.display_message(generated_text)".

[1093] Step 5:

[1094] The device uses an emotion engine to recognize the user's emotions.

[1095] Input: User's tone of voice, facial expression, and word choice

[1096] Data processing: Analyze user emotions using an emotion engine.

[1097] Output: Estimated user sentiment

[1098] Specific operation: The device uses OpenCV to analyze the user's facial expressions and estimate their emotions. For example, it executes code like "emotion = emotion_recognition.detect_face_expression(user_image)".

[1099] Step 6:

[1100] The server generates text suggestions tailored to the user's emotions, based on information obtained from the emotion engine.

[1101] Input: Estimated user sentiment, trained machine learning model

[1102] Data processing: Generate prompt sentences that respond to emotions, input them into the model, and generate suggested sentences.

[1103] Output: Suggested text based on emotions

[1104] Specific operation: Based on the output of the emotion engine, the server inputs appropriate prompt text to the generative AI model. For example, it executes code such as "generate_text('Generate a suggestion text for when the user is angry')".

[1105] (Application Example 1)

[1106] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[1107] Conventional sales support systems made it difficult for salespeople to quickly refer to past proposals and contract information when interacting with customers, and to make optimal proposals. Furthermore, it was difficult to recognize customer emotions in real time and respond accordingly. In addition, in physical stores, while store staff are required to make optimal proposals based on past sales data and proposal content, there was a lack of efficient means to do so. To solve these problems, a system is needed that recognizes customer emotions and generates proposal text that responds accordingly.

[1108] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1109] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, an emotion engine that recognizes the user's emotions, a means for generating proposal texts that correspond to the user's emotions, and a means installed in smart glasses to support customer service. This enables salespeople and store clerks to quickly make optimal proposals based on past proposals and contract information when interacting with customers, and to make proposals that correspond to the customer's emotions in real time.

[1110] "Past information" refers to data that includes information on contracts and proposals made by corporate sales representatives in the past.

[1111] A "database" is a system that stores historical information and allows you to retrieve that information as needed.

[1112] "Artificial intelligence" is a technology that uses machine learning to automatically generate draft text for proposals and contracts.

[1113] "Interactive presentation methods" refer to methods of presenting generated draft text to salespeople through chatbots, voice assistants, or similar means.

[1114] An "emotion engine" is a technology that estimates emotions from a user's tone of voice, facial expressions, and word choice.

[1115] "Means for generating suggestion text that responds to user emotions" refers to a method for generating appropriate suggestion text based on emotions estimated by an emotion engine.

[1116] "Smart glasses" are wearable devices equipped with displays and cameras that can display information and capture video.

[1117] "Means of supporting customer service" refers to a method of using smart glasses to display optimal suggestion text when store employees interact with customers.

[1118] The system for carrying out this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, an emotion engine that recognizes the user's emotions, means for generating proposal text that corresponds to the user's emotions, and means installed in smart glasses to support customer service.

[1119] The server retrieves information from a database containing historical data and automatically generates draft proposal and contract texts using machine learning. Specifically, it uses deep learning models and reinforcement learning models trained with TensorFlow. The generated draft texts are presented to sales representatives through interactive interfaces such as chatbots and voice assistants.

[1120] The device uses smart glasses to capture the customer's facial expressions and voice tone, and an emotion engine recognizes the user's emotions. The emotion engine uses an emotion recognition model based on TensorFlow to estimate emotions from the user's voice tone, facial expressions, and word choice. Based on the estimated emotions, a generative AI model is used to generate the most appropriate suggested text. The generative AI model uses a natural language generation model such as GPT-2.

[1121] When a user wears smart glasses and interacts with customers, the smart glasses' display shows optimal proposal text based on past sales data and previous suggestions. This allows the user to make suggestions tailored to the customer's emotions in real time.

[1122] For example, if a customer is smiling while looking at a product, the emotion recognition model will determine that the customer is "happy" and generate a suggestion message such as, "This product is especially popular. We are currently offering a special discount, so please give it a try." Conversely, if a customer has a confused expression, the emotion recognition model will determine that the customer is "confused" and generate a suggestion message such as, "If there is anything we can help you with, please let us know."

[1123] Examples of prompt messages include the following:

[1124] "The optimal proposal text for when the customer is happy: Past proposal data"

[1125] "The best proposal text when a customer is confused: Past proposal data"

[1126] This allows salespeople and staff to quickly provide optimal proposals based on past proposals and contract information when interacting with customers, and to make proposals that are tailored to the customer's emotions in real time.

[1127] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[1128] Step 1:

[1129] The server retrieves information from a database that stores historical data.

[1130] Input: Past proposals and contract information stored in the database

[1131] Output: Retrieved past proposals and contract information

[1132] Specific operation: The server uses SQL queries to extract the necessary information from the database and load it into memory.

[1133] Step 2:

[1134] The server uses machine learning to automatically generate draft text for proposals and contracts based on the acquired information.

[1135] Input: Past proposals and contract information

[1136] Output: Generated proposal / contract draft

[1137] Specific operation: The server inputs data into deep learning models and reinforcement learning models trained using TensorFlow, and generates optimal text suggestions.

[1138] Step 3:

[1139] The server presents the generated draft text to the salesperson through an interactive interface.

[1140] Input: Generated proposal / contract draft

[1141] Output: Draft text presented to the salesperson

[1142] Specific operation: The server sends the generated draft text to the salesperson's terminal via a chatbot or voice assistant and displays it.

[1143] Step 4:

[1144] The device uses smart glasses to capture the customer's facial expressions and voice tone.

[1145] Input: Customer's facial expressions and tone of voice

[1146] Output: Data of captured customer facial expressions and voice tone.

[1147] Specific operation: The camera and microphone built into the smart glasses on the device are used to capture the customer's facial expressions and voice in real time.

[1148] Step 5:

[1149] The device inputs the captured data into an emotion engine to recognize the user's emotions.

[1150] Input: Data of captured customer facial expressions and voice tone.

[1151] Output: Estimated user sentiment

[1152] Specific operation: The device inputs data into an emotion recognition model using TensorFlow and estimates the user's emotions.

[1153] Step 6:

[1154] The server generates the optimal suggested text using a generative AI model based on the estimated emotions.

[1155] Input: Estimated user sentiment, past suggestion data

[1156] Output: Suggestion text tailored to your emotions

[1157] Specific operation: The server inputs a prompt sentence into a natural language generation model such as GPT-2 and generates the most suitable suggested sentence.

[1158] Step 7:

[1159] The device displays the generated proposal text on the smart glasses' display.

[1160] Input: Generated proposal text

[1161] Output: Proposal text displayed on the smart glasses' screen

[1162] Specific operation: The device sends the generated proposal text to the smart glasses' display and displays it.

[1163] This allows users to make suggestions in real time that are tailored to the customer's emotions.

[1164] (Example 2)

[1165] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1166] In traditional sales activities, salespeople spent a great deal of time and effort drafting proposals and contract documents. Furthermore, it was difficult to generate appropriate wording that resonated with the user's emotions, making improving customer satisfaction a challenge.

[1167] The identification processing performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for acquiring information from a database storing past information and automatically generating draft text for the proposal and contract sections using machine learning, means for presenting the generated draft text interactively, an emotion engine for analyzing the user's emotions, and means for adjusting the draft text based on the user's emotions. This reduces the burden on salespeople and enables the generation of appropriate draft text that responds to the user's emotions.

[1168] A "database" is a system that stores historical information and retrieves it as needed.

[1169] "Artificial intelligence" is a technology that uses machine learning to automatically generate draft text for proposals and contracts.

[1170] "Interactive presentation methods" refer to methods of interactively displaying generated draft text to the user.

[1171] An "emotion engine" is a technology that analyzes a user's emotions and identifies their emotional state.

[1172] "Methods for adjusting draft text" refers to methods for appropriately modifying draft text that has been generated based on user sentiment.

[1173] Modes for carrying out the invention

[1174] This invention relates to a system that reduces the burden on business operators and generates appropriate text drafts that respond to the user's emotions. Specific embodiments of this system are described below.

[1175] System Configuration

[1176] This system consists of the following main components:

[1177] 1. Database: A system for storing historical information and retrieving it as needed.

[1178] 2. Artificial Intelligence (AI): This is a technology that uses machine learning to automatically generate draft text for proposals and contracts. Specifically, it uses a generative AI model (for example, OpenAI's GPT-4).

[1179] 3. Interactive presentation methods: These are methods of interactively displaying the generated draft text to the user. For example, using a web interface or mobile application.

[1180] 4. Emotion Engine: This is a technology for analyzing a user's emotions and identifying their emotional state. Specifically, it involves using an emotion engine.

[1181] 5. Means for adjusting the draft text: This refers to methods for appropriately modifying the draft text that was generated based on the user's emotions.

[1182] System operation

[1183] The server collects basic information such as the proposed content and contract terms entered by the user on the terminal. For example, if a user proposes a new marketing strategy, they would enter "New Marketing Strategy" as the proposed content and "Annual Contract" as the contract terms.

[1184] Next, the server generates prompt messages for the AI ​​model based on the data entered by the user. For example, it generates prompt messages like the following:

[1185] Proposal: New marketing strategy

[1186] Contract terms: Annual contract

[1187] Please generate a draft of the text.

[1188] The server sends the generated prompt text to the AI ​​model, which then generates suggested text based on the prompt text. For example, it might generate a suggested text such as, "This new marketing strategy will lead your business to even greater success."

[1189] The server then collects the user's emotional data and sends it to the emotion engine. The emotion engine analyzes the user's emotions and identifies emotional states such as happy or angry.

[1190] The server adjusts the generated text based on the analysis results of the emotion engine. For example, if the user is happy, the text is adjusted to include positive language. If the user is angry, the text is adjusted to include apologies.

[1191] Finally, the server sends the final draft text to the terminal, where the user can review and revise it as needed.

[1192] Specific example

[1193] As a concrete example, consider the following scenario:

[1194] Scenario 1: When the user is happy

[1195] Input data: Proposal content "New marketing strategy", Contract terms "Annual contract"

[1196] Output from the generating AI model: Draft text: "This new marketing strategy will lead your business to even greater success."

[1197] Emotion engine adjustment: Since the user is happy, it generates text suggestions that include positive language.

[1198] Scenario 2: When the user is angry

[1199] Input data: Proposal content "Service improvement plan", Contract terms "Monthly contract"

[1200] Output from the generated AI model: Draft text: "We sincerely apologize that our service did not meet your expectations."

[1201] Emotion engine adjustment: Since the user is angry, generate a draft message that includes an apology.

[1202] In this way, the system operates by having the server, terminal, and user work together to generate draft documents that reduce the workload on sales representatives.

[1203] The flow of the specific processing in Example 2 will be explained using Figure 19.

[1204] Step 1: Collecting user input data

[1205] The user enters basic information such as the proposal details and contract terms into the terminal. For example, if the user proposes a new marketing strategy, they would enter "New Marketing Strategy" for the proposal details and "Annual Contract" for the contract terms. The entered data is then sent to the server.

[1206] Step 2: Generating prompts for the generative AI model

[1207] The server generates prompts for the AI ​​model based on the data entered by the user. For example, it generates prompts like the following:

[1208] Proposal: New marketing strategy

[1209] Contract terms: Annual contract

[1210] Please generate a draft of the text.

[1211] This prompt message is sent to the generating AI model.

[1212] Step 3: Generating draft text using a generative AI model

[1213] The server sends the generated prompt text to a generative AI model (for example, OpenAI's GPT-4). The generative AI model generates a text draft based on the prompt text. For example, it might generate a text draft such as, "This new marketing strategy will lead your business to even greater success." The generated text draft is then returned to the server.

[1214] Step 4: User sentiment analysis using an emotion engine

[1215] The server collects user emotion data and sends it to the emotion engine. The emotion engine analyzes the user's emotions and identifies emotional states such as happiness or anger. The analysis results are then returned to the server.

[1216] Step 5: Emotion-based adjustments to the draft text

[1217] The server adjusts the generated text drafts based on the analysis results of the emotion engine. For example, if the user is happy, the drafts are adjusted to include positive language. If the user is angry, the drafts are adjusted to include apologies. The adjusted drafts are saved on the server.

[1218] Step 6: Output of the final draft

[1219] The server sends the final draft document to the terminal. The user can review the final draft document on the terminal and make revisions as needed. After the user's review, the final draft document is used as the official proposal or contract document.

[1220] (Application Example 2)

[1221] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1222] In traditional sales activities, salespeople spend a significant amount of time and effort drafting proposals and contract documents. Furthermore, they need to respond appropriately to customer emotions, which is difficult to do manually. This increases the burden on salespeople and can lead to decreased customer satisfaction. Therefore, there is a need for a system that reduces the burden on salespeople and automatically generates appropriate document drafts that respond to customer emotions.

[1223] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1224] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, a means for analyzing the user's emotions, and a means for adjusting the draft texts based on the user's emotions. This reduces the burden on salespeople and enables the automatic generation of appropriate draft texts that respond to customer emotions.

[1225] A "database" is a system that stores historical information and retrieves it as needed.

[1226] "Machine learning" is a technology that allows computers to learn patterns from data and perform predictions and classifications.

[1227] A "draft of the proposal / contract section" refers to a draft of a document used when making proposals or entering into contracts with customers during sales activities.

[1228] "Interactive presentation methods" refer to interfaces for displaying generated text drafts while interacting with the user.

[1229] "Methods for analyzing user emotions" refer to technologies for analyzing emotions from user input and behavior.

[1230] "Means for adjusting text drafts based on user emotions" refers to technologies for appropriately modifying text drafts according to the analyzed user emotions.

[1231] The system for implementing this invention consists of three main elements: a server, a terminal, and a user. The server includes a database storing historical information, a machine learning model, a sentiment analysis engine, and a generative AI model. The terminal is a device operated by the user, such as a smartphone or a personal computer. The user is a salesperson or customer support representative.

[1232] The server first retrieves historical information from the database. This information includes past proposals, contract texts, and customer feedback. Next, it uses a machine learning model to automatically generate draft proposal and contract texts from the retrieved information. These generated draft texts are then presented interactively to the user's terminal.

[1233] Users can review draft text and make revisions as needed. Furthermore, the server uses a sentiment analysis engine to analyze user emotions from their input and actions. This sentiment analysis engine can utilize existing sentiment analysis tools.

[1234] Once the user's emotions are analyzed, a generative AI model adjusts the draft text based on those emotions. For example, if the user is happy, it generates a draft text that includes positive language; if the user is angry, it generates a draft text that includes apologies. Generative AI models such as OpenAI's GPT-3 are used in this process.

[1235] As a concrete example, the prompt message when the user is satisfied with the product is as follows:

[1236] Prompt message (suggested message when the user is pleased): Thank you for your recent purchase. We hope you are satisfied with our service.

[1237] By inputting this prompt into the AI ​​model, it is possible to generate appropriate text suggestions that reflect the user's emotions.

[1238] This system reduces the workload on sales staff and enables the automatic generation of appropriate message drafts tailored to customer sentiment. This is expected to improve customer satisfaction and sales efficiency.

[1239] The flow of a specific process in Application Example 2 will be explained using Figure 20.

[1240] Step 1:

[1241] The server retrieves historical information from the database. Inputs include data such as past proposals, contract documents, and customer feedback. Outputs are the retrieved historical information. Specifically, the server executes SQL queries to extract the necessary information from the database.

[1242] Step 2:

[1243] The server uses a machine learning model to automatically generate draft text for proposals and contracts from the acquired information. The input is the historical information acquired in Step 1. The output is the generated draft text. Specifically, the server executes a machine learning algorithm to generate new draft text based on the historical information.

[1244] Step 3:

[1245] The server interactively presents the generated draft text to the terminal. The input is the draft text generated in step 2. The output is the draft text displayed on the terminal. Specifically, the server sends the draft text to the terminal in JSON format, and the terminal displays it in its user interface.

[1246] Step 4:

[1247] The user reviews the draft text via the terminal and makes revisions as needed. The input is the draft text presented in step 3. The output is the draft text revised by the user. Specifically, the user edits the draft text using the input fields on the terminal.

[1248] Step 5:

[1249] The server uses a sentiment analysis engine to analyze emotions from user input and actions. The input consists of the revised text draft from step 4 and the user's input data. The output is the user's sentiment score. Specifically, the server calls an API to analyze the user's input data and obtain the sentiment score.

[1250] Step 6:

[1251] The server uses a generative AI model to adjust the text drafts based on the user's emotions. The input is the emotion score obtained in step 5 and the modified text draft in step 4. The output is the text draft adjusted according to the emotions. Specifically, the server inputs a prompt sentence into OpenAI's GPT-3 model and generates an emotion-appropriate text draft.

[1252] Step 7:

[1253] The server presents the revised text draft to the terminal again. The input is the revised text draft from step 6. The output is the final text draft displayed on the terminal. Specifically, the server sends the revised text draft to the terminal in JSON format, and the terminal displays it in its user interface.

[1254] (Example 3)

[1255] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1256] Traditional systems could generate draft documents based on past information, but they couldn't generate documents that reflected the user's emotions. Therefore, it was difficult to automatically generate documents such as proposals and apologies that were appropriate to the user's feelings. Furthermore, the lack of concrete means to generate documents that approved by the approver resulted in an inefficient approval process.

[1257] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[1258] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, an emotion engine for detecting the user's emotions, and a means for generating draft texts based on emotion data obtained from the emotion engine. This makes it possible to automatically generate appropriate draft texts according to the user's emotions and provide draft texts that are likely to satisfy the approver.

[1259] A "database" is a system that stores historical information and retrieves it as needed.

[1260] "Artificial intelligence" is a technology that uses machine learning to analyze data and automatically generate draft text for proposals and contracts.

[1261] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing the user to review and modify the draft text.

[1262] An "emotion engine" is a technology that detects emotions from a user's facial expressions, voice, etc., and analyzes that emotional data.

[1263] "Emotional data" refers to data that indicates the user's emotional state as detected by the emotion engine.

[1264] A "draft document" is a draft of a proposal or contract document, generated by artificial intelligence.

[1265] This invention is a system that combines a database storing past information, artificial intelligence that generates text drafts using machine learning, and an emotion engine that detects user emotions. Specific embodiments of this system are described below.

[1266] System Configuration

[1267] database

[1268] The server uses a database that stores past approval history and approver trends. This database can be a relational database management system (RDBMS), such as MySQL. The database stores approval history data and approver trend data.

[1269] artificial intelligence

[1270] The server uses a generative AI model (e.g., GPT-4) to generate draft text based on information retrieved from the database. The generative AI model learns from past approval history and approver tendencies, and generates new draft text based on that.

[1271] Emotional Engine

[1272] The device uses an emotion engine to detect the user's emotions in real time. The emotion engine analyzes the user's facial expressions and voice to determine their emotions and sends the results to the server.

[1273] System operation

[1274] Data collection

[1275] The server retrieves historical approval history data and approver trend data from the database. For example, the server retrieves data by executing an SQL query such as "SELECT FROM approval_history WHERE approver_id = 'X'".

[1276] Data Learning

[1277] The server inputs the acquired data into a generating AI model to train the model. During the training process, it analyzes past approval history and approver tendencies to extract patterns.

[1278] Text draft generation

[1279] The server inputs a prompt message into the trained generative AI model, which then generates a new draft text. The prompt message includes specific instructions based on the approver's tendencies. For example, it might input a prompt message such as, "Generate a draft text that will satisfy approver X. Please include specific numbers."

[1280] Emotion detection

[1281] The device captures the user's facial expressions with its camera and sends the analysis results to the server by calling an API. For example, if the user is happy, the emotion engine detects that emotion and sends data indicating "the user is happy" to the server.

[1282] Emotion-based text drafting

[1283] The server generates appropriate text suggestions based on the received sentiment data. For example, if the user is happy, the server inputs a prompt message to the generation AI model saying, "Generate a suggestion text suggestion for when the user is happy." The generation AI model then generates a text suggestion such as, "This suggestion will lead your business to even greater success."

[1284] Specific example

[1285] Specific examples of data collection

[1286] The server executes the SQL query "SELECT FROM approval_history WHERE approver_id = 'X'" to retrieve data from the database.

[1287] Examples of text draft generation

[1288] The server inputs the prompt message, "Generate a draft document that will satisfy approver X. Please include specific numerical values," into the AI ​​model and receives the generated draft document.

[1289] Examples of generating text drafts based on emotions

[1290] The server inputs the prompt message "Generate suggested text for when the user is happy" into the AI ​​model and presents the generated text suggestions to the user.

[1291] In this way, the server, terminal, and user work together to generate the optimal draft text that reflects the emotions of the approver and the user. The flow of the specific processing in Example 3 will be explained using Figure 21.

[1292] Step 1: Data Collection

[1293] The server retrieves historical approval history data and approver trend data from the database. It takes an SQL query (e.g., "SELECT FROM approval_history WHERE approver_id = 'X'") as input and loads the retrieved data into memory as output. Specifically, the server establishes a database connection, executes the SQL query, and retrieves the results.

[1294] Step 2: Data Training

[1295] The server inputs the acquired data into a generative AI model (e.g., GPT-4) to train the model. Past approval history data and approver trend data are used as input, and a trained generative AI model is obtained as output. Specifically, the server preprocesses the data, inputs it into the generative AI model, and starts the training process.

[1296] Step 3: Draft document generation

[1297] The server inputs a prompt sentence into a fully trained generative AI model and generates a new draft text. The input is a prompt sentence (e.g., "Generate a draft text that will satisfy approver X. Please include specific numerical values."), and the output is the generated draft text. Specifically, the server inputs the prompt sentence into the generative AI model and receives the generated draft text.

[1298] Step 4: Emotion Detection

[1299] The device captures the user's facial expressions with its camera and calls an API to send the analysis results to the server. It uses the user's facial expression data as input and obtains analyzed emotion data as output. Specifically, the device captures the user's facial expressions with its camera, sends the data to the API, and receives the analysis results.

[1300] Step 5: Emotion-based drafting

[1301] The server generates appropriate text suggestions based on the received sentiment data. It uses sentiment data (e.g., "The user is happy") as input and outputs text suggestions corresponding to that sentiment. Specifically, the server analyzes the sentiment data, inputs a prompt (e.g., "Generate suggestion text suggestions for when the user is happy.") into the AI ​​model, and receives the generated text suggestions.

[1302] In this way, the server, terminal, and user work together to generate the optimal draft text that reflects the emotions of the approver or user.

[1303] (Application Example 3)

[1304] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1305] Conventional draft document generation systems struggled to generate draft documents that considered user emotions, thus failing to improve user satisfaction. Furthermore, the lack of appropriate means to generate draft documents that approved by approvers sometimes led to delays in the approval process. Additionally, draft document generation that reduced the workload on sales representatives was insufficient.

[1306] In Application Example 3, the specific processing performed by the specific processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes artificial intelligence that retrieves information from a database storing past information and automatically generates draft text for the proposal and contract sections using machine learning, means for interactively presenting the generated draft text, an emotion engine that analyzes the user's emotions, and means for adjusting the draft text based on the emotion engine. This makes it possible to generate draft text that takes the user's emotions into consideration, thereby increasing user satisfaction. Furthermore, by generating draft text that approvers can accept, delays in the approval process can be prevented. In addition, draft text can be generated to reduce the workload of sales personnel.

[1307] "Past information" refers to data such as the user's or approver's past behavioral history, transaction history, and approval history.

[1308] A "database" refers to an information aggregation system that systematically stores past information and allows it to be searched and retrieved as needed.

[1309] "Machine learning" refers to a technology that uses computers to analyze large amounts of data, find patterns and regularities, and automatically perform specific tasks.

[1310] "Draft proposal / contract text" refers to a draft document containing the proposed content and contract details that will be presented to users and approvers.

[1311] "Artificial intelligence" refers to computer systems that use technologies such as machine learning and natural language processing to imitate and support human intellectual tasks.

[1312] "Interactive presentation methods" refer to interfaces that interactively display generated draft text to users and approvers and receive their feedback.

[1313] An "emotion engine" refers to a software module that analyzes a user's emotions and takes appropriate action based on the results.

[1314] "Means for adjusting draft text" refers to a function for appropriately modifying and adjusting the generated draft text based on the analysis results of the emotion engine.

[1315] A system for carrying out this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, an emotion engine that analyzes the user's emotions, and means for adjusting the draft text based on the emotion engine.

[1316] The server first retrieves information from the database, such as the user's and approver's past behavioral history, transaction history, and approval history. Based on this information, it uses a machine learning algorithm to generate draft text for the proposal and contract sections. The generated draft text is presented interactively to the user or approver, and feedback can be received.

[1317] Next, the emotion engine analyzes the user's emotions. The emotion engine analyzes the user's emotions from their input and actions, and adjusts the draft text based on the results. For example, if the user is feeling anxious, a reassuring message can be added to the draft text.

[1318] This system requires software such as APIs and sentiment analysis modules. Hardware requirements include servers and smartphones.

[1319] As a concrete example, consider generating a message when a user makes a large payment. If the user ID is 12345 and the payment amount is 10,000 yen, the following prompt message will be generated based on the user's past payment history and sentiment.

[1320] Example of a prompt:

[1321] User ID: 12345

[1322] Payment amount: 10,000 yen

[1323] Past payment history: Past payment history data

[1324] User's emotion: Anxiety

[1325] Please generate an appropriate message.

[1326] By inputting this prompt into the AI ​​generation model, it is possible to generate appropriate messages that take into account the user's emotions. In this way, a system that can increase user satisfaction can be realized.

[1327] The flow of the specific processing in Application Example 3 will be explained using Figure 22.

[1328] Step 1:

[1329] The server retrieves information from the database, such as the past activity history, transaction history, and approval history of users and approvers. Inputs are user IDs and approver IDs, and output is data such as past activity history, transaction history, and approval history. This data is used in subsequent processing.

[1330] Step 2:

[1331] The server generates draft text for the proposal and contract sections using a machine learning algorithm based on the acquired historical information. The input is the historical information acquired in step 1, and the output is the generated draft text. Specifically, the system inputs the historical information into a machine learning model and generates the optimal draft text.

[1332] Step 3:

[1333] The server interactively presents the generated draft text to the user or approver. The input is the draft text generated in step 2, and the output is the feedback from the user or approver. Specifically, it displays the draft text through the user interface and receives comments and revision requests from the user or approver.

[1334] Step 4:

[1335] The server analyzes user emotions using an emotion engine based on user input and behavior. Input consists of user feedback and behavioral data, while output is the user's emotional state. Specifically, it uses an emotion analysis module to analyze the user's emotions.

[1336] Step 5:

[1337] The server adjusts the draft text based on the analysis results of the emotion engine. The input is the user's emotional state obtained in step 4 and the draft text generated in step 2, and the output is the adjusted draft text. Specifically, the server modifies the content of the draft text according to the user's emotions; for example, it adds a reassuring message to a user who is feeling anxious.

[1338] Step 6:

[1339] The server presents the revised draft text to the user or approver again and receives final feedback. The input is the revised draft text from step 5, and the output is the final approval or revision request. Specifically, the draft text is displayed again through the user interface for final confirmation.

[1340] Step 7:

[1341] The server stores the final draft document in a database and uses it to generate draft documents for future proposals and contract sections as needed. The input is the final draft document, and the output is the updated database. Specifically, it stores the final draft document in the database and uses it as training data for future machine learning models.

[1342] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1343] The data generation model 58 is a form of so-called generative AI (Artificial Intelligence). One example of the data generation model 58 is ChatGPT (Internet Search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1344] Other examples of generative AI include Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) are some examples.

[1345] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[1346] [Third Embodiment]

[1347] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[1348] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1349] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1350] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[1351] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1352] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1353] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1354] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1355] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1356] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1357] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1358] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[1359] "Example of form 1"

[1360] In one embodiment of the present invention, there exists a database storing past information. This database includes, for example, information on contracts and proposals made by corporate sales representatives in the past. There exists an AI that retrieves information from this database and automatically generates draft text for proposals and contracts using machine learning. This AI is trained using machine learning techniques such as deep learning and reinforcement learning. The generated draft text is presented to the sales representative through an interactive interface. This interface is implemented, for example, in the form of a chatbot or a voice assistant. "Example of Form 2"

[1361] In another embodiment of the present invention, AI generates draft documents to reduce the workload of salespeople. Specifically, the AI ​​automates some of the tasks that salespeople normally perform. For example, it significantly reduces the time and effort salespeople spend creating draft proposals and contract documents.

[1362] "Example of form 3"

[1363] In a further embodiment of the present invention, the AI ​​generates draft text that is expected to be acceptable to the approver. Specifically, the AI ​​learns past approval history and approver tendencies, and generates draft text based on that. This makes it possible to provide draft text that is highly likely to be acceptable to the approver.

[1364] The following describes the processing flow for each example of the form.

[1365] "Example of form 1"

[1366] Step 1: Retrieve the necessary information from a database containing past data. This information could include, for example, past contracts and proposals.

[1367] Step 2: Based on the acquired information, machine learning is used to automatically generate draft text for the proposal and contract sections. This AI is trained using machine learning techniques such as deep learning and reinforcement learning.

[1368] Step 3: The generated draft text is presented to the salesperson through an interactive interface. This interface can be implemented, for example, as a chatbot or voice assistant.

[1369] "Example of form 2"

[1370] Step 1: The AI ​​generates draft text to reduce the workload of salespeople. Specifically, the AI ​​automates some of the tasks that salespeople normally perform.

[1371] Step 2: For example, significantly reduce the time and effort salespeople spend creating proposals and draft contract documents.

[1372] "Example of form 3"

[1373] Step 1: The AI ​​generates draft text that is expected to satisfy the approver. Specifically, the AI ​​learns from past approval history and the approver's tendencies.

[1374] Step 2: Generate a draft document based on that. This makes it possible to provide a draft document that is more likely to be accepted by the approver.

[1375] (Example 1)

[1376] Next, we will describe Embodiment 1 of Embodiment Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1377] In traditional corporate sales, a major problem was the significant time and effort required to create proposals and contracts. Furthermore, it was difficult for salespeople to efficiently utilize information from past proposals and contracts, sometimes resulting in a decline in the quality of proposals. Additionally, creating documents that would satisfy approvers required experience and skill, placing a significant burden on new salespeople.

[1378] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1379] In this invention, the server includes means for retrieving information from a database storing historical information, means for training a machine learning model using the retrieved information, means for automatically generating draft text for proposals and contracts using the trained machine learning model, and means for interactively presenting the generated draft text. This reduces the burden on salespeople and enables the efficient and high-quality creation of proposals and contracts. Furthermore, by automatically generating draft text that is acceptable to approvers, the approval process can also be expedited.

[1380] A "database" is a system that stores historical information and allows you to retrieve that information as needed.

[1381] "Means of acquiring information" refers to the methods and techniques for extracting necessary information from a database.

[1382] A "machine learning model" is an algorithm or framework that learns from data and automatically performs a specific task.

[1383] "Training methods" refer to the methods and techniques used to train machine learning models using data.

[1384] "Methods for automatically generating draft documents" refer to methods and technologies that use trained machine learning models to automatically create drafts of new proposals and contracts.

[1385] "Interactive presentation methods" refer to methods and technologies for interactively displaying generated draft text to the user.

[1386] "Salesperson" refers to the person or staff member who handles corporate sales.

[1387] An "approver" refers to a person who has the authority to review and approve the contents of a proposal or contract.

[1388] Modes for carrying out the invention

[1389] This invention is a system designed to streamline the creation of proposals and contracts in corporate sales, thereby reducing the burden on salespeople. A specific embodiment of this system is described below.

[1390] Retrieve information from the database.

[1391] The server retrieves the necessary information from a database containing information on past contracts and proposals made by corporate sales representatives. This database includes contract details, proposal details, dates, and customer information. The server extracts information from the database using SQL queries. For example, executing a query such as "SELECT FROM contracts WHERE date > '2020-01-01'" retrieves contract information from 2020 onwards.

[1392] Training machine learning models

[1393] The server uses the acquired information to train a machine learning model. This model employs techniques such as deep learning and reinforcement learning. Specifically, it uses the TensorFlow framework. The server cleans and normalizes the data, and then trains the model using the preprocessed data. For example, it trains a text generation model using an LSTM (Long Short-Term Memory) network.

[1394] Automatic generation of draft text

[1395] The server uses a trained LSTM model to automatically generate draft proposals and contract documents. When a user enters a prompt such as "Generate a new proposal," the server inputs this prompt into the model and retrieves the generated draft document. For example, it might generate a document stating, "This proposal provides the best solution for your needs."

[1396] Presentation to the business operator

[1397] The generated draft text is presented to the salesperson through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant. The salesperson accesses the chatbot through their device and reviews the generated draft text. For example, if the salesperson types "Show me the draft text of the proposal" into the chatbot, the chatbot will display the text "This proposal provides the best solution for your company's needs."

[1398] Specific examples and prompt statements

[1399] Suppose a user needs to create a new proposal. The user accesses the system via their device and instructs it to generate draft text based on information from similar past proposals. The server retrieves past proposal information from the database and uses a machine learning model to generate a new draft text. The generated draft text is then presented to the user via a chatbot.

[1400] Example of a prompt:

[1401] "Please generate a draft of a new proposal document based on past corporate sales proposal data."

[1402] In this way, users can efficiently create new proposals.

[1403] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1404] Step 1:

[1405] The server retrieves information on past contracts and proposals from the database. SQL queries are used as input. For example, executing the query "SELECT FROM contracts WHERE date > '2020-01-01'" will output contract information from 2020 onwards. The server stores the retrieved data in memory.

[1406] Step 2:

[1407] The server preprocesses the acquired data. Raw data retrieved from the database is used as input. Preprocessing includes data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). This results in output data in a format suitable for training machine learning models.

[1408] Step 3:

[1409] The server trains a machine learning model using preprocessed data. Preprocessed data is used as input. Specifically, it trains an LSTM (Long Short-Term Memory) network using the TensorFlow framework. The training results in a model for generating proposal and contract texts.

[1410] Step 4:

[1411] The server uses a trained LSTM model to automatically generate drafts of new proposals and contract documents. The input is a prompt sentence entered by the user. For example, the prompt "Generate a new proposal" might be entered. The server inputs this prompt into the model and outputs a generated draft document. For example, it might generate a document stating, "This proposal provides the best solution for your needs."

[1412] Step 5:

[1413] The server presents the generated draft text to the salesperson through an interactive interface. The generated draft text is used as input. The interactive interface is implemented in the form of a chatbot or voice assistant. For example, if the salesperson types "Show me the draft text of the proposal" into the chatbot, the chatbot will display the text "This proposal provides the best solution for your company's needs."

[1414] (Application Example 1)

[1415] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1416] In traditional sales activities, sales representatives must manually refer to past contract and proposal information to create new proposals and contracts, which is time-consuming and labor-intensive. Furthermore, maintaining a consistent level of quality is difficult because the quality of proposals and contracts depends on the sales representative's experience and skills. There is a growing need for a way for sales representatives to efficiently create proposals even when they are on the go.

[1417] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1418] This invention includes a server that retrieves information from a database storing historical data and automatically generates draft proposal and contract documents using machine learning; a server that interactively presents the generated draft documents; and a server that presents the generated draft documents to sales representatives via smartphone. This enables sales representatives to quickly generate high-quality proposal and contract documents based on historical data and to conduct sales activities efficiently even when on the go.

[1419] A "database containing past information" is a data storage system that stores information on past contracts and proposals made by corporate sales representatives, allowing them to search and retrieve the information as needed.

[1420] "Machine learning" is a technology in which computers learn patterns and rules from data and use that knowledge to make predictions and decisions about new data.

[1421] "Artificial intelligence that automatically generates draft text for proposals and contracts" refers to a program that uses machine learning to analyze past data and automatically generate new proposal and contract documents.

[1422] "Interactive presentation methods" refer to interfaces that interactively display generated draft text to the user, allowing the user to review and modify its content.

[1423] The "means of presenting information to sales representatives via smartphone" refers to a function that displays generated draft text on the sales representative's smartphone, enabling them to conduct sales activities efficiently even when they are out of the office.

[1424] A system for implementing this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, and means for presenting the generated draft text to sales representatives via smartphone.

[1425] System Configuration

[1426] 1. Database: The server has a database that stores information on past contracts and proposals made by corporate sales representatives. This database is built using a publicly known database management system.

[1427] 2. Artificial Intelligence: The server includes artificial intelligence that uses machine learning (e.g., OpenAI's GPT-3) to automatically generate new proposals and contracts based on past data. This artificial intelligence is implemented using a publicly known programming language.

[1428] 3. Interactive Interface: The generated draft text is presented to the sales representative through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant.

[1429] 4. Smartphone Application: The generated draft text is displayed on the sales representative's smartphone. The smartphone application runs on the Android platform, enabling sales representatives to conduct sales activities efficiently even when they are on the go.

[1430] Processing flow

[1431] 1. Data Retrieval: The server retrieves past proposals and contracts for a specific client from the database.

[1432] 2. Prompt generation: The server generates prompts to input into the artificial intelligence based on previously retrieved suggestion sentences.

[1433] 3. Document draft generation: The artificial intelligence generates new proposals and contracts based on the generated prompts.

[1434] 4. Presentation of draft text: The generated draft text is presented to the sales representative through an interactive interface.

[1435] 5. Smartphone display: The generated draft text is displayed on the sales representative's smartphone, allowing them to conduct sales activities efficiently even when they are out of the office.

[1436] Hardware and software to be used

[1437] Hardware: Servers, smartphones

[1438] software

[1439] Specific example

[1440] For example, if a past proposal from client ID 123 was "This product is high-quality and offers excellent value for money," the newly generated proposal might be "This product uses the latest technology and is highly reliable."

[1441] Example of a prompt

[1442] Previous proposal: This product is high quality and offers excellent value for money. Please generate a new proposal:

[1443] In this way, sales representatives can easily generate new proposals using their smartphones and make effective proposals to clients.

[1444] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1445] Step 1:

[1446] The server retrieves past proposals and contracts for a specific client from the database. The input is the client ID, and the output is a list of past proposals and contracts. Specifically, the server executes an SQL query to extract proposals related to the corresponding client ID from the database.

[1447] Step 2:

[1448] The server generates prompts for the artificial intelligence based on the previously retrieved proposals. The input is a list of past proposals, and the output is a prompt. Specifically, the server combines the past proposals into a single text and creates a prompt with the preface "Past proposals: ".

[1449] Step 3:

[1450] The server sends the generated prompt text to the artificial intelligence, which then generates new proposal or contract texts. The input is the prompt text, and the output is the generated proposal text. Specifically, the server calls the OpenAI GPT-3 API, passing the prompt text as input and receiving the generated text.

[1451] Step 4:

[1452] The server presents the generated proposal to the sales representative through an interactive interface. The input is the generated proposal, and the output is an interface that the sales representative can review and modify. Specifically, the server displays the generated proposal in a chatbot format, allowing the sales representative to review and modify its contents.

[1453] Step 5:

[1454] The server displays the generated proposal on the sales representative's smartphone. The input is the generated proposal, and the output is the proposal displayed on the smartphone screen. Specifically, the server sends the generated proposal to the smartphone application, allowing the sales representative to review and use the proposal even when they are away from the office.

[1455] (Example 2)

[1456] Next, we will describe Example 2 of the Form Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1457] There is a need to reduce the time and effort salespeople spend creating proposals and contract drafts, and to proceed with their work more efficiently. It is also crucial to maintain a consistent quality of draft documents and to quickly generate content that approves. A system is needed to address these challenges.

[1458] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1459] In this invention, the server includes a generation AI model that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, means for generating prompt texts based on information input by the user, and means for interactively presenting the generated draft texts. This reduces the burden on sales personnel and makes it possible to efficiently generate high-quality draft texts.

[1460] A "database" is a system that stores historical information and retrieves it as needed.

[1461] "Machine learning" is a technology that allows computers to learn patterns from data and perform predictions and classifications.

[1462] A "generative AI model" is an artificial intelligence model that uses machine learning to automatically generate draft text for proposals and contracts.

[1463] A "prompt statement" is a sentence used to give instructions to an AI model based on information entered by the user.

[1464] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing for confirmation and modification.

[1465] A "salesperson" is a person responsible for creating proposals and drafting contract documents.

[1466] A "draft document" is a draft of a document related to a proposal or contract.

[1467] An "approver" is a person who has the authority to review and approve the generated draft document.

[1468] This invention is a system that reduces the burden on salespeople and efficiently generates high-quality proposals and contract drafts. This system operates through the cooperation of three parties: a server, a terminal, and a user.

[1469] First, the server maintains a database containing past information and retrieves it as needed. The database stores information about past proposals and contracts. The server uses a generative AI model (for example, OpenAI's GPT-4) that uses machine learning to automatically generate draft text for proposals and contracts.

[1470] Users input basic information about proposals and contracts via the terminal. Specifically, they input information such as the customer name, proposal details, and contract terms (e.g., initial fees, monthly fees). For example, when a user creates a proposal for a new customer, they would input the following information into the terminal:

[1471] Customer name: XYZ Co., Ltd.

[1472] Proposal: Proposal for a new marketing strategy

[1473] Contract terms: Initial fee 1 million yen, monthly fee 100,000 yen

[1474] The server generates prompt messages based on the information entered by the user. These prompt messages include instructions for the generating AI model. For example, it generates prompt messages like the following:

[1475] "Please prepare a proposal for a new marketing strategy for XYZ Corporation. The proposal should include contract terms of an initial fee of 1 million yen and a monthly fee of 100,000 yen."

[1476] The generative AI model receives a prompt and generates draft proposals and contract documents. The goal is to produce draft documents of the same quality as those manually created by sales representatives. For example, it can generate draft documents like the following:

[1477] "XYZ Corporation"

[1478] Thank you for reviewing our proposal for a new marketing strategy. The proposal details and contract terms are outlined below.

[1479] Proposal details:

[1480] Introducing a new marketing strategy

[1481] Initial cost: 1 million yen

[1482] Monthly fee: 100,000 yen

[1483] Thank you for your consideration.

[1484] Sincerely"

[1485] The generated draft text is sent from the server to the terminal and presented to the user interactively. The user can review the displayed draft text and make revisions as needed. This reduces the workload on sales staff and enables the efficient generation of high-quality draft text.

[1486] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1487] Step 1:

[1488] Users input basic information about proposals and contracts via the terminal. Specifically, they input information such as the customer name, proposal details, and contract terms (e.g., initial fees, monthly fees). For example, when a user creates a proposal for a new customer, they would input the following information into the terminal:

[1489] Customer name: XYZ Co., Ltd.

[1490] Proposal: Proposal for a new marketing strategy

[1491] Contract terms: Initial fee 1 million yen, monthly fee 100,000 yen

[1492] Input: Customer name, proposal details, contract terms

[1493] Output: Basic information entered by the user

[1494] Step 2:

[1495] The server generates prompt messages based on the information entered by the user. These prompt messages include instructions for the AI ​​model that generates them. The server analyzes the input information and creates appropriate prompt messages. For example, it might generate a prompt message such as, "Please create a proposal for a new marketing strategy for XYZ Corporation. The proposal should include contract terms of an initial fee of 1 million yen and a monthly fee of 100,000 yen."

[1496] Input: Basic information entered by the user

[1497] Output: Generated prompt message

[1498] Step 3:

[1499] The server sends a prompt to a generative AI model (e.g., OpenAI's GPT-4) and generates a draft document. The generative AI model then creates a draft proposal or contract document based on the prompt. For example, it might generate a draft document like this:

[1500] "XYZ Corporation"

[1501] Thank you for reviewing our proposal for a new marketing strategy. The proposal details and contract terms are outlined below.

[1502] Proposal details:

[1503] Introducing a new marketing strategy

[1504] Initial cost: 1 million yen

[1505] Monthly fee: 100,000 yen

[1506] Thank you for your consideration.

[1507] Sincerely"

[1508] Input: Generated prompt message

[1509] Output: Generated draft text

[1510] Step 4:

[1511] The server sends the generated draft text to the terminal and presents it to the user interactively. The user can review the displayed draft text and make corrections as needed. The user checks the draft text on the terminal screen, verifies that there are no errors, and makes corrections as necessary.

[1512] Input: Generated draft text

[1513] Output: Draft text presented to the user

[1514] In this way, the server, terminals, and users work together, reducing the workload on sales personnel and enabling the efficient generation of high-quality draft documents.

[1515] (Application Example 2)

[1516] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as a "server," and the headset-type terminal 314 will be referred to as a "terminal."

[1517] A problem is that salespeople spend a great deal of time and effort creating proposals and contracts. Furthermore, communication with customers and managing the progress of sales activities are also time-consuming, making efficient sales activities difficult. There is a need to solve these problems, reduce the burden on salespeople, and achieve more efficient sales activities.

[1518] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1519] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract documents using machine learning, a means for interactively presenting the generated draft documents, a means for automating communication with customers, a means for automatically generating proposals and contracts, a means for managing and analyzing customer information, and a means for managing the progress of sales activities. This significantly reduces the workload on salespeople and enables efficient sales activities.

[1520] "Past information" refers to past data and records related to sales activities.

[1521] A "database" refers to an information management system that systematically stores information and allows it to be searched and retrieved as needed.

[1522] "Machine learning" refers to the technology that allows computers to learn patterns from data and perform predictions and classifications.

[1523] "Draft text for proposals and contracts" refers to draft proposals and contracts that are presented to customers during sales activities.

[1524] "Artificial intelligence" refers to computer systems that mimic human intelligence and perform learning and reasoning.

[1525] "Interactive presentation methods" refer to methods in which users and systems exchange information in a dialogue format.

[1526] "Methods for automating communication with customers" refers to methods for automating communication with customers, such as email and chat.

[1527] "Methods for automatically generating proposals and contracts" refers to methods for automatically creating proposals and contracts necessary for sales activities.

[1528] "Means for managing and analyzing customer information" refers to methods for organizing and analyzing data related to customers.

[1529] "Means of managing the progress of sales activities" refers to methods of tracking and managing the progress of the sales process.

[1530] The system for implementing this invention is configured as follows: The server retrieves information from a database containing historical data and automatically generates draft proposal and contract documents using machine learning. The generated draft documents are presented to the user interactively. It also has the functionality to automate communication with customers and automatically generate proposals and contracts. Furthermore, it includes functions to manage and analyze customer information and manage the progress of sales activities.

[1531] Hardware and software to be used

[1532] Hardware: Servers, user terminals (smartphones, personal computers, etc.)

[1533] Software: APIs, database management systems (e.g., MySQL)

[1534] Data processing and data calculation

[1535] 1. Retrieving information from the database: The server retrieves data related to past sales activities from the database. This data includes past proposals, contracts, and communication history with customers.

[1536] 2. Text draft generation using machine learning: Based on the acquired data, a machine learning model (e.g., OpenAI's GPT-3) is used to generate draft text for proposals and contracts.

[1537] 3. Interactive Presentation: The generated draft text is presented interactively on the user's terminal. The user can review the draft text and make modifications as needed.

[1538] 4. Automated communication with customers: The server automates email and chat communication with customers based on the generated draft messages.

[1539] 5. Automatic generation of proposals and contracts: The server automatically generates formal proposals and contracts based on the generated draft documents.

[1540] 6. Customer Information Management and Analysis: The server organizes and analyzes customer information to improve the efficiency of sales activities.

[1541] 7. Sales activity progress management: The server tracks and manages the progress of the sales process.

[1542] Specific example

[1543] For example, if the customer name is "ABC Corporation" and the product details are "latest AI-equipped smartphone," the following prompt message will be sent to the AI ​​model.

[1544] Example of a prompt

[1545] Customer Name: ABC Co., Ltd.

[1546] Product Details: Latest AI-powered smartphone

[1547] Please prepare your proposal based on the following information.

[1548] Based on this prompt, the AI ​​model generates the text of the proposal, reducing the workload on salespeople. The generated proposal is presented interactively on the user's terminal, allowing the user to review the content and make revisions as needed.

[1549] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1550] Step 1:

[1551] The server retrieves data on past sales activities from a database. Inputs include past proposals, contracts, and customer communication history stored in the database. Output is a collection of the retrieved data, which is used for subsequent processing.

[1552] Step 2:

[1553] The server uses a machine learning model (e.g., OpenAI's GPT-3) to generate draft proposal and contract documents based on the acquired data. The input is the data acquired in step 1. The output is the generated draft documents. Specifically, the data is converted into prompt sentences and sent to the generation AI model.

[1554] Step 3:

[1555] The server interactively presents the generated draft text to the user terminal. The input is the draft text generated in step 2. The output is the draft text displayed on the user terminal. Specifically, the server displays the draft text in the user interface, allowing the user to review and modify it.

[1556] Step 4:

[1557] The server automates email and chat communication with customers based on the generated draft text. The input is the draft text generated in step 2. The output is the email or chat message sent to the customer. Specifically, it converts the draft text into email or chat format and sends it automatically.

[1558] Step 5:

[1559] The server automatically generates formal proposals and contracts based on the generated draft documents. The input is the draft documents generated in step 2. The output is the formal proposal or contract document. Specifically, it formats the draft documents into a formal document format and converts them into a format that can be saved and printed.

[1560] Step 6:

[1561] The server organizes and analyzes customer information. The input is customer information stored in a database. The output is the analysis results and the organized customer information. Specifically, it organizes customer information by category and applies analytical algorithms to extract useful insights.

[1562] Step 7:

[1563] The server tracks and manages the progress of the sales process. Inputs are data and progress information related to sales activities. Outputs are progress reports. Specifically, it tracks each stage of sales activities and generates dashboards to visualize progress.

[1564] (Example 3)

[1565] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1566] Conventional draft document generation systems failed to adequately consider past approval history and approver tendencies, making it difficult to efficiently create draft documents that approved by approvers. Furthermore, draft document generation that reduced the workload on sales staff was insufficient. This resulted in delays in the approval process and decreased operational efficiency.

[1567] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes means for the user to input a prompt sentence, means for the terminal to send the prompt sentence to the server, means for the server to obtain past approval history and approver tendencies from a database, means for generating a draft text using a generation AI model, means for sending the generated draft text to the terminal, means for the terminal to display the draft text to the user, and means for the user to check the draft text and modify it as necessary. This makes it possible to efficiently generate draft text that takes into account past approval history and approver tendencies, and to provide draft text that is likely to be accepted by the approver. In addition, it is possible to reduce the burden on sales personnel and improve operational efficiency.

[1568] A "database" is a system that stores historical information and retrieves it as needed.

[1569] "Artificial intelligence" refers to a program or system that uses machine learning to automatically generate draft text.

[1570] "Interactive presentation methods" refer to interfaces that display generated draft text to the user in a conversational format, allowing the user to review and modify the text.

[1571] A "prompt message" is text that the user enters, containing instructions and conditions for generating a draft of the text.

[1572] A "terminal" is a device used by a user to input prompt messages and communicate with a server.

[1573] A "server" is a computer system that retrieves information from a database, generates draft text using a generative AI model, and sends it to a terminal.

[1574] A "generative AI model" is an artificial intelligence model that learns from past data and generates draft text.

[1575] "Approval history" refers to data that includes previously approved documents and feedback from approvers.

[1576] "Approver tendencies" refer to patterns based on the types of documents approvers have approved in the past or the types of feedback they have provided.

[1577] A "draft document" is a draft of a document generated by a generative AI model that is likely to be accepted by the approver.

[1578] A "user" is someone who enters prompt text and reviews and modifies the generated draft text.

[1579] This invention relates to a system for generating draft documents that take into account past approval history and the tendencies of approvers. Specific embodiments of this system are described below.

[1580] First, the user enters a prompt using the terminal. The prompt is text containing instructions and conditions for generating a draft document. For example, it might say, "Please create a document requesting budget approval for a new project. Based on past approval history, approvers tend to prioritize specific figures and clear objectives."

[1581] The terminal sends the entered prompt message to the server. This transmission uses communication protocols such as HTTP POST requests.

[1582] The server accesses the database to retrieve past approval history and approver trends. The database stores previously approved documents and approver feedback. For example, it might execute a query like "SELECT FROM approval_history WHERE approver_id = '12345'" using SQL.

[1583] Next, the server runs a generative AI model based on the acquired data to generate draft text. For example, OpenAI's GPT-4 is used as the generative AI model. The generative AI model has learned from past data and generates text that is likely to be accepted by the approver.

[1584] The generated draft text is sent from the server to the terminal. Communication protocols such as HTTP responses are used for transmission.

[1585] The terminal displays the received draft text to the user. The user interface includes a text area and display widgets. The user can review the displayed draft text and make revisions as needed. The revised text is used as the final approval request.

[1586] This system makes it possible to efficiently generate draft documents that take into account past approval history and approver tendencies, and to provide draft documents that are more likely to be accepted by approvers. Furthermore, it can reduce the workload on sales staff and improve operational efficiency. The flow of the specific processing in Example 3 will be explained using Figure 15.

[1587] Step 1:

[1588] The user enters a prompt message.

[1589] The user enters a prompt in the terminal's input field. For example, they might enter, "Please draft a document requesting budget approval for a new project. Based on past approval history, approvers tend to prioritize specific figures and clear objectives."

[1590] Input: Prompt message

[1591] Output: Input prompt message

[1592] Step 2:

[1593] The terminal sends a prompt message to the server.

[1594] The terminal sends the entered prompt text to the server as an HTTP POST request.

[1595] Input: Prompt message

[1596] Output: Prompt message sent to the server

[1597] Step 3:

[1598] The server retrieves past approval history and approver trends from the database.

[1599] The server accesses the database to retrieve past approval history and approver trends. For example, it might use an SQL query like "SELECT FROM approval_history WHERE approver_id = '12345'".

[1600] Input: Prompt message, database

[1601] Output: Past approval history and approver trends

[1602] Step 4:

[1603] The server generates draft text using an AI model.

[1604] The server runs a generative AI model (for example, OpenAI's GPT-4) based on the acquired data to generate draft text. The generative AI model has learned from past data and generates text that is likely to be accepted by the approver.

[1605] Input: Past approval history and approver trends

[1606] Output: Generated draft text

[1607] Step 5:

[1608] The server sends the generated draft text to the terminal.

[1609] The server sends the generated draft text to the terminal as an HTTP response.

[1610] Input: Generated draft text

[1611] Output: Draft text sent to the terminal

[1612] Step 6:

[1613] The device displays a draft text to the user.

[1614] The device displays the received draft text in a text area. The user can then review the displayed text.

[1615] Input: Draft text sent to the terminal

[1616] Output: Draft text displayed to the user

[1617] Step 7:

[1618] The user reviews the draft text and makes revisions as needed.

[1619] The user reviews the displayed draft text and makes any necessary revisions. For example, they might change "Personnel costs: 5 million yen" to "Personnel costs: 6 million yen." The revised text will then be used as the final approval request.

[1620] Input: Draft text displayed to the user

[1621] Output: Revised draft text

[1622] (Application Example 3)

[1623] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1624] Traditional electronic payment systems had the problem of requiring a great deal of time and effort to create documents that would satisfy approvers. Furthermore, because the documents were created without considering the approver's preferences, the approval process often suffered from delays. This resulted in inefficient expense reimbursement and large-scale transaction approval processes within companies, leading to a decline in overall business efficiency.

[1625] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[1626] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning; a server that interactively presents the generated draft texts; and a server that learns the approver's tendencies and generates draft texts that are likely to be accepted by the approver. This makes it possible to quickly generate draft texts that are easily accepted by the approver and streamline the electronic payment approval process.

[1627] "Past information" refers to data and history that the system has collected in the past.

[1628] A "database" is a collection of information that systematically stores past information and makes it searchable and retrievalable as needed.

[1629] "Machine learning" is a technology in which computers analyze large amounts of data and automatically learn patterns and rules.

[1630] "Draft text for proposals and contracts" refers to a draft of the text included in proposals and contracts.

[1631] "Artificial intelligence" is a system that uses technologies such as machine learning to mimic human intelligence and automatically perform specific tasks.

[1632] "Interactive presentation methods" refer to interfaces that allow users and systems to exchange information in a conversational format.

[1633] "Approver tendencies" refer to data that shows what criteria or patterns a particular approver has used to make approvals in the past.

[1634] A "draft document that is likely to satisfy the approver" refers to a draft document that is likely to be approved, created considering the approver's past tendencies.

[1635] As an example of how to implement this invention, an electronic payment approval assistant system will be described. This system operates in cooperation with three parties: a server, a terminal, and a user.

[1636] First, the server maintains a database containing historical information, including past approval history and data on approver tendencies. The server retrieves information from this database and uses machine learning to automatically generate draft text for proposals and contracts. Specifically, the server uses Python, Pandas, and OpenAI APIs to analyze past approval history and learn approver tendencies.

[1637] Next, the generated draft text is presented interactively on the device. This device is a user-operated device such as a smartphone or tablet, through which the user can review the generated draft text and make modifications as needed. The interactive interface is designed for easy user operation.

[1638] Furthermore, the server learns the approver's tendencies and generates draft text that is more likely to satisfy the approver. In this process, a generative AI model is used to generate draft text based on features extracted from past approval history. Examples of specific prompt texts are as follows:

[1639] Example of a prompt:

[1640] Based on the approver's tendencies, please generate the following draft text.

[1641] This business trip involves an important meeting with a new client and requires prompt approval.

[1642] Approver tendencies: {Characteristics extracted from the approver's past approval history}

[1643] This system enables the rapid generation of draft documents that are easily accepted by approvers, thereby streamlining the electronic payment approval process. For example, in the expense reimbursement process within a company, when an employee applies for travel expense reimbursement and requires supervisor approval, this system can automatically generate a draft document that is easily accepted by the supervisor, leading to quicker approval.

[1644] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[1645] Step 1:

[1646] The server retrieves approval history data from a database that stores past information.

[1647] Input: Past approval history data stored in the database

[1648] Output: Acquired approval history data

[1649] Specific operation: The server uses SQL queries to extract past approval history data from the database.

[1650] Step 2:

[1651] The server inputs the acquired approval history data into a machine learning model to learn the tendencies of the approvers.

[1652] Input: Acquired approval history data

[1653] Output: Learning results regarding approver tendencies

[1654] Specific operation: The server preprocesses the data using publicly available libraries and inputs it into a machine learning model. The model learns past approval patterns and extracts the tendencies of approvers.

[1655] Step 3:

[1656] The server inputs prompt sentences into the AI ​​model based on the learning results and generates suggested sentences.

[1657] Input: Learning results regarding approver tendencies, prompt text

[1658] Output: Generated draft text

[1659] Specific operation: The server uses the OpenAI API to input prompt text and training results into the AI ​​model, which then generates suggested text.

[1660] Step 4:

[1661] The server sends the generated draft text to the terminal and presents it interactively.

[1662] Input: Generated draft text

[1663] Output: Draft text displayed on the terminal

[1664] Specific operation: The server sends the generated draft text to the terminal via an HTTP request, and the terminal displays it in the user interface.

[1665] Step 5:

[1666] The user reviews the draft text generated through their device and makes revisions as needed.

[1667] Input: Draft text displayed on the terminal

[1668] Output: Revised draft text

[1669] Specific operation: The user reviews the draft text using the terminal interface and makes revisions in a text editor as needed.

[1670] Step 6:

[1671] The server saves the revised draft as the final approved document.

[1672] Input: Revised draft text

[1673] Output: Saved final approval document

[1674] Specific operation: The server saves the revised draft text to the database and completes the approval process.

[1675] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1676] "Example of form 1"

[1677] In one embodiment of the present invention, the AI ​​includes an emotion engine that recognizes the user's emotions. This emotion engine estimates the user's emotions from the tone of their voice, facial expressions, and word choice. Specifically, it generates positive suggestions when the user is happy, and suggestions that include apologies when the user is angry.

[1678] "Example of form 2"

[1679] Furthermore, the emotion engine adjusts the draft text based on the user's emotions. Specifically, it generates draft text that includes positive language when the user is happy and apologies when the user is angry. For example, when the user is happy, it generates text such as "This proposal will lead your business to even greater success," and when the user is angry, it generates text such as "We deeply apologize that our service did not meet your expectations."

[1680] "Example of form 3"

[1681] Furthermore, the emotion engine automatically generates draft text for proposals and contracts based on the user's emotions. Specifically, it generates positive proposals when the user is happy and proposals that include apologies when the user is angry. For example, when the user is happy, it generates a proposal such as "This proposal will lead your business to even greater success," and when the user is angry, it generates a proposal such as "We deeply apologize that our service did not meet your expectations."

[1682] The following describes the processing flow for each example of the form.

[1683] "Example of form 1"

[1684] Step 1: The AI ​​activates an emotion engine that estimates the user's emotions based on their tone of voice, facial expressions, and word choice.

[1685] Step 2: The emotion engine recognizes the user's emotions and determines the content of the suggestions based on those emotions.

[1686] Step 3: Based on the results of the emotion engine, the AI ​​generates and presents suggestions to the user. (Example 2)

[1687] Step 1: The AI ​​activates an emotion engine that recognizes the user's emotions.

[1688] Step 2: The emotion engine recognizes the user's emotions and determines the content of the draft text based on those emotions.

[1689] Step 3: Based on the results of the emotion engine, the AI ​​generates and presents text drafts to the user.

[1690] "Example of form 3"

[1691] Step 1: The AI ​​activates an emotion engine that recognizes the user's emotions.

[1692] Step 2: The emotion engine recognizes the user's emotions and determines the content of the proposal and contract based on those emotions.

[1693] Step 3: Based on the results of the emotion engine, the AI ​​generates proposal and contract sections and presents them to the user.

[1694] (Example 1)

[1695] Next, we will describe Embodiment 1 of Embodiment Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1696] In traditional sales activities, salespeople had to manually refer to past contract and proposal information to create new proposals. This process was time-consuming and labor-intensive, increasing the workload on salespeople. Furthermore, it was difficult to make appropriate proposals that resonated with the user's emotions, making it challenging to improve customer satisfaction.

[1697] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1698] In this invention, the server includes means for acquiring information from a database storing past information, means for training a machine learning model using the acquired information, means for automatically generating draft text for proposals and contracts using the trained machine learning model, means for presenting the generated draft text through an interactive interface, means including an emotion engine that recognizes the user's emotions, and means for adjusting the draft text based on information obtained from the emotion engine. This reduces the burden on salespeople and makes it possible to provide optimal proposals that are tailored to the user's emotions.

[1699] A "database" is an information management system that stores historical information and allows for the efficient retrieval of necessary information.

[1700] A "machine learning model" is an algorithm that learns patterns and rules based on data and uses them to make predictions and classifications on new data.

[1701] "Training" is the process of improving the performance of a machine learning model by having it learn from a large amount of data.

[1702] A "draft document" is a draft of a document that outlines the contents of a proposal or contract.

[1703] An "interactive interface" is a user interface that allows the user and the system to communicate in both directions.

[1704] An "emotion engine" is a system that estimates a user's emotions based on their tone of voice, facial expressions, and word choice.

[1705] "Adjustment" means changing the content or settings according to specific conditions or circumstances.

[1706] Modes for carrying out the invention

[1707] This invention is a system that retrieves information from a database containing historical data and automatically generates draft text for proposals and contracts using machine learning. A specific embodiment of this system is described below.

[1708] Retrieve information from the database.

[1709] The server retrieves the necessary information from a database that stores information on past contracts and proposals made by corporate sales representatives. This database includes contract details, proposal details, and customer responses. The server uses SQL queries to extract past contract information from the database. For example, it might execute a query like "SELECT FROM contracts WHERE date > '2020-01-01'".

[1710] Training machine learning models

[1711] The server uses the acquired information to train a machine learning model. This model has the ability to automatically generate draft text for proposals and contracts using techniques such as deep learning and reinforcement learning. Specifically, it uses a framework such as TensorFlow. The server builds a deep learning model using TensorFlow and trains the model using historical contract data. For example, it executes code like "model.fit(training_data, epochs=50)".

[1712] Generating a draft document

[1713] The server uses a trained machine learning model to generate draft text for the proposal and contract sections. This draft text is designed to include optimal content based on historical data. The server prompts the generation AI model with a text prompt and generates the draft text. For example, it executes code like "generate_text('Generate a new proposal text')".

[1714] The interface presents draft text.

[1715] The terminal presents the generated message draft to the salesperson through an interactive interface. This interface is implemented in the form of a chatbot or voice assistant. The terminal displays the generated message draft in the chatbot's UI. For example, it executes code such as "chatbot.display_message(generated_text)".

[1716] User emotion recognition by an emotion engine

[1717] The device recognizes the user's emotions using an emotion engine. This emotion engine estimates emotions from the user's tone of voice, facial expressions, and word choice. The device uses OpenCV to analyze the user's facial expressions and estimate emotions. For example, it executes code like "emotion = emotion_recognition.detect_face_expression(user_image)".

[1718] Adjusting the wording to suit the emotions.

[1719] The server generates text suggestions tailored to the user's emotions based on information obtained from the emotion engine. When the user is happy, it generates positive suggestions; when the user is angry, it generates suggestions that include apologies. Based on the output of the emotion engine, the server inputs appropriate prompt sentences to the generation AI model. For example, it might execute code like "generate_text('Generate a suggestion sentence for when the user is angry')".

[1720] Specific example

[1721] Example 1: Generating a proposal

[1722] Based on the user's past contract information, the following proposal statement is generated.

[1723] Example of a prompt:

[1724] "Please generate a new proposal based on past contract information."

[1725] Example of a generated proposal:

[1726] "Customer, based on your past contract, we would like to offer you a new proposal with similar terms. We appreciate your consideration."

[1727] Example 2: Adjusting the wording to suit the emotions.

[1728] If the user is angry, the emotion engine recognizes that emotion and generates a draft message that includes an apology.

[1729] Example of a prompt:

[1730] "Please generate a suggestion message for when the user is angry."

[1731] Example of a generated proposal:

[1732] "We sincerely apologize for the inconvenience caused, customer. In this proposal, we have addressed past issues and will offer you better terms."

[1733] In this way, the system utilizes past data to automatically generate optimal suggestions tailored to the user's emotions.

[1734] The flow of the specific processing in Example 1 will be explained using Figure 17.

[1735] Step 1:

[1736] The server retrieves information from a database that stores historical data.

[1737] Input: Database query (Example: "SELECT FROM contracts WHERE date > '2020-01-01'")

[1738] Data processing: Execute SQL queries to extract past contract information.

[1739] Output: Dataset of acquired contract information

[1740] Specific operation: The server extracts historical contract information from the database using SQL queries. For example, it executes a query like "SELECT FROM contracts WHERE date > '2020-01-01'".

[1741] Step 2:

[1742] The server uses the acquired information to train a machine learning model.

[1743] Input: Header of acquired contract information

[1744] Data processing: Preprocess the dataset and convert it into a format suitable for machine learning models.

[1745] Output: Trained machine learning model

[1746] Specific operation: The server uses TensorFlow to build a deep learning model and trains the model using historical contract data. For example, it executes code like "model.fit(training_data, epochs=50)".

[1747] Step 3:

[1748] The server generates draft text for the proposal and contract sections using a trained machine learning model.

[1749] Input: A trained machine learning model, a prompt sentence (e.g., "Generate a new suggestion sentence")

[1750] Data processing: Input prompt messages into the model and generate suggested text options.

[1751] Output: Generated draft text

[1752] Specific operation: The server takes a prompt message as input to the generation AI model and generates a text proposal. For example, it executes code like "generate_text('Please generate a new proposal text')".

[1753] Step 4:

[1754] The terminal presents the generated draft text to the salesperson through an interactive interface.

[1755] Input: Generated draft text

[1756] Data processing: Convert the draft text into a format suitable for an interactive interface.

[1757] Output: Draft text to be presented to the business operator

[1758] Specific operation: The device displays the generated message draft in the chatbot's UI. For example, it executes code like "chatbot.display_message(generated_text)".

[1759] Step 5:

[1760] The device uses an emotion engine to recognize the user's emotions.

[1761] Input: User's tone of voice, facial expression, and word choice

[1762] Data processing: Analyze user emotions using an emotion engine.

[1763] Output: Estimated user sentiment

[1764] Specific operation: The device uses OpenCV to analyze the user's facial expressions and estimate their emotions. For example, it executes code like "emotion = emotion_recognition.detect_face_expression(user_image)".

[1765] Step 6:

[1766] The server generates text suggestions tailored to the user's emotions, based on information obtained from the emotion engine.

[1767] Input: Estimated user sentiment, trained machine learning model

[1768] Data processing: Generate prompt sentences that respond to emotions, input them into the model, and generate suggested sentences.

[1769] Output: Suggested text based on emotions

[1770] Specific operation: Based on the output of the emotion engine, the server inputs appropriate prompt text to the generative AI model. For example, it executes code such as "generate_text('Generate a suggestion text for when the user is angry')".

[1771] (Application Example 1)

[1772] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1773] Conventional sales support systems made it difficult for salespeople to quickly refer to past proposals and contract information when interacting with customers, and to make optimal proposals. Furthermore, it was difficult to recognize customer emotions in real time and respond accordingly. In addition, in physical stores, while store staff are required to make optimal proposals based on past sales data and proposal content, there was a lack of efficient means to do so. To solve these problems, a system is needed that recognizes customer emotions and generates proposal text that responds accordingly.

[1774] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1775] This invention includes a server that retrieves information from a database storing past information and automatically generates draft proposal and contract texts using machine learning, a means for interactively presenting the generated draft texts, an emotion engine that recognizes the user's emotions, a means for generating proposal texts that correspond to the user's emotions, and a means installed in smart glasses to support customer service. This enables salespeople and store clerks to quickly make optimal proposals based on past proposals and contract information when interacting with customers, and to make proposals that correspond to the customer's emotions in real time.

[1776] "Past information" refers to data that includes information on contracts and proposals made by corporate sales representatives in the past.

[1777] A "database" is a system that stores historical information and allows you to retrieve that information as needed.

[1778] "Artificial intelligence" is a technology that uses machine learning to automatically generate draft text for proposals and contracts.

[1779] "Interactive presentation methods" refer to methods of presenting generated draft text to salespeople through chatbots, voice assistants, or similar means.

[1780] An "emotion engine" is a technology that estimates emotions from a user's tone of voice, facial expressions, and word choice.

[1781] "Means for generating suggestion text that responds to user emotions" refers to a method for generating appropriate suggestion text based on emotions estimated by an emotion engine.

[1782] "Smart glasses" are wearable devices equipped with displays and cameras that can display information and capture video.

[1783] "Means of supporting customer service" refers to a method of using smart glasses to display optimal suggestion text when store employees interact with customers.

[1784] The system for carrying out this invention includes a database storing past information, artificial intelligence that automatically generates draft text for proposals and contracts using machine learning, means for interactively presenting the generated draft text, an emotion engine that recognizes the user's emotions, means for generating proposal text that corresponds to the user's emotions, and means installed in smart glasses to support customer service.

[1785] The server retrieves information from a database containing historical data and automatically generates draft proposal and contract texts using machine learning. Specifically, it uses deep learning models and reinforcement learning models trained with TensorFlow. The generated draft texts are presented to sales representatives through interactive interfaces such as chatbots and voice assistants.

[1786] The device uses smart glasses to capture the customer's facial expressions and voice tone, and an emotion engine recognizes the user's emotions. The emotion engine uses an emotion recognition model based on TensorFlow to estimate emotions from the user's voice tone, facial expressions, and word choice. Based on the estimated emotions, a generative AI model is used to generate the most appropriate suggested text. The generative AI model uses a natural language generation model such as GPT-2.

[1787] When a user wears smart glasses and interacts with customers, the smart glasses' display shows optimal proposal text based on past sales data and previous suggestions. This allows the user to make suggestions tailored to the customer's emotions in real time.

[1788] For example, if a customer is smiling while looking at a product, the emotion recognition model will determine that the customer is "happy" and generate a suggestion message such as, "This product is especially popular. We are currently offering a special discount, so please give it a try." Conversely, if a customer has a confused expression, the emotion recognition model will determine that the customer is "confused" and generate a suggestion message such as, "If there is anything we can help you with, please let us know."

[1789] Examples of prompt messages include the following:

[1790] "The optimal proposal text for when the customer is happy: Past proposal data"

[1791] "The best proposal text when a customer is confused: Past proposal data"

[1792] This allows salespeople and staff to quickly provide optimal proposals based on past proposals and contract information when interacting with customers, and to make proposals that are tailored to the customer's emotions in real time.

[1793] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[1794] Step 1:

[1795] The server retrieves information from a database that stores historical data.

[1796] Input: Past proposals and contract information stored in the database

[1797] Output: Retrieved past proposals and contract information

[1798] Specific operation: The server uses SQL queries to extract the necessary information from the database and load it into memory.

[1799] Step 2:

[1800] The server uses machine learning to automatically generate draft text for proposals and contracts based on the acquired information.

[1801] Input: Past proposals and contract information

[1802] Output: Generated proposal / contract draft

[1803] Specific operation: The server inputs data into deep learning models and reinforcement learning models trained using TensorFlow, and generates optimal text suggestions.

[1804] Step 3:

[1805] The server presents the generated draft text to the salesperson through an interactive interface.

[1806] Input: Generated proposal / contract draft

[1807] Output: Draft text presented to the salesperson

[1808] Specific operation: The server sends the generated draft text to the salesperson's terminal via a chatbot or voice assistant and displays it.

[1809] Step 4:

[1810] The device uses smart glasses to capture the customer's facial expressions and voice tone.

[1811] Input: Customer's facial expressions and tone of voice

[1812] Output: Data of captured customer facial expressions and voice tone.

[1813] Specific operation: The camera and microphone built into the smart glasses on the device are used to capture the customer's facial expressions and voice in real time.

[1814] Step 5:

[1815] The device inputs the captured data into an emotion engine to recognize the user's emotions.

[1816] Input: Data of captured customer facial expressions and voice tone.

[1817] Output: Estimated user sentiment

[1818] Specific operation: The device inputs data into an emotion recognition model using TensorFlow and estimates the user's emotions.

[1819] Step 6:

[1820] The server generates the optimal suggested text using a generative AI model based on the estimated emotions.

[1821] Input: Estimated user sentiment, past suggestion data

[1822] Output: Suggestion text tailored to your emotions

[1823] Specific operation: The server inputs a prompt sentence into a natural language generation model such as GPT-2 and generates the most suitable suggested sentence.

[1824] Step 7:

[1825] The device disp...

Claims

1. A system comprising a data processing device having a processor and a terminal, The processor of the data processing device is From a database that stores past information for each approver ID, retrieve the past approval history and approver trends corresponding to the specified approver ID. Based on the acquired past approval history and the approver's tendencies, a first prompt sentence is generated for the AI ​​model to generate a draft document that is likely to satisfy the approver. The generated first prompt sentence is input to the generation AI model to generate the draft text. The generated draft text is transmitted to the terminal. A system that includes this.

2. The processor is The system retrieves data from the database that includes information on past contracts and proposals made by the user of the terminal. The emotion engine estimates the customer's emotions based on their tone of voice, facial expressions, and word choice. Based on the aforementioned data and the customer's emotions, a second prompt sentence is generated to cause the AI ​​model to generate a suggestion text that corresponds to the customer's emotions. The generated second prompt sentence is input to the generation AI model to generate the proposed sentence. The generated proposed text is displayed on the display of the smart glasses, which serve as the terminal. The system according to claim 1.

3. The processor is If the customer's emotion indicates happiness, the second prompt statement is adjusted to output a suggestion statement containing positive words; if the customer's emotion indicates anger, the second prompt statement is adjusted to output a suggestion statement containing words of apology. The system according to claim 2.

Images