Implementation method, system and equipment of sales support tool and storage medium

Abstract

The invention relates to an implementation method, system and device of a sales support tool and a storage medium, and the method comprises the steps that a question sent by a user can be received, a question text corresponding to the question is sent to a first large language model, and the first large language model is used for providing a strategy for solving the question; receiving a target tool which is determined by the first large language model according to the question text and needs to be used for solving the question; calling the target tool to determine target information required for solving the user problem; a question text corresponding to the question and target information needed for solving the user question are sent to a second big language model, and the second big language model is used for obtaining reply content for the question text; and receiving the reply content sent by the second large language model, and sending the reply content to the user. According to the invention, the working efficiency can be improved, and the confidentiality of user data information is ensured.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular, to a method, system, device, and storage medium for implementing a sales support tool. Background Art

[0002] Currently, in the business scenarios of insurance companies, there are different roles such as customers, agents, and sales support staff. Agents provide services such as product explanations, demand analysis, and business handling for customers, with the ultimate goal of enabling customers to purchase insurance. The sales support staff provides support for agents, such as coordinating various professional resources, answering various professional questions, analyzing data, and statistics on the business development of each agent.

[0003] With the development of artificial intelligence technology, more and more enterprises are gradually exploring the combination of business and artificial intelligence. Therefore, there is a need to provide an artificial intelligence-based sales support tool to replace the work of traditional sales support positions. The sales support tool can answer various professional questions through an artificial intelligence model and complete some data statistical analysis work. The large language model (LLM) is a general language generation model trained using deep learning technology and can generate relevant output text based on the input text. Through the large language model, question answers matching the user's questions can be output. However, since a lot of data in the insurance business contains user information and is not convenient to be directly sent to an external large language model for analysis, therefore, how to develop a sales support tool suitable for the insurance industry is a technical problem that urgently needs to be solved. Summary of the Invention

[0004] Based on this, the present application provides a method, system, device, and storage medium for implementing a sales support tool to solve the problems existing in the prior art.

[0005] In a first aspect, a method for implementing a sales support tool is provided, and the method includes:

[0006] Receiving a question sent by a user, and sending the question text corresponding to the question to a first large language model, where the first large language model is used to provide a strategy for solving the question;

[0007] Receiving a target tool determined by the first large language model for solving the question based on the question text;

[0008] Invoking the target tool to determine the target information required to solve the user's question;

[0009] Sending the question text corresponding to the question and the target information required to solve the user's question to a second large language model, where the second large language model is used to obtain a reply content for the question text;

[0010] Receive the reply content sent by the second large language model and send the reply content to the user.

[0011] According to an implementable manner in the embodiments of the present application, the receiving the question sent by the user and sending the question text corresponding to the question to the first large language model includes:

[0012] Based on the chain of thought reasoning strategy, disassemble the question text into a number of reasoning questions;

[0013] Send the number of reasoning questions to the first large language model.

[0014] According to an implementable manner in the embodiments of the present application, the receiving the question sent by the user and sending the question text corresponding to the question to the first large language model further includes:

[0015] Determine a set of tools to be used corresponding to the number of reasoning questions, where the set of tools to be used includes local resource data and a search engine tool;

[0016] Generate a prompt according to the number of reasoning questions and the set of tools to be used;

[0017] Input the prompt into the first large language model.

[0018] According to an implementable manner in the embodiments of the present application, the set of tools to be used includes: a vector knowledge base tool and a business system tool;

[0019] Among them, the vector knowledge base tool includes local resource data and a semantic recognition tool, and the business system tool includes local resource data and a search engine tool.

[0020] According to an implementable manner in the embodiments of the present application, the first large language model is used to provide a strategy for solving the problem, including:

[0021] The first large language model is used to receive the prompt;

[0022] According to the number of reasoning questions and the set of tools to be used included in the prompt, determine the target tools that need to be used to solve the number of reasoning questions;

[0023] According to the number of reasoning questions and the target tools, obtain a strategy for solving the problem.

[0024] According to an implementable manner in the embodiments of the present application, the business system tool includes: a user information query tool and corresponding user information data, and an order query tool and corresponding order data.

[0025] In a second aspect, an implementation system of a sales support tool is provided, and the system includes:

[0026] A first transceiver module: configured to receive a problem sent by a user, and send the problem text corresponding to the problem to a first large language model, where the first large language model is used to provide a strategy for solving the problem;

[0027] A first large language module: configured to receive a target tool required to solve the problem determined by the first large language model according to the problem text;

[0028] An invocation module: configured to invoke the target tool to determine target information required to solve the user's problem;

[0029] A second large language module: configured to send the problem text corresponding to the problem and the target information required to solve the user's problem to a second large language model, where the second large language model is used to obtain a reply content for the problem text;

[0030] A second transceiver module: configured to receive the reply content sent by the second large language model, and send the reply content to the user.

[0031] According to an implementable manner in the embodiments of the present application, the first transceiver module is further configured to:

[0032] Based on a chain of thought reasoning strategy, disassemble the problem text into a plurality of reasoning problems;

[0033] Determine a set of tools to be used corresponding to the plurality of reasoning problems, where the set of tools to be used includes local resource data and a search engine tool;

[0034] Generate a prompt according to the plurality of reasoning problems and the set of tools to be used;

[0035] Input the prompt into the first large language model.

[0036] In a third aspect, a computer device is provided, including:

[0037] At least one processor; and

[0038] A memory communicatively connected to the at least one processor; wherein,

[0039] The memory stores computer instructions executable by the at least one processor, and the computer instructions are executed by the at least one processor so that the at least one processor can execute the method involved in the first aspect above.

[0040] Fourthly, a computer-readable storage medium is provided, on which computer instructions are stored, characterized in that the computer instructions are used to cause a computer to execute the method involved in the first aspect above.

[0041] According to the technical content provided by the embodiments of the present application, the present application receives a question sent by a user, sends the question text corresponding to the question to an external first large language model, and the first large language model provides a strategy for solving the question and determines the target tool required to solve the question; further, the target tool is called to determine the target information required to solve the user's question, and then the question text and the target information are sent to a second large language model deployed internally; further, the second large language model obtains a reply content for the question text; finally, the reply content is sent to the user. By combining the use of two large language models, the first large language model is a third-party large language model deployed externally in the cloud, which is only used to provide a strategy for solving the question, and then the core central controller queries data information internally, and then sends the data information and the question to the second large language model deployed internally to generate the final reply content, which can improve the efficiency of sales support work and reduce costs, while ensuring the confidentiality of user data information. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic flow chart of an implementation method of a sales support tool in an embodiment;

[0043] Figure 2 It is an architecture diagram of an implementation method of a sales support tool in an embodiment;

[0044] Figure 3 It is a structural block diagram of an implementation system of a sales support tool in an embodiment;

[0045] Figure 4 It is a schematic structural diagram of a computer device in an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The following further describes the present application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] Figure 1 It is a flow chart of an implementation method of a sales support tool provided by an embodiment of the present application. As Figure 1 shown, the method may include the following steps:

[0048] Step 101: Receive a question sent by a user, and send the question text corresponding to the question to the first large language model, where the first large language model is used to provide a strategy for solving the question.

[0049] Specifically, asFigure 2 As shown Figure 2 It is an architectural diagram of an implementation method of a sales support tool. This technical architecture provides three interaction windows, namely, a web interface serving the sales support position, a WeChat enterprise chat box serving agents, and an API interface serving agents.

[0050] An implementation method of a sales support tool provided by an embodiment of this application is applied to Figure 2 the core central controller in. The core central controller receives questions sent by users from three interaction windows, and sends the question text corresponding to the questions to the first large language model, where the first large language model is used to provide strategies for solving the questions. The first large language model is Figure 2 a third-party large language model in the cloud. The third-party large language model in the cloud can be launched by major technology companies currently, with a relatively large scale and a relatively high level of intelligence. We use it as an "advisor" to do complex problem analysis and decision-making, but they are not responsible for the execution of specific decisions. This part of the execution work needs to be completed by a self-deployed large model.

[0051] Step 102: Receive the target tool required to solve the problem determined by the first large language model according to the question text.

[0052] Specifically, the first large language model is the third-party large language model in the cloud. The third-party large language model is used to determine the target tool required to solve the problem according to the question text, and send the strategy for solving the problem to the core central controller. The core central controller receives the target tool required to solve the problem determined by the first large language model according to the question text.

[0053] For example, the core central controller sends the following question text to the first large language model: The user's question is "How old is Zhang San?" I now have the following several tools. What tool should I use to solve the user's problem? The tools include: user information query API; vector knowledge base; order query API.

[0054] The answer returned by the first large language model: You should use the [user information API] to query the user information of [Zhang San], and then obtain Zhang San's age from it.

[0055] The core central controller receives the target tool required to solve the problem determined by the first large language model according to the question text and the solution method "You should use the [user information API] to query the user information of [Zhang San], and then obtain Zhang San's age from it."

[0056] Step 103: Invoke the target tool to determine the target information required to solve the user's problem.

[0057] Specifically, the core central controller determines the target information applicable to solving the user's problem based on the target tool required to solve the problem sent by the first large language model. For example, if the target tool is [User Information API], the core central controller calls [User Information Query API], inputs [Zhang San], and obtains that the target information of Zhang San is "Zhang San, from Beijing, born in 2000, graduated from XXX University".

[0058] Step 104: Send the problem text corresponding to the problem and the target information required to solve the user's problem to the second large language model, where the second large language model is used to obtain a reply content for the problem text.

[0059] Specifically, the core central controller sends the problem text corresponding to the user's problem and the target information required to solve the user's problem obtained by calling the target tool to the second large language model. The second large language model is a self-deployed large language model that contains the specific business rules and professional knowledge within different organizations, which are usually not possessed by the first large language model. The second large language model is used to handle some professional problems within the organization or problems that contain customer information and are not convenient to be sent to the first large language model.

[0060] For example, the core central controller assembles a new problem and sends the following content to the second large language model: "Given that Zhang San is from Beijing, born in 2000, and graduated from XXX University. Question: How old is Zhang San?"

[0061] Step 105: Receive the reply content sent by the second large language model and send the reply content to the user.

[0062] Specifically, the second large language model generates a reply content based on the problem text sent by the core central controller and the target information required to solve the user's problem. For example, the second large language model receives the content "Given that Zhang San is from Beijing, born in 2000, and graduated from XXX University. Question: How old is Zhang San?" The second large language model generates the reply content "23 years old". The second large language model sends the reply content "23 years old" to the core central controller, and the core central controller receives the reply content sent by the second large language model and sends the reply content to the user.

[0063] It can be seen that in the embodiments of the present application, by receiving the questions sent by the user, the question text corresponding to the questions is sent to an external first large language model. The first large language model provides strategies for solving the questions and determines the target tools required to solve the questions. Further, the target tools are called to determine the target information required to solve the user's questions, and then the question text and the target information are sent to a second large language model deployed autonomously internally. Further, the second large language model generates a reply content for the question text. Finally, the reply content is sent to the user. By combining the use of two large language models in the present application, the first large language model is a cloud third-party large language model deployed externally, which is only used to provide strategies for solving questions. Then the core central controller queries data information internally, and then sends the data information and the questions to the second large language model deployed internally to generate the final reply content, which can improve the efficiency of sales support work and reduce costs, while ensuring the confidentiality of user data information.

[0064] In one embodiment of the present application, receiving the questions sent by the user and sending the question text corresponding to the questions to the first large language model in step 101 includes: decomposing the question text into a number of reasoning questions based on the chain of thought reasoning strategy; and sending the number of reasoning questions to the first large language model.

[0065] Specifically, the chain of thought, that is, the COT (Chain of thought, COT) chain of thought, refers to a series of logically related thinking steps that form a complete thinking process. Here, the chain of thought reasoning strategy is a reasoning strategy for decomposing complex problems by decomposing complex problems into one sub-problem after another, which reflects the reasoning process of complex problems. Receiving the questions sent by the user and sending the question text corresponding to the questions to the first large language model includes: decomposing the complex question text into a number of reasoning questions based on the chain of thought reasoning strategy; and sending the number of reasoning questions to the first large language model. The core central controller receives the request sent from the front end and performs operations such as identity authentication on it, and then performs a series of scheduling operations in combination with the COT chain of thought. The standardized LLM interaction interface usually adopts industry-standard technical framework components, such as fast Chat, which we use to encapsulate the differences of different large language models for facilitating our future switching of large language models in the architecture. The COT chain of thought includes the following parts: question (i.e., the original question provided by the customer), thought (a self-prompt word to guide the AI to think about what it should do), action (the set of tools that the LLM can use), action input (the parameters corresponding to the tools), Observation (the results obtained by using the tools), and final answer (the final answer to the original question).

[0066] In an embodiment of the present application, receiving the question sent by the user in step 101 and sending the question text corresponding to the question to the first large language model further includes: determining a tool set for use corresponding to a number of reasoning questions, where the tool set for use includes local resource data and a search engine tool; generating a prompt according to the number of reasoning questions and the tool set for use; and inputting the prompt into the first large language model.

[0067] Specifically, after the core central controller disassembles the question text into a number of reasoning questions, it also determines a tool set for use corresponding to the number of reasoning questions according to the number of reasoning questions. For example, the tool set for use includes tools such as a user information query API, a vector knowledge base, and an order query API. The core central controller generates a prompt according to the number of reasoning questions and the tool set for use, and inputs the prompt into the first large language model. Prompt represents a prompt word, which refers to displaying a prompt message to the user when running a program. Before waiting for the user to input information, the user is required to input information.

[0068] In a specific example: centered on the standardized API interface layer, it receives a request from the front end, such as "What is the regular deduction frequency of the insurance company?". Next, it will place this question in the COT template to form a complete prompt word. The general idea of this prompt word is as follows: The user asks what the regular deduction frequency of the insurance company is. To answer this question, what should I do? I have 2 tools, which are a vector knowledge base and a query interface for calling the order system respectively. The core central controller sends this prompt to an external large language model (the first large language model) for decision-making. The first large language model determines what tool to select to solve this problem according to the user's question and the preset thinking process.

[0069] In an embodiment of the present application, the first large language model in step 101 is used to provide a strategy for solving problems, including: the first large language model is used to receive the prompt; determine the target tool that needs to be used to solve a number of reasoning questions according to the number of reasoning questions and the tool set for use included in the prompt; and obtain a strategy for solving the problem according to the number of reasoning questions and the target tool.

[0070] Specifically, the first large language model determines what tool to select to solve this problem according to the user's question and the preset thinking process, and obtains a strategy for solving the problem according to the number of reasoning questions and the target tool.

[0071] For example, the first large language model receives a prompt, and the general idea of this prompt is as follows: The user asks what is the frequency of the insurance company's regular deductions? To answer this question, what should I do? I have two tools, namely a vector knowledge base and a query interface for calling the order system. The first large language model analyzes the problem-solving strategy: "You should query the vector knowledge base", and the target tool is the "vector knowledge base".

[0072] By decomposing complex problems through the COT thinking chain in the embodiments of the present application, the efficiency of problem-solving can be improved.

[0073] In an embodiment of the present application, the tool set used includes: a vector knowledge base tool and a business system tool; among them, the vector knowledge base tool includes local resource data and a semantic recognition tool, and the business system tool includes local resource data and a search engine tool.

[0074] Specifically, as Figure 2 shown, the vector knowledge base tool contains various types of company system documents and has the ability of semantic comparison. The vector knowledge base includes local resource data and a semantic recognition tool, and the semantic recognition tool is used to recognize semantics and perform matching according to semantics. The local resource data contains various types of company system document data, such as the attendance system, etc. The business system tool usually refers to a set of API interfaces opened to this architecture. The standardized API interface layer can call the interfaces opened by these business systems. The business system tool includes: a user information query tool and the corresponding user information data, and an order query tool and the corresponding order data. For example, a user information query API and user information data, an order query API and order data.

[0075] It can be seen that in the embodiments of the present application, by receiving the problem sent by the user, the problem text corresponding to the problem is sent to the external first large language model, and the first large language model provides the problem-solving strategy and determines the target tool required to solve the problem; further, the target tool is called to determine the target information required to solve the user's problem, and then the problem text and the target information are sent to the second large language model deployed internally; further, the second large language model obtains the reply content for the problem text; finally, the reply content is sent to the user. By combining the use of two large language models in the present application, the first large language model is a cloud third-party large language model deployed externally, which is only used to provide the problem-solving strategy, and then the core central controller queries the data information internally, and then sends the data information and the problem to the second large language model deployed internally to generate the final reply content, which can improve the efficiency of sales support work and reduce costs, while ensuring the confidentiality of user data information.

[0076] It should be understood that although Figure 1The steps in the flowchart are shown in sequence according to the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this application, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least a part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0077] Figure 3 FIG. is a schematic structural diagram of an implementation system of a sales support tool provided by an embodiment of this application. As Figure 3 shown, the system may include:

[0078] The first transceiver module 301: is configured to receive a problem sent by a user, and send the problem text corresponding to the problem to a first large language model, where the first large language model is used to provide a strategy for solving the problem;

[0079] The first large language module 302: is configured to receive a target tool required to solve the problem determined by the first large language model according to the problem text;

[0080] The calling module 303: is configured to call the target tool to determine the target information required to solve the user's problem;

[0081] The second large language module: is configured to send the problem text corresponding to the problem and the target information required to solve the user's problem to a second large language model, where the second large language model is used to obtain a reply content for the problem text;

[0082] The second transceiver module 304: is configured to receive the reply content sent by the second large language model, and send the reply content to the user.

[0083] In an embodiment of this application, the first transceiver module 301 is further configured to:

[0084] Based on the chain of thought reasoning strategy, disassemble the problem text into a number of reasoning problems;

[0085] Determine a set of tools to be used corresponding to the number of reasoning problems, where the set of tools to be used includes local resource data and a search engine tool;

[0086] Generate a prompt according to the number of reasoning problems and the set of tools to be used;

[0087] Input the prompt into the first large language model.

[0088] According to the specific embodiments provided in this application, the technical solutions provided in this application may have the following advantages:

[0089] By receiving the question sent by the user, the question text corresponding to the question is sent to the external first large language model. The first large language model provides a strategy for solving the problem and determines the target tools required to solve the problem. Further, the target tools are called to determine the target information required to solve the user's problem, and then the question text and the target information are sent to the second large language model deployed internally. Further, the second large language model generates a reply content for the question text. Finally, the reply content is sent to the user. In this application, two large language models are used in combination. The first large language model is a third-party large language model deployed externally in the cloud, which is only used to provide a strategy for solving the problem. Then the core central controller queries data information internally, and then sends the data information and the question to the second large language model deployed internally to generate the final reply content, which can improve the efficiency of sales support work and reduce costs, while ensuring the confidentiality of user data information.

[0090] For the same or similar parts between the above embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment.

[0091] It should be noted that the use of user data may be involved in the embodiments of this application. In actual applications, user-specific personal data can be used in the solutions described in this article within the scope permitted by applicable laws and regulations (such as when the user clearly consents, is effectively notified to the user, and the user clearly authorizes, etc.) in compliance with the applicable laws and regulations of the country where it is located.

[0092] According to the embodiments of this application, this application also provides a computer device and a computer-readable storage medium. This application also provides a computer device, including at least one processor and a memory communicatively connected to at least one processor; wherein, the memory stores computer instructions executable by at least one processor, and the computer instructions are executed by at least one processor so that at least one processor can execute the implementation method of the sales support tool described in any of the above embodiments.

[0093] Such as Figure 4As shown, it is a block diagram of a computer device according to an embodiment of the present application. The computer device is intended to represent various forms of digital computers or mobile systems. Among them, digital computers may include desktop computers, portable computers, workstations, personal digital assistants, servers, mainframe computers, and other suitable computers. Mobile systems may include tablet computers, smart phones, wearable devices, etc.

[0094] As Figure 4 shown, the computer device 400 includes a computing unit 401, a ROM 402, a RAM 403, a bus 404, and an input / output (I / O) interface 405. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via the bus 404. The input / output (I / O) interface 405 is also connected to the bus 404.

[0095] The computing unit 401 can execute various processes in the method embodiments of the present application according to the computer instructions stored in the read-only memory (ROM) 402 or the computer instructions loaded from the storage unit 408 into the random access memory (RAM) 403. The computing unit 401 can be various general-purpose and / or dedicated processing components with processing and computing capabilities. The computing unit 401 may include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. In some embodiments, the method provided by the embodiments of the present application can be implemented as a computer software program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 408.

[0096] The RAM 404 can also store various programs and data required for the operation of the device 400. Part or all of the computer programs can be loaded and / or installed onto the device 400 via the ROM 402 and / or the communication unit 409.

[0097] The input unit 406, the output unit 407, the storage unit 408, and the communication unit 409 in the computer device 400 can be connected to the I / O interface 405. Among them, the input unit 406 can be, for example, a keyboard, a mouse, a touch screen, a microphone, etc.; the output unit 407 can be, for example, a display, a speaker, an indicator light, etc. The device 400 can exchange information, data, etc. with other devices through the communication unit 409.

[0098] It should be noted that the device may also include other components necessary for normal operation. It may also only include the components necessary to implement the solution of the present application, and does not necessarily include all the components shown in the figure.

[0099] The various embodiments of the systems and techniques described herein can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof.

[0100] The computer instructions for implementing the methods of the present application can be written in any combination of one or more programming languages. These computer instructions can be provided to the computing unit 401 such that when the computer instructions are executed by the computing unit 401, such as a processor, the steps involved in the method embodiments of the present application are executed.

[0101] The present application also provides a computer-readable storage medium having stored thereon computer instructions for causing a computer to execute the implementation method of the sales support tool described in any of the above embodiments.

[0102] The computer-readable storage medium provided by the present application can be a tangible medium that can contain or store computer instructions for executing the steps involved in the method embodiments of the present application. The computer-readable storage medium can include, but is not limited to, storage media in the form of electronic, magnetic, optical, electromagnetic, and the like.

[0103] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present application shall be included within the protection scope of the present application.

Claims

1. A method for implementing a sales support tool, characterized in that, the method includes: Receiving a question sent by a user, and sending the question text corresponding to the question to a first large language model, wherein the first large language model is used to provide a strategy for solving the question; Receiving a target tool determined by the first large language model according to the question text for solving the question; Invoking the target tool to determine the target information required to solve the user's question; Sending the question text corresponding to the question and the target information required to solve the user's question to a second large language model, wherein the second large language model is used to obtain a reply content for the question text; Receiving the reply content sent by the second large language model, and sending the reply content to the user.

2. The method for implementing a sales support tool according to claim 1, characterized in that, the receiving a question sent by a user, and sending the question text corresponding to the question to a first large language model includes: Based on a chain of thought reasoning strategy, decomposing the question text into a number of reasoning questions; Sending the number of reasoning questions to the first large language model.

3. The method for implementing a sales support tool according to claim 2, characterized in that, the receiving a question sent by a user, and sending the question text corresponding to the question to a first large language model further includes: Determining a set of tools to be used corresponding to the number of reasoning questions, wherein the set of tools to be used includes local resource data and a search engine tool; Generating a prompt according to the number of reasoning questions and the set of tools to be used; Inputting the prompt into the first large language model.

4. The method for implementing a sales support tool according to claim 2, characterized in that, the set of tools to be used includes: a vector knowledge base tool and a business system tool; wherein the vector knowledge base tool includes local resource data and a semantic recognition tool, and the business system tool includes local resource data and a search engine tool.

5. The method for implementing a sales support tool according to claim 3, characterized in that, the first large language model is used to provide a strategy for solving the question, including: the first large language model is used to receive the prompt; According to the number of reasoning questions and the set of tools to be used included in the prompt, determining a target tool to be used for solving the number of reasoning questions; Deriving a strategy for solving the question according to the number of reasoning questions and the target tool.

6. The method for implementing a sales support tool according to claim 4, characterized in that, the business system tool includes: a user information query tool and corresponding user information data, and an order query tool and corresponding order data.

7. A system for implementing a sales support tool, characterized in that, the system includes: A first transceiver module: used for receiving a question sent by a user, and sending the question text corresponding to the question to a first large language model, wherein the first large language model is used to provide a strategy for solving the question; The first large language module: configured to receive the target tools required to solve the problem determined by the first large language model according to the problem text; The invocation module: configured to invoke the target tools to determine the target information required to solve the user's problem; The second large language module: configured to send the problem text corresponding to the problem and the target information required to solve the user's problem to the second large language model, wherein the second large language model is used to obtain the reply content for the problem text; The second transceiver module: configured to receive the reply content sent by the second large language model and send the reply content to the user.

8. The implementation system of the sales support tool according to claim 7, wherein, the first transceiver module is further configured to: Based on the chain of thought reasoning strategy, disassemble the problem text into a number of reasoning questions; Determine the set of tools to be used corresponding to the number of reasoning questions, wherein the set of tools to be used includes local resource data and search engine tools; Generate a prompt according to the number of reasoning questions and the set of tools to be used; Input the prompt into the first large language model.

9. A computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores computer instructions executable by the at least one processor, and the computer instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1-6.

10. A computer-readable storage medium, on which computer instructions are stored, wherein, the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 6.

Images