AI-based partner training dialogue method and device, and medium

Abstract

The invention provides an AI-based partner training dialogue method and device and a medium. The method comprises the following steps: reading a key and endpoint information from a configuration center in a configuration class injection mode, generating a remote client, acquiring script data from a third-party AI platform by using a remote adapter, and establishing a mapping binding relationship between the script data and a preset task in a local service system; a session is initiated on a front-end interaction interface through an integrated AI SDK, and user interaction data including voice or video is transparently transmitted by using the remote adapter in a multi-round dialogue drilling process; monitoring a session state in real time; and finally, asynchronously acquiring original evaluation data of the third-party platform through a timed task, and generating a multi-dimensional partner training evaluation report in combination with local business assessment dimensions. According to the invention, through integration of the third-party capability and deep decoupling of a local system, the maintenance cost of the system is reduced, a highly immersive actual combat drill environment is provided for property consultants, and a quantitative assessment basis is provided for managers.

Description

Technical Field

[0001] This invention relates to the field of real estate consultant training platform technology, and more specifically, to a method, device and medium based on AI-based training dialogue. Background Technology

[0002] In the field of real estate marketing, the professional expression and communication skills of sales consultants are core factors determining project sales results. Current industry practice involves sales consultants learning relevant marketing scripts and then directly serving clients. However, due to a lack of real-world practical experience, this model reveals significant problems in actual application.

[0003] On the one hand, when real estate consultants are conveying the value of their pitches, they often demonstrate a lack of experience and an inability to effectively handle unexpected situations such as customer questions and counter-questions, which can easily lead to the loss of potential clients. On the other hand, project managers currently lack effective methods to quantify and assess the degree to which real estate consultants have mastered their pitches. Managers cannot accurately ascertain the depth of their consultants' understanding of business knowledge, nor can they grasp their adaptability and reaction mechanisms when facing real communication pressure.

[0004] This situation results in a slow and costly process for real estate consultants to accumulate practical experience. It also makes it difficult for project management to propose precise improvement measures targeting the weaknesses of real estate consultants, leading to a need to improve overall training efficiency and management effectiveness. Therefore, providing a technical means that can simulate real-world customer interaction scenarios, support high-frequency drills, and quantify evaluation data is a pressing issue that needs to be addressed in this field. Summary of the Invention

[0005] The present invention aims to solve at least one of the aforementioned technical problems existing in the prior art.

[0006] Therefore, the first aspect of the present invention provides a method based on AI-assisted dialogue.

[0007] A second aspect of the present invention provides a computer device.

[0008] A third aspect of the present invention provides a computer-readable storage medium.

[0009] This invention provides a method for AI-based tutoring dialogue, including: Task Configuration: Obtain the exercise configuration request, retrieve script data from a third-party AI platform via remote API, establish a mapping and binding relationship between the script data and preset tasks in the local business system, and generate the target exercise task; Session Initiation: In response to the start command for the target training task, the integrated AISDK is invoked in the front-end interactive interface to initialize the training environment, and a session creation request is initiated to the third-party AI platform according to the mapping and binding relationship; Dialogue Interaction: During multi-round dialogue drills, user interaction data collected at the front end is transmitted to the third-party AI platform through a preset remote adapter, and simulated feedback data returned by the third-party AI platform is received and displayed in real time. The user interaction data includes voice data or video data. Status monitoring: Listen to the status changes of the multi-round dialogue exercise. When the session end signal is detected, trigger the local task status to be written back and the exercise record to be synchronized. Report generation: The system asynchronously acquires the original evaluation data of the third-party AI platform for this session through a scheduled task, and analyzes and processes the original evaluation data in conjunction with local business assessment dimensions to generate a multi-dimensional coaching evaluation report.

[0010] The AI-based tutoring dialogue method according to the above-described technical solution of the present invention may also have the following additional technical features: In the above technical solution, the task configuration includes: Obtain the script library details of the third-party AI platform through the system resource library; Based on the script identifier returned from the front end, the corresponding third-party script is saved as a local file and bound to the exercise task in the local business system.

[0011] In the above technical solution, during the session initiation step, the front-end interactive interface is implemented using the SDK capabilities of the third-party AI platform embedded within H5.

[0012] In the above technical solution, in the dialogue interaction step, the interaction data includes audio stream data in the smart phone practice scenario, or audio and video composite stream data in the video practice scenario.

[0013] In the above technical solution, the construction logic of the remote adapter includes: The key and endpoint information of the third-party AI platform are read from the configuration center through configuration class injection. Generate a remote client and encapsulate a unified interface for the business layer to call. The interface includes SDK message forwarding, task creation, script query and report retrieval interfaces.

[0014] In the above technical solution, the status monitoring step includes: SDK messages are transmitted through the front-end interactive interface, and the session status is monitored in real time according to a preset strategy. When a session end or closing action is detected, the policy triggers a write-back update of the local task state.

[0015] In the above technical solution, during the report generation step, a scheduled task is used to batch retrieve sessions in a pending state to obtain raw evaluation data from the third-party AI platform.

[0016] In the above technical solution, the training evaluation report adopts a multi-dimensional scoring model, and the scoring dimensions include: comprehensive score, coverage of key assessment points, expression ability, coach comments, and targeted improvement suggestions.

[0017] The present invention proposes a computer device, including a processor and a memory, wherein the memory stores a computer program, and when the computer program is loaded and executed by the processor, it implements the AI-based tutoring dialogue method as described in any of the above technical solutions.

[0018] The present invention proposes a computer-readable storage medium storing a program that, when loaded by a processor, implements the AI-based tutoring dialogue method as described in any of the above technical solutions.

[0019] In summary, due to the adoption of the above-mentioned technical features, the beneficial effects of the present invention are: The present invention provides a method for AI-based coaching dialogue, the benefits of which are first reflected in the integration of third-party AI training tools and the use of data mapping between the cloud and local systems and remote invocation to complete script and task configuration. This significantly reduces the system's R&D and maintenance costs and effectively accelerates the implementation of AI training functions.

[0020] Secondly, by integrating SDK capabilities into the front end, this invention can initiate multi-round dialogue drills for different training tasks and quickly generate training reports, enabling the rapid application and practice of AI-assisted training capabilities. During this process, the system supports simulating real-world intelligent phone or video dialogue scenarios, creating a more immersive communication environment for real estate consultants. This helps them improve their responsiveness and experience in real customer interactions through targeted scenario drills, effectively preventing potential customer loss due to insufficient experience.

[0021] Furthermore, by setting up drills in different scenarios, this invention can quantify the sales consultants' mastery and comprehension of their sales pitches, as well as their response mechanisms to unexpected situations. Through a multi-dimensional scoring model, the system can generate detailed coaching comments and targeted improvement suggestions from dimensions such as overall score, key assessment points, and communication skills. This enables project managers to grasp the team's overall business performance in real time and propose improvement measures.

[0022] Finally, the script task viewing platform and training report query platform provided by this invention support real-time viewing and sharing of training results, which not only facilitates self-reflection by real estate consultants, but also enables management to understand the training situation in a timely manner and optimize and improve strategies accordingly, thus realizing a closed loop of training and management.

[0023] Additional aspects and advantages of the invention will become apparent in the following description or may be learned by practice of the invention. Attached Figure Description

[0024] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart of an AI-based tutoring dialogue method according to an embodiment of the present invention. Detailed Implementation

[0025] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0026] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0027] The following reference Figure 1 This describes the methods, devices, and media based on AI-assisted dialogue provided according to some embodiments of the present invention.

[0028] Some embodiments of this application provide a method based on AI-assisted dialogue.

[0029] like Figure 1 As shown, the first embodiment of the present invention proposes a method based on AI-assisted dialogue, which mainly includes five parts: task configuration, session initiation, dialogue interaction, status monitoring and report generation, corresponding to the following steps S1 to S5.

[0030] S1. Task Configuration: Obtain the exercise configuration request, retrieve script data from a third-party AI platform via remote API, establish a mapping and binding relationship between the script data and preset tasks in the local business system, and generate the target exercise task. S2. Session Initiation: In response to the start command for the target training task, the integrated AISDK is called in the front-end interactive interface to initialize the training environment, and a session creation request is initiated to the third-party AI platform according to the mapping and binding relationship. S3. Dialogue Interaction: During the multi-round dialogue exercise, the user interaction data collected at the front end is transmitted to the third-party AI platform through a preset remote adapter, and the simulated feedback data returned by the third-party AI platform is received and displayed in real time. The user interaction data includes voice data or video data. S4. Status monitoring: Listen to the status changes of the multi-round dialogue exercise. When the session end signal is detected, trigger the local task status to be written back and synchronize the exercise record. S5. Report Generation: The original evaluation data of the third-party AI platform for this session is obtained asynchronously through a scheduled task. The original evaluation data is then parsed and processed in conjunction with local business assessment dimensions to generate a multi-dimensional coaching evaluation report.

[0031] Step S1 primarily involves the in-depth implementation of the task configuration phase. In some embodiments, during initialization, the system securely reads the access credentials required by the third-party AI platform (such as Alibaba Cloud AICoach) from a preset configuration center via configuration class injection. These credentials include, but are not limited to, AccessKey ID, AccessKey Secret, endpoint information, and a specific version number. This sensitive information is loaded into memory, and the system generates a high-performance remote client instance. To decouple business logic from underlying communication, this embodiment constructs a unified remote adapter. This adapter acts as a protocol conversion layer, encapsulating complex OpenAPI calls into standardized methods that are easy for the local business layer to call. Through this adapter, the system can perform a series of core actions, including SDK message forwarding, creation of exercise tasks or sessions, real-time querying of third-party playlists and details, and acquisition of exercise evaluation reports.

[0032] Specifically, when a manager initiates a script configuration request in the local business system, the system calls the query interface of the aforementioned adapter to retrieve available script metadata from the cloud script library of the third-party AI platform. These scripts typically include the sales presentation background, customer personality settings (such as picky, professional, or easygoing), preset dialogue nodes, and key scoring points for the assessment. After obtaining this script data, managers can select a target script based on the sales strategy of the current project. The system then establishes a mapping and binding relationship between the unique identifier of the third-party script and the preset task ID (TaskID) in the local business system. It is worth noting that this binding process is not a simple recording; the system stores the core parameters of the third-party script as local files in a local resource database, thereby ensuring that the system can still maintain basic task display logic when there is no network or the third-party interface is abnormal. This "third-party script → local file" synchronization mechanism achieves a deep integration of cloud-based general capabilities and local private business needs.

[0033] In the session initiation phase described in step S2, this embodiment employs a front-end integration solution to optimize user experience. In some embodiments, the real estate consultant enters the exercise portal via a mobile app or WeChat mini-program. The front-end interactive interface is primarily built on the H5 technology stack and deeply embeds the SDK capabilities of a third-party AI platform (such as the AI ​​Coach Web SDK). When the user clicks the "Start Exercise" command, the front-end first checks the locally stored login token. After successful verification, the front-end calls the initialized SDK and, based on the aforementioned established mapping relationship, transmits the corresponding script ID to initiate a session creation request to the third-party platform. During this process, the local application service performs secondary verification on the request, records the session start time, user ID, and device environment information, and stores this data in conjunction with the session identifier (Session ID) returned by the third party for subsequent end-to-end state tracking.

[0034] In the actual dialogue interaction process corresponding to step S3, this disclosure supports multiple highly realistic scenarios. In some embodiments, in the intelligent telephone practice scenario, the front end will enable audio acquisition permissions, process the real estate consultant's voice input through a noise reduction algorithm, and convert it into audio stream data. In other embodiments, in the video practice scenario, the front end will simultaneously turn on the camera to capture the user's facial expressions, gestures, and background environment, forming composite audio and video stream data. This interaction data is transmitted to the interaction engine of a third-party AI platform in a very low-latency manner through a remote adapter. The third-party platform uses a large language model and an emotion computing model to simulate the tone, intonation, and logic of real customers for real-time feedback. For example, when a real estate consultant mentions the price in a sales presentation, the AI ​​may simulate a customer asking a question: "This price is not competitive compared to surrounding projects. What are your specific advantages?" This simulated feedback data is distributed in real time through the SDK and displayed on the front-end interface in the form of voice broadcasts or dynamic video images, thereby creating a highly immersive communication environment. In this interaction logic, the remote adapter also acts as a state machine, continuously monitoring the number of dialogue rounds, sensitive word coverage, and the logical flow of the dialogue.

[0035] To ensure the integrity of the exercise process, this embodiment employs a robust monitoring mechanism to correspond with the status monitoring in step S4. Specifically, the front-end SDK not only handles data transmission but also listens for session status changes in real time. When it detects that a user clicks "End Call," the connection is interrupted due to network fluctuations, or the conversation reaches the preset maximum number of rounds, the SDK immediately captures the corresponding end signal. The system's preset strategy automatically intercepts these actions and sends a status write-back instruction to the local server, updating the current exercise task's status from "In Progress" to "Pending Evaluation" or "Abnormal Termination." If the session is interrupted due to non-human factors, the system also supports status recovery and reconnection within a certain timeframe to ensure that the real estate consultant's exercise progress is not lost.

[0036] For evaluating the effectiveness of the exercise, this disclosure employs an asynchronous batch processing architecture in step S5. In some embodiments, considering that the third-party platform requires a certain amount of computation time to generate a detailed in-depth evaluation report, the system does not use a synchronous waiting method, but instead utilizes a scheduled task to process the data in the background. The scheduled task scans the session records in the local database that are in the "pending evaluation" state at fixed intervals (e.g., every minute). For records that meet the criteria, the scheduled task requests the corresponding raw evaluation data from the third-party platform through the adapter's reporting interface. This raw data includes raw indicators such as transcribed text dialogue, sentiment score, key knowledge point matching degree, and pronunciation clarity.

[0037] After acquiring the raw data, the core processing logic of this invention involves secondary processing and localized analysis of this data. The system combines a weighting system preset by the project management team, for example, setting the weight of "adaptability" at 40%, "accuracy of speech" at 30%, and "affability" at 30%, to calculate the final comprehensive score. The resulting AI coaching report adopts a multi-dimensional scoring model, enabling refined evaluation from dimensions such as overall performance, key assessment points (e.g., whether core selling points were mentioned), and expression ability (whether the speaking speed was too fast, whether there were too many interjections). In addition, the system will also provide targeted improvement suggestions based on the AI-generated coaching comments, such as: "When facing price inquiries, it is recommended to focus more on the product premium and subsequent property services, rather than repeatedly emphasizing the basic unit price."

[0038] Finally, based on the method of this invention, a feature-rich exercise report sharing and management platform can be provided. This platform integrates data visualization components, enabling project managers to view the overall exercise profile of the current project, including the activity level of real estate consultants, average scores, and radar charts of capabilities across various dimensions. Individual real estate consultants can view all past exercise records in their own files at any time and can generate long exercise charts or links with a single click to share with mentors or colleagues, further solidifying exercise effectiveness through interactive evaluation. Through this closed-loop design throughout the entire process, this invention not only solves the problem of insufficient practical experience among real estate consultants but also provides project managers with a precise and quantitative evaluation system, significantly improving the professional competence of the sales team and customer acquisition conversion efficiency.

[0039] Other embodiments of the present invention provide a computer device including a processor and a memory, wherein the memory stores a computer program that, when loaded and executed by the processor, implements the AI-based tutoring dialogue method as described in any of the above embodiments.

[0040] Some embodiments of the present invention provide a computer-readable storage medium storing a program that, when loaded by a processor, implements the AI-based tutoring dialogue method as described in any of the above embodiments.

[0041] In this specification, the illustrative expressions of the terms used do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0042] Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this invention shall be included within the scope of protection of this invention.

Claims

1. A method based on AI-assisted practice dialogue, characterized in that, include: Task Configuration: Obtain the exercise configuration request, retrieve script data from a third-party AI platform via remote API, establish a mapping and binding relationship between the script data and preset tasks in the local business system, and generate the target exercise task; Session Initiation: In response to the start command for the target training task, the integrated AISDK is invoked in the front-end interactive interface to initialize the training environment, and a session creation request is initiated to the third-party AI platform according to the mapping and binding relationship; Dialogue Interaction: During multi-round dialogue drills, user interaction data collected at the front end is transmitted to the third-party AI platform through a preset remote adapter, and simulated feedback data returned by the third-party AI platform is received and displayed in real time. The user interaction data includes voice data or video data. Status monitoring: Listen to the status changes of the multi-round dialogue exercise. When the session end signal is detected, trigger the local task status to be written back and the exercise record to be synchronized. Report generation: The system asynchronously acquires the original evaluation data of the third-party AI platform for this session through a scheduled task, and analyzes and processes the original evaluation data in conjunction with local business assessment dimensions to generate a multi-dimensional coaching evaluation report.

2. The method based on AI-assisted dialogue according to claim 1, characterized in that, The task configuration includes: Obtain the script library details of the third-party AI platform through the system resource library; Based on the script identifier returned from the front end, the corresponding third-party script is saved as a local file and bound to the exercise task in the local business system.

3. The method based on AI-assisted dialogue according to claim 1, characterized in that, In the session initiation step, the front-end interactive interface is implemented using the SDK capabilities of the third-party AI platform embedded in H5.

4. The method based on AI-assisted dialogue according to claim 1, characterized in that, In the dialogue interaction step, the interaction data includes audio stream data in a smart phone practice scenario, or audio and video composite stream data in a video practice scenario.

5. The method based on AI-assisted dialogue according to claim 1, characterized in that, The construction logic of the remote adapter includes: The key and endpoint information of the third-party AI platform are read from the configuration center through configuration class injection. Generate a remote client and encapsulate a unified interface for the business layer to call. The interface includes SDK message forwarding, task creation, script query and report retrieval interfaces.

6. The method based on AI-assisted dialogue according to claim 1, characterized in that, The status monitoring steps include: SDK messages are transmitted through the front-end interactive interface, and the session status is monitored in real time according to a preset strategy. When a session end or closing action is detected, the policy triggers a write-back update of the local task state.

7. The method based on AI-assisted dialogue according to claim 1, characterized in that, In the report generation step, a scheduled task is used to batch retrieve sessions in a pending state to obtain raw evaluation data from the third-party AI platform.

8. The method based on AI-assisted dialogue according to claim 1, characterized in that, The training evaluation report adopts a multi-dimensional scoring model, which includes: overall score, coverage of key assessment points, expression ability, coach comments, and targeted improvement suggestions.

9. A computer device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when loaded and executed by the processor, implements the AI-based coaching dialogue method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The system stores a program that, when loaded by a processor, implements the AI-based coaching dialogue method as described in any one of claims 1 to 8.

Images