A method and device for controlling an artificial intelligence workbench

Abstract

The present invention discloses a control method for an artificial agent AI workbench, including step 301, monitoring the working state of an AI dialogue robot; wherein the working state of the AI ​​dialogue robot includes the busy and idle state information of the AI ​​dialogue robot, and the dialogue record of the AI ​​dialogue robot when the AI ​​dialogue robot is in a busy state; step 302, receiving a display instruction, and based on the display instruction, displaying the label information of the AI ​​dialogue robot, the working state and dialogue record of the AI ​​dialogue robot on the display device of the artificial agent AI workbench; step 303, judging whether a preset pop-up screen trigger condition is met, and if the preset pop-up screen trigger condition is met, triggering pop-up screen scheduling; step 304, based on the pop-up screen scheduling, selecting an artificial agent, and displaying the pop-up screen information on the display device of the AI ​​workbench of the artificial agent. Through the method of the present invention, the coupling and scheduling between the robot customer service and the artificial agent can be coordinated to improve the efficiency of outbound calls.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of information processing technology, and in particular to a method, device, terminal device, and computer-readable storage medium for controlling an artificial agent AI workstation. Background Art

[0002] With the development of science and technology and the prosperity of the Internet industry, call centers are an important means for Internet companies to connect with customers in order to provide business support and service optimization and maintain communication with customers. Traditional call centers are mainly staffed by customer service personnel. In the information age, the number of online customers is very large, so the requirements for online customer service are very high, and customers need to be connected in time and respond quickly. To meet this demand, the traditional way is to purchase more lines from operators and expand customer service personnel, but this requires a lot of resources and labor costs. With the development of artificial intelligence technology, artificial intelligence has been applied in the customer service field of call centers, and intelligent customer service tools such as robot customer service have emerged. Many simple and repetitive services of manual seats can be replaced by intelligent voice customer service, saving labor costs. However, how to coordinate the coupling and scheduling between robot customer service and manual seats, increase the number of simultaneous dialing of the control center, increase the customer contact speed per unit time, reduce call losses, increase the intervention work time of manual seats, and realize efficient outbound call services are urgent problems to be solved in this field. Summary of the invention

[0003] In view of the problems in the prior art, the present invention provides a method, device, terminal equipment and computer-readable storage medium for controlling an artificial agent AI workstation.

[0004] The present invention provides an artificial agent AI workbench control method, which is characterized by:

[0005] Step 301, monitoring the working status of the AI ​​dialogue robot; wherein the working status of the AI ​​dialogue robot includes the busy and idle status information of the AI ​​dialogue robot, and the dialogue record of the AI ​​dialogue robot when the AI ​​dialogue robot is in a busy state;

[0006] Step 302: receiving a display instruction, and displaying the label information of the AI ​​dialogue robot, the working status of the AI ​​dialogue robot, and the dialogue record on the display device of the artificial agent AI workstation based on the display instruction;

[0007] Step 303, determining whether a preset pop-up screen triggering condition is met, and if the preset pop-up screen triggering condition is met, triggering pop-up screen scheduling;

[0008] Step 304: Based on the pop-up screen scheduling, a manual agent is selected, and the pop-up screen information is displayed on the display device of the AI ​​workstation of the manual agent.

[0009] The step 301 of monitoring the working status of the AI ​​dialogue robot includes:

[0010] Store the conversation records of the AI ​​conversation robot in real time. The conversation records may directly store voice information and / or store corresponding text information through text conversion;

[0011] Indexing the conversation record of the AI ​​conversation robot based on the AI ​​conversation robot ID;

[0012] The conversation information of the AI ​​conversation robot includes recorded voice information and text information converted in real time based on the voice confidence, and both the voice information and the text information are recorded with time marks.

[0013] The step 302, receiving a display instruction, and based on the display instruction, displaying the label information of the AI ​​dialogue robot, the working status of the AI ​​dialogue robot, and the dialogue record on the display device of the artificial agent AI workstation specifically includes:

[0014] When the AI ​​dialogue robot is in a busy state, a dialogue display instruction is triggered;

[0015] Based on the display indication, the dialogue interface corresponding to the AI ​​dialogue robot is displayed in the form of a card, and the dialogue interface corresponding to the AI ​​dialogue robot displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue records of the AI ​​dialogue robot.

[0016] The step 304 further includes

[0017] Step 305: Receive a first input for the pop-up screen information, where the first input is first operation information based on the pop-up screen information.

[0018] The pop-up screen reminder may be configured to include a first control, and when the first manual agent is unable to intervene, the pop-up screen scheduling may be triggered by operating the first control.

[0019] The present invention also provides an artificial agent AI workbench control device, which is characterized by:

[0020] A real-time conversation monitoring module monitors the working status of the AI ​​conversation robot; wherein the working status of the AI ​​conversation robot includes the busy and idle status information of the AI ​​conversation robot, and the conversation record of the AI ​​conversation robot when the AI ​​conversation robot is in a busy state;

[0021] A display control module receives a display instruction and displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue record of the AI ​​dialogue robot on a display device of the artificial agent AI workstation based on the display instruction;

[0022] A pop-up screen trigger judgment module is used to judge whether a preset pop-up screen trigger condition is met, and if the preset pop-up screen trigger condition is met, the pop-up screen scheduling is triggered;

[0023] The pop-up screen scheduling module selects a manual agent based on the pop-up screen scheduling, and displays the pop-up screen information on the display device of the AI ​​workbench of the manual agent.

[0024] The real-time conversation monitoring module further includes a conversation storage module, wherein the conversation storage module is used to

[0025] Store the conversation records of the AI ​​conversation robot in real time. The conversation records may directly store voice information and / or store corresponding text information through text conversion;

[0026] Wherein, the conversation record of the AI ​​conversation robot is indexed based on the AI ​​conversation robot ID;

[0027] The conversation information of the AI ​​conversation robot includes recorded voice information and text information converted in real time based on the voice confidence, and both the voice information and the text information are recorded with time marks.

[0028] The display control module is further used to

[0029] When the AI ​​dialogue robot is in a busy state, a dialogue display instruction is triggered;

[0030] Based on the display indication, the dialogue interface corresponding to the AI ​​dialogue robot is displayed in the form of a card, and the dialogue interface corresponding to the AI ​​dialogue robot displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue records of the AI ​​dialogue robot.

[0031] The device also includes

[0032] An input module receives a first input for the pop-up screen information, where the first input is first operation information based on the pop-up screen information;

[0033] The pop-up screen reminder may be configured to include a first control, and when the first manual agent is unable to intervene, the pop-up screen scheduling may be triggered by operating the first control.

[0034] The present invention also provides a terminal device, characterized in that the terminal device includes a processor and a memory, the memory stores a computer program that can be run on the processor, and the computer program implements the method as described above when executed by the processor.

[0035] The present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program that can be run on a processor, and the computer program implements the method as described above when executed.

[0036] Through the method of the present invention, the coupling scheduling between robot customer service and manual seats can be effectively coordinated, call losses can be reduced, outbound call efficiency can be improved, and user experience can be enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 The present invention is a human-machine coupling calling method in one embodiment of the present invention.

[0039] Figure 2 It is a system architecture of a human-machine coupling call in one embodiment of the present invention.

[0040] Figure 3 The diagram is a diagram showing a dialog interface in one embodiment of the present invention.

[0041] Figure 4 The figure is an illustration of the pop-up screen information display in one embodiment of the present invention.

[0042] Figure 5 The invention relates to a method for dispatching a human-machine coupled outbound call in one embodiment of the present invention.

[0043] Figure 6 The present invention provides an artificial agent AI workstation control method in one embodiment of the present invention.

[0044] Figure 7 The figure is a functional diagram of an artificial agent AI workstation in one embodiment of the present invention.

[0045] Figure 8 It is a human-machine coupling calling device in one embodiment of the present invention.

[0046] Fig. 9 The invention is a human-machine coupled outbound calling device in one embodiment of the present invention.

[0047] Fig.10 It is an artificial agent AI workbench device in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail with reference to the accompanying drawings. The embodiments of the present invention and the specific features of the embodiments are detailed descriptions of the technical solutions of the embodiments of the present invention, rather than limitations on the technical solutions of the present specification. In the absence of conflict, the embodiments of the present invention and the technical features of the embodiments can be combined with each other.

[0049] Embodiment 1

[0050] refer to Figure 1 , Embodiment 1 of the present invention provides a human-machine coupling calling method, characterized in that,

[0051] Step 101, determining whether multiple calls initiated at the same time are connected, if connected, executing step 102;

[0052] Step 102, distributing the connected one or more calls to an artificial intelligence AI dialogue robot, which makes an initial response, and the AI ​​dialogue robot has a binding relationship with the human agent;

[0053] Step 103, monitoring the response of the AI ​​dialogue robot to determine whether manual intervention is required; if manual intervention is required, executing step 104;

[0054] Step 104: Push a notification message to the human agent, prompting the human agent that the AI ​​dialogue robot needs to intervene in the dialogue.

[0055] Reference Figure 2 The control center can control the simultaneous dialing of multiple customer calls. When there is a connected call, the control center distributes it to the manual seat. The system includes M manual seats, such as Figure 2 The human agents A, B, C, D, etc. shown in the figure each correspond to an AI workstation, and N AI dialogue robots are configured for each human agent. For example, AI 1, AI 2, AI 3, AI 4 and other AI dialogue robots are bound to human agent A, that is, AI 1, AI 2, AI 3, AI 4 and other AI dialogue robots correspond to the AI ​​workstation of human agent A.

[0056] Specifically, in step 102, distributing the connected one or more calls to the artificial intelligence AI dialogue robot specifically includes:

[0057] Step 1021, query the AI ​​dialogue robot resource pool;

[0058] Step 1022: based on a predetermined rule, select one or more AI dialogue robots from the AI ​​dialogue robot resource pool.

[0059] Step 1023: distribute the connected one or more calls to the selected AI dialogue robot.

[0060] The AI ​​conversational robot resource pool may be pre-built, and the pre-building of the AI ​​conversational robot resource pool includes the following steps:

[0061] Poll the N AI dialogue robots corresponding to M human seats respectively,

[0062] Determine the busy / idle status of the manual agent, that is, whether the manual agent is idle or in conversation. If the manual agent is idle, mark the N AI dialogue robots corresponding to the manual agent respectively, for example, set the AI ​​dialogue robot flag, the AI ​​dialogue robot flag corresponding to the idle manual agent is "0", and the AI ​​dialogue robot flag corresponding to the busy manual agent is "1".

[0063] Determine whether the AI ​​dialogue robot is idle. If so,

[0064] The idle AI dialogue robot is added to the AI ​​dialogue robot resource pool.

[0065] The AI ​​dialogue robot resource pool is updated in real time.

[0066] Furthermore, the real-time updating of the AI ​​dialogue robot resource pool specifically includes:

[0067] Monitor the status of the AI ​​conversation robot in real time. When the AI ​​conversation robot releases the session and enters an idle state, add the AI ​​conversation robot that has entered the idle state to the AI ​​conversation robot resource pool;

[0068] When an AI dialogue robot in the resource pool is selected and enters a dialogue state, the selected AI dialogue robot is deleted from the resource pool.

[0069] Furthermore, an AI dialogue robot state array may be constructed, which includes parameter values ​​such as the AI ​​dialogue robot ID, the bound human agent ID, the current state of the AI ​​dialogue robot, and the current state of the bound human agent. As shown in Table 1 below:

[0070] Table 1: AI conversational robot status array

[0071]

[0072] Furthermore, the AI ​​dialogue robot state array may also include AI dialogue robot and / or human agent conversation party data; the conversation party data may further include conversation party account, conversation party attributes, conversation party level, current conversation category, etc.

[0073] For example, the AI ​​conversational robot AIA2 is currently in a conversation, the account of the conversation partner is 137XXXXXXX, the attribute of the conversation partner is a registered customer, the level of the conversation partner is intermediate, and the current conversation category is business promotion.

[0074] The current state of the AI ​​dialogue robot and the current state of the bound artificial agent in the AI ​​dialogue robot state array can be further refined into: current dialogue state and dialogue direction information.

[0075] For example, for the current state of the AI ​​conversation robot, the current conversation state can be in conversation or idle, and the conversation trend can be the need for manual intervention, such as the need for manual intervention is graded. For the current state of the human agent, the current conversation state can be in conversation or idle, and the conversation trend information can be continued communication, hang-up tendency, etc.

[0076] Furthermore, an AI dialogue robot status attribute field can be constructed, and the AI ​​dialogue robot status attribute field is indexed by the AI ​​dialogue robot ID to store parameter values ​​related to the AI ​​dialogue robot status attributes, for example, 1 or 2 bits are used to indicate the current state of the AI ​​dialogue robot, for example, "0" represents idle, that is, the idle state, and "1" represents the in-conversation, that is, the busy state; or further use "00" to represent the idle state, "11" to represent the busy state, "01" to represent the upcoming release, and "10" to represent waiting for intervention; set 3-4 bits to indicate the bound artificial agent ID; and use 1 or 2 bits to indicate the current state of the artificial agent.

[0077] Monitoring the response of the AI ​​dialogue robot in step 103 and determining whether manual intervention is required further includes:

[0078] The human agent bound to the AI ​​conversation robot monitors the conversation information of the AI ​​conversation robot and chooses whether to intervene;

[0079] The monitoring method is to monitor the real-time conversation text or voice information between AI and customers, and the customer's status can be obtained through the voice information.

[0080] Specifically, see Figure 3Through the AI ​​workstation corresponding to the human agent, the conversation process of each AI dialogue robot can be seen or heard in real time. The AI ​​workstation corresponding to the human agent can choose to intervene in the conversation of the AI ​​dialogue robot at any time based on monitoring the conversation process of the AI ​​dialogue robot.

[0081] For example, the conversation process of the AI ​​conversation robot is recorded in real time. The conversation process can directly store and play voice information, or be converted into text and presented in real time on the conversation interface corresponding to the AI ​​conversation robot.

[0082] The AI ​​workbench of the human agent extracts corresponding keyword data and / or user status information based on multiple rounds of conversations with the AI ​​conversation robot, and analyzes the intervention needs of the current conversation.

[0083] For example, based on multiple rounds of conversations with the AI ​​conversational robot, it is identified that Customer 1 has an intention to subscribe to the service and needs to further communicate the details. The AI ​​conversational robot cannot provide further services, so Customer 1 has an intervention need.

[0084] Furthermore, the intervention needs may be graded according to the urgency of the intervention needs, wherein the urgency may be determined according to one or more of the conversation duration, conversation content, conversation party attributes, conversation party level, and current conversation category.

[0085] Specifically, in step 104, the push notification message to the human agent further includes:

[0086] Push notification messages to human agents who have a binding relationship with the AI ​​conversational robot;

[0087] Specifically, see Figure 4 The AI ​​workbench can preset pop-up screen rules. When the AI ​​conversation enters a certain stage and requires human agent intervention, the AI ​​workbench will notify the human agent through a pop-up screen.

[0088] Furthermore, the pop-up screen is controlled to be displayed on a display device corresponding to the manual agent, wherein the display device synchronously displays one or more dialogue interfaces corresponding to the AI ​​dialogue robots corresponding to the manual agent.

[0089] Furthermore, the dialogue interface corresponding to the AI ​​dialogue robot that requires human intervention is controlled to be highlighted; wherein the highlighted display includes shaking of the dialogue interface, deepening of the dialogue interface border color, enlarged display of the dialogue interface, etc.

[0090] The method further includes step 105, determining whether the first human agent bound to the AI ​​dialogue robot is idle; if idle, executing step 106, otherwise executing step 107;

[0091] The step of determining whether the first human agent bound to the AI ​​dialogue robot is idle includes:

[0092] Query the AI ​​dialogue robot state array to determine the first human agent bound to the AI ​​dialogue robot;

[0093] Obtain the current status of the first human agent bound to the AI ​​dialogue robot.

[0094] Determine whether the first human agent bound to the AI ​​dialogue robot is idle;

[0095] Step 106: The first human agent replaces the AI ​​dialogue robot to complete the customer service dialogue.

[0096] Specifically, for example, when human agent A performs manual intervention, a new human agent dialogue interface is created, and the history record of the current dialogue of the AI ​​dialogue robot that is manually intervened is pushed to the human agent dialogue interface. The AI ​​dialogue robot resources are released, and the AI ​​dialogue robot state array and the AI ​​dialogue robot resource pool are updated.

[0097] Furthermore, in step 107, it is determined whether there is an idle second manual agent among other manual agents that have no binding relationship with the AI ​​dialogue robot. If there is an idle second manual agent, the idle second manual agent is notified, and the idle second manual agent takes over the AI ​​dialogue robot to complete the customer service dialogue.

[0098] Specifically, a notification message is pushed to the idle second manual agent, and the notification message is a pop-up message. When the idle second manual agent takes over the AI ​​dialogue robot, a second manual agent dialogue interface is created, and the historical record of the current dialogue of the AI ​​dialogue robot with manual intervention is pushed to the second manual agent dialogue interface.

[0099] Based on the above solution, it is possible to coordinate the coupling between robot customer service and human agents, increase the number of simultaneous dialing by the control center, reach customers faster per unit time, and increase output.

[0100] Embodiment 2

[0101] Based on the first embodiment, a method for scheduling a human-machine coupled outbound call is provided. Figure 5 , characterized in that,

[0102] Step 201, determining the condition parameters of the AI ​​dialogue robot, the condition parameters of the AI ​​dialogue robot include a first condition parameter, a second condition parameter, and a third condition parameter, wherein the first condition parameter indicates the state of the artificial seat corresponding to the AI ​​dialogue robot, the second condition parameter indicates the state of the AI ​​dialogue robot, and the third condition parameter indicates whether the AI ​​dialogue robot is in a key dialogue link;

[0103] Step 202: determining the candidate priority of the AI ​​dialogue robot according to the first condition parameter, the second condition parameter, and the third condition parameter;

[0104] Step 203, selecting N AI dialogue robots that meet the distribution conditions according to the candidate priorities of the AI ​​dialogue robots;

[0105] Step 204, generate a distribution instruction to distribute the connected lines to the N AI dialogue robots that meet the distribution conditions.

[0106] The state of the human agent and the state of the AI ​​dialogue robot corresponding to the human agent in step 201 can be obtained through the AI ​​dialogue robot state array;

[0107] Or obtain the AI ​​conversation robot status corresponding to the human agent through the AI ​​conversation robot resource pool; determine the human agent status through the AI ​​conversation robot flag in the AI ​​conversation robot resource pool, or determine the human agent status by polling the human agent.

[0108] Preferably, the step 201 specifically further includes:

[0109] Step 2012, determining whether the artificial seat meets the first preset condition, if there is an artificial seat that meets the first preset condition, executing step 2013,

[0110] Step 2013: the AI ​​dialogue robots corresponding to the human agents that meet the first preset condition form a first candidate cluster; the AI ​​dialogue robots corresponding to the human agents that do not meet the first preset condition form a second candidate cluster;

[0111] Step 2014, determining whether the AI ​​dialogue robots in the first candidate cluster and the second candidate cluster meet the second preset condition;

[0112] Step 2015: performing secondary clustering on the first candidate cluster and the second candidate cluster according to whether the AI ​​dialogue robot meets the second preset condition, and performing a first grouping mark on the AI ​​dialogue robot;

[0113] Step 2016, determine whether the AI ​​dialogue robot that does not meet the second preset condition meets the third preset condition; based on the third preset condition, mark the AI ​​dialogue robot as a first sub-group.

[0114] Specifically, the first preset condition is that the human agent is idle, the second preset condition is that the AI ​​dialogue robot is idle, and the third preset condition is that the AI ​​dialogue robot is in a critical dialogue link.

[0115] Specifically, the step 2013 further includes constructing candidate tag information of the AI ​​dialogue robot based on the clustering.

[0116] The candidate tag information of the AI ​​dialogue robot includes the AI ​​dialogue robot ID and the first-level clustering value. For example, the AI ​​dialogue robots AIN1~Ni corresponding to the artificial seats that meet the first preset condition form the first candidate cluster; the AI ​​dialogue robots AI P1~Pj corresponding to the artificial seats that do not meet the first preset condition form the second candidate cluster; the corresponding AI dialogue robot candidate tag information is generated and stored: {AI dialogue robot ID, first-level clustering value}, for example {AIN2, 0}, {AI P1, 1}, where the first-level clustering value "0" represents that the first preset condition is met, that is, the artificial seat corresponding to the AI ​​dialogue robot is idle, and the first-level clustering value "1" represents that the first preset condition is not met, that is, the artificial seat corresponding to the AI ​​dialogue robot is busy.

[0117] If the number of AI dialogue robots that meet / do not meet the first preset condition is 0, the corresponding candidate cluster members are empty.

[0118] Specifically, the step 2015, according to whether the AI ​​dialogue robot meets the second preset condition, performs secondary clustering on the first candidate cluster and the second candidate cluster, and performs the first grouping marking on the AI ​​dialogue robot further includes:

[0119] According to the second preset condition, the first candidate cluster is divided into a first candidate group and a second candidate group; and the second candidate cluster is divided into a third candidate group and a fourth candidate group.

[0120] If the number of AI dialogue robots that meet / do not meet the second preset condition is 0, the corresponding candidate group members are empty.

[0121] Preferably, the corresponding AI dialogue robot candidate tag information is updated and stored according to the first grouping tag;

[0122] Specifically, after making a judgment on the second preset condition, for example, the AI ​​dialogue robots AI N1~Ni corresponding to the manual seats that meet the first preset condition; the AI ​​dialogue robots AI P1~Pj corresponding to the manual seats that do not meet the first preset condition; are further divided into multiple groups, and the corresponding AI dialogue robot candidate label information is updated and stored: {AI dialogue robot ID, first-level clustering value, second-level grouping value}, for example {AI N2, 0, 1}, {AI P1, 1, 0}, where the second-level grouping value "0" represents that the second preset condition is met, that is, the AI ​​dialogue robot is currently idle, and the second-level grouping value "1" represents that the second preset condition is not met, that is, the AI ​​dialogue robot is busy.

[0123] Specifically, step 2016, determining whether the AI ​​dialogue robot that does not meet the second preset condition meets the third preset condition; based on the third preset condition, marking the AI ​​dialogue robot with a first sub-group specifically includes:

[0124] The current conversation record of the AI ​​conversation robot that does not meet the second preset condition is obtained in real time. Specifically, the conversation process of each AI conversation robot can be watched or listened to in real time through the AI ​​workstation corresponding to the artificial seat.

[0125] For example, the conversation process of the AI ​​conversation robot is recorded in real time. The conversation process can directly store and play voice information, or be converted into text and presented in real time on the conversation interface corresponding to the AI ​​conversation robot.

[0126] The AI ​​workbench of the human agent extracts corresponding keyword data and / or user status information based on multiple rounds of conversations with the AI ​​conversation robot, and analyzes the intervention needs of the current conversation.

[0127] For example, the actual intention of the user can be determined by extracting the corresponding keyword data in the conversation, or the current status of the user can be determined by analyzing the tone and voice in the user's voice information. For example, if the user has the intention to further communicate and subscribe, but is in an irritable state in the machine conversation, it is determined that the current AI conversation robot is in a critical conversation link, that is, the customer has a need to intervene.

[0128] Furthermore, the intervention needs may be graded according to the urgency of the intervention needs, wherein the urgency may be determined according to one or more of the conversation duration, conversation content, conversation party attributes, conversation party level, and current conversation category.

[0129] When it is determined that the current AI dialogue robot is in a critical dialogue link, a first sub-grouping mark is performed on the AI ​​dialogue robot in the critical dialogue link.

[0130] For example, for the first candidate grouping, the second candidate grouping, the third candidate grouping, and the fourth candidate grouping further grouped after the second preset condition judgment, the AI ​​dialogue robot in the key dialogue link in the group is marked as a first sub-grouping, and the corresponding AI dialogue robot candidate marking information is updated and stored: {AI dialogue robot ID, first-level clustering value, second-level grouping value, third-level sub-grouping value}, for example, {AI N2, 0, 1, 1}. Preferably, the AI ​​dialogue robot candidate marking information of other AI dialogue robots that have not been marked as the first sub-grouping can be synchronously updated, for example, {AI P1, 1, 0, 0}, or {AI N1, 0, 1, 0}; the third-level sub-grouping value "1" represents that the third preset condition is met, that is, the AI ​​dialogue robot is currently in the key dialogue link, and the third-level sub-grouping value "0" represents that the third preset condition is not met, that is, the AI ​​dialogue robot is busy but not in the key dialogue link. In addition, when the second-level grouping value "0" represents that the second preset condition is met, that is, the AI ​​dialogue robot is currently idle, the third-level sub-grouping value is updated to "0" based on the default.

[0131] Specifically, the step 202, determining the candidate priority of the AI ​​dialogue robot according to the first condition parameter, the second condition parameter, and the third condition parameter further includes:

[0132] Obtain candidate tag information of the AI ​​dialogue robot, and determine the priority of the candidate AI dialogue robot based on the first-level clustering value, the second-level grouping value, and the third-level sub-grouping value in the candidate tag information of the AI ​​dialogue robot.

[0133] The first condition parameter, the second condition parameter, and the third condition parameter are respectively mapped to the first-level clustering value, the second-level grouping value, and the third-level sub-grouping value.

[0134] For example, for {AI N2, 0, 1, 1}, {AI P1, 1, 0, 0}, {AI N1, 0, 1, 0};

[0135] According to the first-level clustering value, the second-level grouping value, and the third-level sub-grouping, 3-bit information is determined: 000, 010, 011, 100, 110, 111; for example, three levels of priority are set, where 000 is the first priority, 100 is the second priority, 011 is the third priority, and 110 and 111 are the fourth priority.

[0136] Preferably, the step 202 of determining the candidate priority of the AI ​​dialogue robot according to the first condition parameter, the second condition parameter, and the third condition parameter further comprises:

[0137] Determining the candidate priority of the AI ​​dialogue robot according to the first condition parameter, the second condition parameter, the third condition parameter, and the fourth condition parameter further includes:

[0138] Furthermore, the third-level sub-grouping values ​​of other AI dialogue robots in the artificial seat corresponding to the AI ​​dialogue robot are counted, and the third-level sub-grouping values ​​of the other AI dialogue robots are calculated to determine the fourth-level association value. The fourth condition parameter indicates the artificial seat state prediction, which has a mapping relationship with the fourth-level association value.

[0139] For example, for AI A2, if it belongs to human agent A, the three-level sub-grouping values ​​of the AI ​​dialogue robot corresponding to human agent A are traversed. When the three-level sub-grouping value is 1, the four-level association value is determined to be 1. If the three-level sub-grouping values ​​of the AI ​​dialogue robot corresponding to human agent A are all 0, then the four-level association value is 0.

[0140] Specifically, in this scenario, the AI ​​dialogue robot candidate tag information includes the AI ​​dialogue robot ID, and 4-bit tag information: for example, 0000, 0001, 0101, 0100, 0111, 1000, 1001, 1100, 1101, 1111, with multiple priority levels, 0000 as the first priority level, 1000, 0001 as the second priority level, 0100 as the third priority level, and 0111, 1100, 1101, 1111 as the fourth priority level.

[0141] Specifically, step 204, generating a distribution instruction to distribute the connected line to the N AI dialogue robots that meet the distribution conditions, includes:

[0142] Distributing connected lines based on the candidate priorities;

[0143] When the number of AI conversation robots of the first priority level is greater than or equal to the lines to be connected, the AI ​​conversation robot of the first priority level is selected;

[0144] When the number of AI conversational robots of the first priority level is less than the lines to be connected, the excess lines will be distributed to AI conversational robots of lower priority levels.

[0145] After step 204, the method further includes:

[0146] Step 205, monitoring the response of the AI ​​dialogue robot to determine whether manual intervention is required; if manual intervention is required, executing step 206;

[0147] Step 206: Push a notification message to the human agent, prompting the human agent that the AI ​​dialogue robot needs to intervene in the dialogue.

[0148] The above steps 205-206 are processed similarly to steps 103-104 in the first embodiment.

[0149] The method may further include step 207, determining whether the first human agent bound to the AI ​​dialogue robot is idle.

[0150] Specifically, determining candidate tag information of the AI ​​dialogue robot, and determining a first human agent bound to the AI ​​dialogue robot;

[0151] Obtain the current status of the first human agent bound to the AI ​​dialogue robot.

[0152] Determine whether the first human agent bound to the AI ​​dialogue robot is idle;

[0153] Step 208: The first human agent replaces the AI ​​dialogue robot to complete the customer service dialogue.

[0154] Specifically, for example, when human agent A performs manual intervention, a new human agent dialogue interface is created, and the history record of the current dialogue of the AI ​​dialogue robot that is manually intervened is pushed to the human agent dialogue interface. The AI ​​dialogue robot resources are released, and the AI ​​dialogue robot state array and the AI ​​dialogue robot resource pool are updated.

[0155] Furthermore, in step 209, it is determined whether there is an idle second manual agent among other manual agents that have no binding relationship with the AI ​​dialogue robot. If there is an idle second manual agent, the idle second manual agent is notified, and the idle second manual agent takes over the AI ​​dialogue robot to complete the customer service dialogue.

[0156] Preferably, the step 206 of pushing a notification message to the human agent to prompt the human agent that the AI ​​dialogue robot needs dialogue intervention may further include:

[0157] Determine the candidate tag information of the AI ​​dialogue robot, such as {AI dialogue robot ID, first-level clustering value, second-level grouping value, third-level sub-grouping value, fourth-level association value},

[0158] If the first-level clustering value and the fourth-level association value corresponding to the AI ​​dialogue robot are both "0", the first human agent corresponding to the AI ​​dialogue robot is determined, and a pop-up screen reminder is initiated to the first human agent.

[0159] When the first-level clustering value corresponding to the AI ​​dialogue robot is 1 or the fourth-level association value is 1, query the AI ​​dialogue robot ID whose first-level clustering value and fourth-level association value are both "0" in the AI ​​dialogue robot candidate tag information, determine the third manual agent corresponding to the AI ​​dialogue robot ID, and initiate a pop-up screen reminder to the third manual agent.

[0160] Furthermore, the first human agent corresponding to the AI ​​dialogue robot can also be directly determined, and a pop-up screen reminder can be initiated to the first human agent. The pop-up screen reminder can be configured with a first control. When the first human agent is unable to intervene, the pop-up screen scheduling can be triggered by operating the first control.

[0161] When the pop-up screen scheduling is triggered, the AI ​​dialogue robot ID whose first-level clustering value and fourth-level association value are both "0" in the AI ​​dialogue robot candidate tag information can also be queried to determine the third manual agent corresponding to the AI ​​dialogue robot ID, and then initiate a pop-up screen reminder to the third manual agent.

[0162] The above-mentioned manual intervention method can also be applied to the first embodiment to achieve similar manual intervention.

[0163] Through the human-machine coupled outbound call scheduling method of Example 2, by predicting the status of the agents, based on the dialing situation of the AI ​​dialogue robot of each agent, the status of the AI ​​dialogue robot, and whether the AI ​​dialogue robot is in the key dialogue link, the connected lines are distributed predictively, thereby improving the working time of human agents and reducing call losses.

[0164] Embodiment 3

[0165] Based on the above embodiments, Figure 6 , Embodiment 3 provides a method for controlling an artificial agent AI workstation, which is characterized in that,

[0166] Step 301, monitoring the working status of the AI ​​dialogue robot; wherein the working status of the AI ​​dialogue robot includes the busy and idle status information of the AI ​​dialogue robot, and the dialogue record of the AI ​​dialogue robot when the AI ​​dialogue robot is in a busy state;

[0167] Step 302: receiving a display instruction, and displaying the label information of the AI ​​dialogue robot, the working status of the AI ​​dialogue robot, and the dialogue record on the display device of the artificial agent AI workstation based on the display instruction;

[0168] Step 303, determining whether a preset pop-up screen triggering condition is met, and if the preset pop-up screen triggering condition is met, triggering pop-up screen scheduling;

[0169] Step 304: Based on the pop-up screen scheduling, a manual agent is selected, and the pop-up screen information is displayed on the display device of the AI ​​workstation of the manual agent.

[0170] See also Figure 7 As shown, the AI ​​workbench of the human agent can monitor the working status of each AI dialogue robot, and display key information such as the working status, dialogue records, and tags of each AI in the form of cards. When the AI ​​needs human agent intervention, the AI ​​workbench notifies the human agent through a pop-up screen, and the human agent can also intervene in advance after discovering important customers through cards.

[0171] Specifically,

[0172] Step 301, monitoring the working status of the AI ​​dialogue robot includes:

[0173] Store the conversation records of the AI ​​conversation robot in real time. The conversation records may directly store voice information and / or store corresponding text information through text conversion;

[0174] Indexing the conversation record of the AI ​​conversation robot based on the AI ​​conversation robot ID;

[0175] The conversation information of the AI ​​conversation robot includes recorded voice information and text information converted in real time based on the voice confidence, and both the voice information and the text information are recorded with time marks.

[0176] Get the AI ​​conversational robot status array and / or AI conversational robot candidate tag information,

[0177] The AI ​​dialogue robot state array and / or AI dialogue robot candidate tag information can be determined according to the method in the previous embodiment.

[0178] Determine the busy / idle status information of the AI ​​dialogue robot based on the AI ​​dialogue robot status array and / or the AI ​​dialogue robot candidate tag information,

[0179] When the AI ​​dialogue robot is in a busy state, the dialogue record of the AI ​​dialogue robot is searched through the AI ​​dialogue robot ID.

[0180] Specifically, the step 302, receiving a display instruction, and based on the display instruction, displaying the label information of the AI ​​dialogue robot, the working status of the AI ​​dialogue robot, and the dialogue record on the display device of the artificial agent AI workstation specifically includes:

[0181] When the AI ​​dialogue robot is in a busy state, a dialogue display instruction is triggered;

[0182] Based on the display indication, the dialogue interface corresponding to the AI ​​dialogue robot is displayed in the form of a card, and the dialogue interface corresponding to the AI ​​dialogue robot displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue records of the AI ​​dialogue robot.

[0183] Specifically, based on the label information of the AI ​​dialogue robot, a plurality of the AI ​​dialogue robots in busy states are classified;

[0184] Extract key information of the conversation records of the AI ​​conversation robot in busy state, construct a conversation information graph based on the key information, and optimize and train the AI ​​conversation robot based on the conversation information graph of the conversation records of the AI ​​conversation robot with the same label information, so as to enrich the intelligent processing capability of the AI ​​conversation robot under the label information.

[0185] Furthermore, according to the conversation record displayed by the display device of the artificial agent AI workstation, the artificial agent can actively intervene in the conversation;

[0186] For example, when a human agent customer service is viewing a conversation record displayed on a display device of the human agent AI workstation and finds that the conversation needs to be supplemented, a temporary insertion can be triggered.

[0187] Specifically, a second control is set in the dialogue interface corresponding to the AI ​​dialogue robot, and the second control is used to trigger the insertion of a third-party dialogue;

[0188] Triggering the insertion of a temporary dialogue by operating the second control;

[0189] Specifically, the first temporary dialogue is input through the input control of the dialogue interface, and the first temporary dialogue can be a voice message or a text message.

[0190] Furthermore, the voice information and the text information are voice-converted to generate a first temporary voice response, and the converted first temporary voice response is transmitted to the calling party.

[0191] Preferably, in step 303, judging whether a preset pop-up screen triggering condition is met, if the preset pop-up screen triggering condition is met, triggering pop-up screen scheduling further includes:

[0192] Based on multiple rounds of conversations with the AI ​​conversational robot, corresponding keyword data and / or user status information is extracted to analyze the need for intervention in the current conversation.

[0193] The preset pop-up screen triggering condition is that the customer has an intervention demand.

[0194] The actual intention of the user is determined by extracting the corresponding keyword data in the conversation, or the current status of the user is determined by analyzing the tone and voice in the user's voice information. For example, if the user has the intention to further communicate and subscribe, but is in an irritable state in the machine conversation, it is determined that the current AI conversation robot is in a critical conversation link, that is, the customer has a need to intervene. At this time, the pop-up screen scheduling is triggered.

[0195] Specifically, the step 304 further includes:

[0196] Step 305: Receive a first input for the pop-up screen information, where the first input is first operation information based on the pop-up screen information.

[0197] The pop-up screen reminder may be configured to include a first control, and when the first manual agent is unable to intervene, the pop-up screen scheduling may be triggered by operating the first control.

[0198] For example, when human agent A is in a conversation, and multiple AI conversation robots corresponding to human agent A need human intervention, the pop-up screen information is displayed on the display device of the AI ​​workstation of human agent A, and the pop-up screen information may include a first control, such as a transfer pop-up screen. At this time, human agent A can transfer the pop-up screen information to other idle human agents by clicking the first control.

[0199] Based on the above embodiment, through the artificial agent AI workbench control method, the working status of each AI robot can be monitored intuitively, and through third-party dialogue intervention and pop-up screen scheduling, the customer reach speed per unit time is improved, call losses are reduced, and call efficiency is improved.

[0200] Embodiment 4

[0201] See also Figure 8 , corresponding to the method in the previous embodiment 1, this embodiment provides a human-machine coupling calling device, characterized in that,

[0202] The first judgment module judges whether multiple calls initiated at the same time are connected.

[0203] A distribution control module distributes the connected one or more calls to an artificial intelligence AI dialogue robot, which performs an initial response. The AI ​​dialogue robot has a binding relationship with a human agent;

[0204] A dialogue monitoring module monitors the response of the AI ​​dialogue robot and determines whether manual intervention is required; if manual intervention is required, a first trigger message is sent to a message push module;

[0205] The message push module pushes a notification message to the human agent, prompting the human agent that the AI ​​dialogue robot needs to intervene in the dialogue.

[0206] Preferably, the device further comprises

[0207] Query module, query the AI ​​dialogue robot resource pool;

[0208] The AI ​​dialogue robot selection module selects one or more AI dialogue robots from the AI ​​dialogue robot resource pool based on predetermined rules.

[0209] The distribution control module distributes the connected one or more calls to the selected AI dialogue robot.

[0210] Preferably, the device further comprises

[0211] The second judgment module monitors the conversation information of the AI ​​conversation robot through the human agent bound to the AI ​​conversation robot, and chooses whether to intervene;

[0212] The monitoring method is to monitor the real-time conversation text or voice information between AI and customers, and the customer's status can be obtained through the voice information.

[0213] Preferably, the message push module is used to

[0214] Push notification messages to human agents who have a binding relationship with the AI ​​conversational robot;

[0215] Among them, the human agent is notified through a pop-up screen; the pop-up screen is controlled to be displayed on a display device corresponding to the human agent, wherein the display device synchronously displays one or more dialogue interfaces corresponding to the AI ​​dialogue robot corresponding to the human agent, and the pop-up screen is displayed at the top layer of the display area.

[0216] Embodiment 5

[0217] See also Fig. 9 , corresponding to the method in the previous embodiment 2, this embodiment provides a scheduling device for human-machine coupling outbound calls, characterized in that the device includes

[0218] A parameter acquisition module determines the condition parameters of the AI ​​dialogue robot, wherein the condition parameters of the AI ​​dialogue robot include a first condition parameter, a second condition parameter, and a third condition parameter, wherein the first condition parameter indicates the state of the artificial seat corresponding to the AI ​​dialogue robot, the second condition parameter indicates the state of the AI ​​dialogue robot, and the third condition parameter indicates whether the AI ​​dialogue robot is in a critical dialogue link;

[0219] A priority determination module, which determines the candidate priority of the AI ​​dialogue robot according to the first condition parameter, the second condition parameter, and the third condition parameter;

[0220] A selection control module selects N AI conversation robots that meet the distribution conditions according to the candidate priorities of the AI ​​conversation robots;

[0221] The distribution control module generates distribution instructions and distributes the connected lines to the N AI dialogue robots that meet the distribution conditions.

[0222] The parameter acquisition module specifically includes:

[0223] The condition judgment module judges whether the artificial seat meets the first preset condition, and if there is an artificial seat that meets the first preset condition, sends a notification message to the classification control module,

[0224] The classification control module is used to form a first candidate cluster with the AI ​​dialogue robots corresponding to the human seats that meet the first preset condition; and form a second candidate cluster with the AI ​​dialogue robots corresponding to the human seats that do not meet the first preset condition;

[0225] The condition judgment module is further used to judge whether the AI ​​dialogue robots in the first candidate cluster and the second candidate cluster meet the second preset condition;

[0226] The classification control module is used to, according to whether the AI ​​dialogue robot meets a second preset condition, perform secondary clustering on the first candidate cluster and the second candidate cluster, and perform a first grouping mark on the AI ​​dialogue robot;

[0227] The condition judgment module is further used to determine whether the AI ​​dialogue robot that does not meet the second preset condition meets the third preset condition; based on the third preset condition, the AI ​​dialogue robot is marked as a first sub-group.

[0228] The priority determination module is further used to:

[0229] Obtain candidate tag information of the AI ​​dialogue robot, and determine the priority of the candidate AI dialogue robot based on the first-level clustering value, the second-level grouping value, and the third-level sub-grouping value in the candidate tag information of the AI ​​dialogue robot.

[0230] The distribution control module is further used to:

[0231] Distributing connected lines based on the candidate priorities;

[0232] When the number of AI conversation robots of the first priority level is greater than or equal to the lines to be connected, the AI ​​conversation robot of the first priority level is selected;

[0233] When the number of AI conversational robots of the first priority level is less than the lines to be connected, the excess lines will be distributed to AI conversational robots of lower priority levels.

[0234] Embodiment 6

[0235] See also Fig.10 , corresponding to the method in the previous embodiment 3, this embodiment provides an artificial agent AI workbench control device, characterized in that,

[0236] A real-time conversation monitoring module monitors the working status of the AI ​​conversation robot; wherein the working status of the AI ​​conversation robot includes the busy and idle status information of the AI ​​conversation robot, and the conversation record of the AI ​​conversation robot when the AI ​​conversation robot is in a busy state;

[0237] A display control module receives a display instruction and displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue record of the AI ​​dialogue robot on a display device of the artificial agent AI workstation based on the display instruction;

[0238] A pop-up screen trigger judgment module is used to judge whether a preset pop-up screen trigger condition is met, and if the preset pop-up screen trigger condition is met, the pop-up screen scheduling is triggered;

[0239] The pop-up screen scheduling module selects a manual agent based on the pop-up screen scheduling, and displays the pop-up screen information on the display device of the AI ​​workbench of the manual agent.

[0240] The real-time conversation monitoring module further includes a conversation storage module, wherein the conversation storage module is used to

[0241] Store the conversation records of the AI ​​conversation robot in real time. The conversation records may directly store voice information and / or store corresponding text information through text conversion;

[0242] Wherein, the conversation record of the AI ​​conversation robot is indexed based on the AI ​​conversation robot ID;

[0243] The conversation information of the AI ​​conversation robot includes recorded voice information and text information converted in real time based on the voice confidence, and both the voice information and the text information are recorded with time marks.

[0244] The display control module is further used to

[0245] When the AI ​​dialogue robot is in a busy state, a dialogue display instruction is triggered;

[0246] Based on the display indication, the dialogue interface corresponding to the AI ​​dialogue robot is displayed in the form of a card, and the dialogue interface corresponding to the AI ​​dialogue robot displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue records of the AI ​​dialogue robot.

[0247] The device also includes

[0248] An input module receives a first input for the pop-up screen information, where the first input is first operation information based on the pop-up screen information;

[0249] The pop-up screen reminder may be configured to include a first control, and when the first manual agent is unable to intervene, the pop-up screen scheduling may be triggered by operating the first control.

[0250] The present invention also provides a terminal device, characterized in that the terminal device includes a processor and a memory, the memory stores a computer program that can be run on the processor, and the computer program implements the method as described above when executed by the processor.

[0251] The present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program that can be run on a processor, and the computer program implements the method as described above when executed.

[0252] Any combination of one or more computer-readable media can be used. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above. Computer-readable storage media can include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), a flash memory, an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by an instruction execution system, device or device or used in combination with it.

[0253] Computer program code for carrying out operations of the present invention may be written in one or more programming languages, or a combination of these.

[0254] The above description is only an example for facilitating understanding of the present invention and is not intended to limit the scope of the present invention. In specific implementation, those skilled in the art may change, increase, or reduce the components of the device according to actual conditions, and may change, increase, reduce, or change the order of the steps of the method according to actual conditions without affecting the functions implemented by the method.

[0255] Although the embodiments of the present invention have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and purpose of the present invention. The scope of the present invention is defined by the claims and their equivalents. Improvements made without creative work should be included in the scope of protection of the present invention.

Claims

1. A method for controlling an artificial agent AI workstation, characterized in that: Step 301, monitoring the working status of the AI ​​dialogue robot; wherein the working status of the AI ​​dialogue robot includes the busy and idle status information of the AI ​​dialogue robot, and the dialogue record of the AI ​​dialogue robot when the AI ​​dialogue robot is in a busy state; Step 302: receiving a display instruction, and displaying the label information of the AI ​​dialogue robot, the working status of the AI ​​dialogue robot, and the dialogue record on the display device of the artificial agent AI workstation based on the display instruction; Step 303, determining whether a preset pop-up screen triggering condition is met, and if the preset pop-up screen triggering condition is met, triggering pop-up screen scheduling; Step 304: Based on the pop-up screen scheduling, select a manual seat, and display the pop-up screen information on the display device of the AI ​​workstation of the manual seat; The step 302, receiving a display instruction, and based on the display instruction, displaying the label information of the AI ​​dialogue robot, the working status of the AI ​​dialogue robot, and the dialogue record on the display device of the artificial agent AI workstation specifically includes: When the AI ​​dialogue robot is in a busy state, a dialogue display instruction is triggered; Based on the display instruction, the dialogue interface corresponding to the AI ​​dialogue robot is displayed in the form of a card, and the dialogue interface corresponding to the AI ​​dialogue robot displays label information including the AI ​​dialogue robot, as well as the working status and dialogue record of the AI ​​dialogue robot; The step 303 determines whether a preset pop-up screen triggering condition is met. If the preset pop-up screen triggering condition is met, triggering the pop-up screen scheduling includes: Based on multiple rounds of conversations with AI chatbots, extract corresponding keyword data and / or user status information to analyze the need for intervention in the current conversation; The preset pop-up screen triggering condition is that the customer has an intervention demand; Extracting corresponding keyword data in the conversation to determine the actual intention of the user, or determining the current state information of the user by analyzing the tone and / or voice in the user's voice information to determine the access requirements of the current conversation; When it is determined that the customer is in an irritable state during the machine conversation, it is determined that the current AI conversation robot is in a critical conversation phase and the customer has a need to intervene, triggering the pop-up screen scheduling; After step 304, the following steps are also included Step 305, receiving a first input for the pop-up screen information, where the first input is first operation information based on the pop-up screen information; The pop-up screen reminder is configured with a first control. When the first manual agent is unable to intervene, the pop-up screen scheduling can be triggered by operating the first control to transfer the pop-up screen information to other idle manual agents.

2. The method according to claim 1, characterized in that: Step 301, monitoring the working status of the AI ​​dialogue robot includes: The conversation records of the AI ​​conversation robot's conversation process are stored in real time, wherein the conversation records directly store voice information and / or store corresponding text information through text conversion; Indexing the conversation record of the AI ​​conversation robot based on the AI ​​conversation robot ID; The conversation information of the AI ​​conversation robot includes recorded voice information and text information converted in real time based on the voice information, and both the voice information and the text information are recorded with time marks.

3. A control device for an artificial agent AI workbench, characterized in that: The device comprises A real-time conversation monitoring module monitors the working status of the AI ​​conversation robot; wherein the working status of the AI ​​conversation robot includes the busy and idle status information of the AI ​​conversation robot, and the conversation record of the AI ​​conversation robot when the AI ​​conversation robot is in a busy state; A display control module receives a display instruction and displays the label information of the AI ​​dialogue robot, as well as the working status and dialogue record of the AI ​​dialogue robot on a display device of the artificial agent AI workstation based on the display instruction; A pop-up screen trigger judgment module is used to judge whether a preset pop-up screen trigger condition is met, and if the preset pop-up screen trigger condition is met, the pop-up screen scheduling is triggered; A pop-up screen scheduling module selects a manual agent based on the pop-up screen scheduling, and displays the pop-up screen information on a display device of the AI ​​workstation of the manual agent; The display control module is also used When the AI ​​dialogue robot is in a busy state, a dialogue display instruction is triggered; Based on the display instruction, the dialogue interface corresponding to the AI ​​dialogue robot is displayed in the form of a card, and the dialogue interface corresponding to the AI ​​dialogue robot displays label information including the AI ​​dialogue robot, as well as the working status and dialogue record of the AI ​​dialogue robot; The pop-up screen trigger judgment module judges whether a preset pop-up screen trigger condition is met. If the preset pop-up screen trigger condition is met, triggering the pop-up screen scheduling includes: Based on multiple rounds of conversations with AI chatbots, extract corresponding keyword data and / or user status information to analyze the need for intervention in the current conversation; The preset pop-up screen triggering condition is that the customer has an intervention demand; Extracting corresponding keyword data in the conversation to determine the actual intention of the user, or determining the current state information of the user by analyzing the tone and / or voice in the user's voice information to determine the access requirements of the current conversation; When it is determined that the customer is in an irritable state during the machine conversation, it is determined that the current AI conversation robot is in a critical conversation phase and the customer has a need to intervene, triggering the pop-up screen scheduling; The device also includes An input module receives a first input for the pop-up screen information, where the first input is first operation information based on the pop-up screen information; The pop-up screen reminder is configured with a first control. When the first manual agent is unable to intervene, the pop-up screen scheduling can be triggered by operating the first control to transfer the pop-up screen information to other idle manual agents.

4. The device according to claim 3, characterized in that The real-time conversation monitoring module also includes a conversation storage module, which is used to The conversation record of the conversation process of the AI ​​conversation robot is stored in real time, wherein the conversation record directly stores the voice information and / or stores the corresponding text information through text conversion; Wherein, the conversation record of the AI ​​conversation robot is indexed based on the AI ​​conversation robot ID; The conversation information of the AI ​​conversation robot includes recorded voice information and text information converted in real time based on the voice information, and both the voice information and the text information are recorded with time marks.

5. A terminal device, characterized in that: The terminal device comprises a processor and a memory, wherein the memory stores a computer program executable on the processor, and when the computer program is executed by the processor, the method according to any one of claims 1 to 2 is implemented.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program that can be run on a processor, and the computer program implements the method according to any one of claims 1 to 2 when executed.

Images