Automated chatbot transfer to live agent
Disclosed are methods, systems, and machine-readable mediums which provide for customer chatbots that detect a customer handoff condition and in response, transferring the customer to a communication session with a live agent. The handoff condition may comprise an inability to understand the customer, an inability to answer the customer's question, expressions of frustration or anger on the part of the customer, a customer's express request to be transferred, or the like. The live agent may receive a complete history of the conversation with the chatbot so that the customer does not have to repeat him or herself to the live agent. The chatbot chat session may be linked to a social networking account of the customer and may take place in association with a social networking profile page of the company.
1. A method for routing a digital communication, the method comprising:
receiving, from a network-based service on a chat page of the network-based service, a first chat message entered into a chat of the network-based service, the first chat message posted by a user as part of a chat conversation, the first chat message received, by a chat router, through a callback web address of the chat router provided to the network-based service;
sending, by the chat router, the first chat message to a live agent service and to an automated chat service;
receiving at the chat router a first reply from the live agent service and a second reply from the automated chat service;
scoring the first reply from the live agent service and the second reply from the automated chat service through a first machine-learning model implemented by the chat router;
ranking the first reply and the second reply based on the scoring; and
posting the second reply from the automated chat service to the chat page based on the ranking indicating that the second reply has a higher score than the first reply, wherein the second reply is posted to the chat page without action by the live agent service.
2. The method of claim 1 , wherein the automated chat service includes a second machine-learning model, and wherein the chat router uses the first chat message and the first reply from the live agent service to refine the second machine-learning model of the automated chat service.
3. The method of claim 1 , wherein posting the first reply or the second reply includes sending the first reply or the second reply to the network-based service using a Hypertext Transfer Protocol (HTTP).
4. The method of claim 1 , further comprising providing a chat history of the user to the live agent service.
5. The method of claim 1 , further comprising:
receiving an indication from the live agent service that future messages of the chat conversation are to be routed to the automated chat service;
responsive to receiving the indication from the live agent service that future messages of the chat conversation should be routed to the automated chat service:
receiving a second chat message;
transmitting the second chat message to the automated chat service, the automated chat service using natural language processing to produce a third reply without human intervention; and
posting the third reply to the chat page.
6. The method of claim 5 , further comprising updating the natural language processing based on the second reply.
7. The method of claim 5 , wherein the natural language processing includes a neural network.
8. At least one non-transitory machine-readable medium including instructions for routing a digital communication, which when executed by processing circuitry cause the processing circuitry to perform operations to:
receive, from a network-based service on a chat page of the network-based service, a chat message entered into the chat page of the network-based service, the chat message posted by a user as part of a chat conversation, the chat message received, by a chat router, through a callback web address of the chat router provided to the network-based service;
send, by the chat router, the chat message to a live agent service and to an automated chat service;
receive at the chat router a first reply from the live agent service and a second reply from the automated chat service;
score the first reply from the live agent service and the second reply from the automated chat service through a first machine-learning model implemented by the chat router;
rank the first reply and the second reply based on the scoring; and
post the second reply from the automated chat service to the chat page based on the ranking indicating that the second reply has a higher score than the first reply, wherein the second reply is posted to the chat page without action by the live agent service.
9. The at least one machine-readable medium of claim 8 , wherein the automated chat service includes a second machine-learning model, and wherein the chat router uses the chat message and the first reply from the live agent service to refine the second machine-learning model of the automated chat service.
10. The at least one machine-readable medium of claim 8 , wherein to post the first reply or the second reply includes sending the first reply or the second reply to the network-based service using a Hypertext Transfer Protocol (HTTP).
11. The at least one machine-readable medium of claim 8 , further comprising providing a chat history of the user to the live agent service.