TL;DR Artificial Intelligence and Machine Learning can’t replace human support agents yet, but can dramatically decrease boring and repetitive tasks human agents are spending their time on.
What is Artificial Intelligence?
Artificial intelligence (AI) is the science of creating machines that can work and react in a way that normally requires human intelligence
Human + AI cooperation in customer support
Customer support is about personal interactions and empathy - that cannot yet be expressed by machines (at least, not today or in the near future).
Robots cannot yet replace human support agents, but they can work together to automate boring and repetitive tasks.
Below, there are a few examples of such tasks - and how they can be automated using AI.
Prioritizing incoming support requests
For a human, prioritizing a request would be an easy task.
Quickly reading a few sentences and understanding who the customer is will be enough to determine the urgency of the request.
However, if the request is urgent, we would want to take action right away. The time that passes until a human reviews the request and determines that it is urgent - might be critical.
That is where machines can help — determine the message is urgent in seconds and even send an alert to the support team.
How can a machine determine if a support request is urgent?
Natural Language Processing(NLP) technology is capable of “understanding” free text (such as support email request) and retrieving useful information, such as tone, feeling expressed by the customer, sense of urgency, and more.
By combining NLP technology and connecting to data sources (such as CRM / Customer support software) machines can determine the urgency of support requests within seconds. There are several ways to do so.
Below, I will provide the parameters we use at Kilometer.io to perform that task.
Assigning incoming support requests to the right person or team
Have you ever sent a support question and got a reply the next day just to notify you it was assigned to the right person or team?
Not only is it boring to assign tickets to teams, but why make the customer wait?
By the time the issue is forwarded to the right destination, it could have been already resolved.
How can a machine determine who should handle a support request?
Machine Learning provides computers with the ability to perform tasks by learning from past examples, without being explicitly programmed to do those specific tasks.
Now, let’s not forget your customer support software is a great source for examples of various support requests and information about who handled them.
Those examples can be fed into Machine Learning algorithms to train the machines to determine who should be assigned to a support request in real time.
Creating, searching, updating, and managing canned responses or templates.
While resolving support issues and interacting with customers can be fun, what is not fun is searching and maintaining canned responses or templates or even looking for similar questions addressed in the past.
Luckily, machines can do such boring tasks with ease and joy.
Natural Language Processing technology enable machines to understand the customer’s intent, what he wants.
For example this technology can “understand” that those questions have the same intent:
- What is the pricing of the product?
- Should I rob a bank to use the product?
This technology also enables the chat-bots revolution the tech world is undergoing those days.
Machine Learning technology can learn how messages with similar intents were addressed in the past, and suggest an appropriate reply.
Such a system can automatically improve itself and adjust in real-time as new data become available (new answers to new questions).
Past support requests should not be the only source of information machines can use to suggest replies. They can also use your knowledge base, FAQ pages etc.
No more searching, no more editing, organizing and maintaining canned responses or templates.
Free businesses of doing boring stuff
Our vision at Kilometer.io is to free businesses from doing boring and repetitive tasks.
To achieve this, we use advanced technologies such as Machine Learning and Artificial Intelligence.
We feel customer support is definitely an area where we can provide a lot of value.
We also develop other product using the same concept:
Data Enrichment, Lead Scoring and Churn Prediction.
All our products seamlessly integrate with tools companies are already using such as: CRMs, Marketing Automation platforms, and Customer Support software.
