August 14, 2020
Using AI Models To Decrease Time-Sucking Customer Support Team Tasks
The time for AI is here. And it doesn’t have to be intimidating or mysterious.
At Forethought, we strongly believe that customer support AI is going to improve the way that support teams predict and anticipate customer needs.
We’re sure you know how much of a resource-draining task it can be trying to predict a ticket’s need as it comes through. Having to route tickets to the right agents takes up valuable human capital and hurts the customer experience. No one wants to have to wait in a queue only to get rerouted again a few days later.
So, what can customer support AI do and how does it help your customers? To answer this question, we have to talk about training models.
You're probably wondering, "What Is An AI Model?"
A model is a black box that you feed inputs into. As Deon Nicholas, our CEO, explained so eloquently, “It gobbles up the information.” When the input comes out of the black box, that piece of information now has a label attached to it.
In the case of customer support tickets, this model would label tickets based on the information provided within the ticket.
Forethought creates different AI models that identify the need of a support ticket. Whatever needs the ticket expresses helps categorize and route them to the right people using labels related to priority, language, skill, department, or resolution.
Each one of these labels is a new model. Let’s get into how you go about using a model to enhance customer support AI.
How Do You Train an AI Model?
Training a model is somewhat like teaching a child about animals.
Let’s say you want to introduce a child to the world of fish. You could go to the pet store and buy a goldfish, and your child would be so happy! But that goldfish is just one example of a fish.
If you want your child to be able to recognize lots of different types of fish, you’ll have to teach them about more types of fish. So you head to the aquarium where you teach the child about all the various sizes, colors, and types of fish. Before you know it, they’re pointing out fish you didn’t even know existed.
And just like that, you’ve taught (trained) a child (a model).
In a sense, we each have models in our brains for identifying fish. We’ve seen enough examples over and over, and now we’re able to identify them easily. Our brains know what shape a fish can take, what characteristics make it a fish, and can identify a fish compared to something that isn’t one.
Similarly, you train an AI model by teaching it lots of previous human experiences to then identify what something new is, without the help of a human! For customer support, we do this with tickets, using historical knowledge to better predict the future.
How Can Organization Leverage AI Models to Improve Customer Support?
For customer support, we use AI models to help us predict a ticket’s need and solve customer issues faster. By creating models based on past issues, tickets, and categorization, you can leverage your past customer experiences to make the future ones even better.
What Would Using AI Models for Customer Support Look Like?
- Catch escalations before they create chaos and hurt your business using a model that prioritizes tickets based on past escalations and ticket priority.
- Funnel out spam and less important tickets from your queue before an agent ever has to look at them using a model that labels tickets as “spam” or “not spam.”
- Route tickets to the right place for resolution the first time without having to bog down talented agents. Do this with a model that learns all the categories or departments where similar, historical tickets have been resolved.
Because we use actual historical knowledge to predict the labels for new issues, our customer support AI can achieve high response accuracy. Additionally, because we work off of historical knowledge, our models are perfectly tailored to a customer support team’s business needs.
At Forethought, we built Triage to make it insanely easy to build AI models on top of your existing tickets.