Impact of Generative AI on the Contact Center Industry
Tod Famous, former SVP Head of Product @Genesys
Tod Famous has over 25 years of experience building contact center software. Most recently, he was the SVP of Product @ Genesys. Where he was responsible for building out their Multicloud and PureEngage products. Prior to that, he held several leadership positions at Cisco. He is an expert in the customer care market, enterprise software sales, channel development, and agile software engineering process.
In today’s interview, we cover:
How AI is currently being used in contact center products
What is the potential of generative AI in the contact center industry
How it will impact customer service and businesses
Some of the challenges with adoption and implementation
How the role of an agent is going to change, and
Why incumbents are better prepared to take advantage of this new technology
What follows below is a condensed and lightly edited version of our interview.
Tod thanks for being a guest. Let's start with the basics. How is AI currently being leveraged for Contact Center products?
Tod Famous: Mustafa thanks for having me.
There are quite a few ways where contact center software is using AI. Let me share with you the three most popular use cases.
The first is in classification. For example, where you call a contact center and the system asks you why you are calling. Understanding what you are saying and guiding you to the right destination is all done with AI technology. And here there are different levels of sophistication - as to how open-ended you want to make it. But in general, AI is very good at translating your voice and classifying it against a set of preconfigured options.
The second is fraud detection. So if somebody is doing something unexpected, AI is very good at picking it up. It is much better at detecting or pattern recognition than humans.
The third is routing. For example, using expert systems, and machine learning systems, to make decisions as to who to assign your contact to. Who's the best person to speak with you about that particular topic.
All three use cases are heavily used in contact center systems. And have been around for quite some time.
Given that context, where do you think Generative AI fits in? What is your perspective on how it will change the contact center industry?
TF: Great question.
I think a good place to start with this is that to recognize this is very new. I mean for many of us, our first exposure to ChatGPT was only earlier this year. So in my mind what is happening is pretty experimental on the enterprise side.
Don't get me wrong, generative AI is definitely in its wow moment. But in my mind, this is just the beginning.
Agreed. You make a good point. This is still very much in its infancy.
TF: Exactly. Which is why it is important to separate the hype from reality.
For example, there is a lot of talk about generative AI getting rid of call centers. After all, if you feed the model all the responses. Then do you even need agents? Customers can get all the answers they want via a generative AI chatbot.
The problem with this hypothesis is that it is not rooted in reality. We all know that ChatGPT hallucinates. Do you think a premium brand like Apple or your bank would risk an AI answering customer service inquiries? What if the model provides the wrong information? What if it hallucinates? Enterprises can’t take on that risk. If a customer asks them a question, the response has to be accurate all the time.
So the theory that generative AI will eliminate contact centers is overblown.
Absolutely makes sense. So then where do you think it will make a difference?
TF: For me, at least in the near term, I see this technology delivering a lot of benefits to the agents and the business.
For example, after every call, an agent usually spends about 2 mins. taking notes. These notes are used for a lot of things, like improving quality, call center performance, or even kicking off a selling opportunity. Now instead of the agent taking note, what if generative AI did it? Yes, it saves only 2 minutes, but then you multiply it by 1000 agents over millions of calls and suddenly you are taking real savings. The agent benefits because they don’t enjoy taking notes. And the business benefits because they can use that 2 mins per agent per call to answer other calls.
Second, is that the agent can use generative AI to better answer the customers' questions. For example, you can put a lot of information about your customer into the prompt. So if a customer asks about Roth vs Traditional IRA. Based on their information, age, demographics, etc. generative AI could create a customized answer and the agent can then use that as a starting point to deliver a better and more personal experience.
Third is what I briefly mentioned above. If generative AI is really good at picking out signals from customers' calls. Signals like I am moving, or buying a house, or in my case adding my daughter to our insurance plan. These are very rich signals that companies can then use to upsell products. Now it has to be done delicately, but the opportunity is there.
The fourth is in translation. Right now AI has a hard time recognizing names, last names, addresses, emails, etc. So we always need an agent to help capture that information. But if generative AI can do the job better, then that eliminates another step in the IVR (Interactive Voice Response) system..
I love the examples. Do you think any of this will be directly available to the customer? Or do you still think it is the agent?
TF: I think it is going to be an agent first.
Giving direct access to the customer sounds tempting, but we still have to deal with hallucinations. And that as we talked about is too risky.
And I know what some people say, that we can just train it on our internal data. But that kind of defeats the purpose. You lose the benefit of the large language model.
For me, having the agent in the middle using the technology to assist the customer is where the real value is today.
And if these technologies do come to fruition, where is the value? How does this translate into benefits?
TF: The value is the key here. So, in my opinion, there are two main groups that benefit.
First is the customer. So when, generative AI does some of the work, like name recognition or better classification. The customers get to the agent faster and can get their issues resolved faster. And that translates to a better customer experience.
Second, is the business. And for them, AI tools help them reduce agent time, improve productivity, and reduce costs. Remember, the agent is the most expensive entity in the entire process. Having an AI do some of the work means that agents spend more time helping fix customers' problems as opposed to doing busy work. That busy work is important, don't get me wrong. But why have agents do it when you have a bot?
Now does this mean there are going to be fewer agents?
TF: I may be an outlier on this point but I don’t see this technology reducing the global customer service labor pool. AI increases productivity and allows the same work to be done with fewer agents but what we’ve seen in the past is that businesses continue to expand and create new service offers; they use the productivity gains for growth. The nature of the work that agents perform keeps changing, for example, they won't be asking you what your name is, they won't be asking you to explain your problem. They'll be having that empathetic customer service conversation and quickly addressing your concerns.
Let’s move on to adoption. What do you see are the biggest challenges of adopting this new technology?
TF: There really are three that come to mind.
First, is what we talked about earlier. There is a hallucination problem that we have to fix. And it will get better, but until then I don’t expect enterprises will be ready to go all in.
Second, is security and privacy. We share all this data into LLM but what happens to that data? What if we end up sharing PII information? Can it be used or tracked? So generative AI really needs to come with guardrails around data privacy and security.
And to add to that, there is the regulatory environment. Where the laws per country are different. Generative AI will have to figure out how to work within these laws.
Third, it is just change management. For LLM applications to work, we may have to re-engineer our internal business processes and IT architecture. That work takes a lot of time in the enterprise market.
Agreed. There are a lot of people working on the first two. But the last one I think is going to be the hardest.
How do you see the landscape changing? Do you see new start-ups taking over the incumbents, or do you see larger incumbents being able to quickly incorporate this technology?
TF: I think the latter, frankly, because some of the use cases we talked about aren't that difficult to implement from a technology point of view. The complexity is more in the go-to-market model and the business process re-engineering.
There is, of course, a potential for disruption. A vendor comes in and completely rethinks this space. But I think that is hard. And at least from what I am seeing in this space, highly unlikely.
Tod, this has been awesome. One last question. What is next for you?
TF: So I recently left Genesys and as I said while we were prepping for this interview, the first thing I plan to do is some fishing (laughing). And of course, spend time with the family. My daughter is about to go to college, so I want to spend as much time as possible with my family during her last summer home.
But after the summer, I want to dive back in. I absolutely love this industry and this space. What I want to really focus on and am very passionate about is improving the customer experience.
A lot of what we talk about really benefits the business. Yes, customers too to some extent. But it is primarily about saving those hard dollars. I want to build a contact center experience that does that and delivers a high-quality customer experience.
That sounds great. I wish you the best of luck. Thank you for being such a great guest.
TF: Mustafa I am glad you reached out to me and connected. This was a lot of fun. Thanks for having me on.