Thrive, an outsourcer of managed services, cloud infrastructure and security based in Foxborough, Mass., has standardized on ServiceNow and has begun rolling out AI agents to automate and broaden customer self-service and support. The goal is to reduce Thrive’s contact center call volume and upskill service employees to solve complex customer issues that AI agents can’t handle.
We discussed with COO Scott Steele how the global managed services provider (MSP) and managed security service provider (MSSP) has deployed AI service agents. He shared how the agents are transforming the company’s contact center, and what lies ahead as AI customer service automation evolves.
Editor’s note: This Q&A was edited for clarity and brevity.
Briefly describe Thrive’s customers.
Scott Steele
Scott Steele: Primarily, our customers are in the SMB space. “Sophisticated SMB” is the way that we would describe it. Technology-forward people are the sweet spot for us — lower midmarket, from a customer environment perspective.
How are you deploying AI to augment contact center operations?
Steele: End-user support is where we generally see most of the contact center volume, and that’s where we’ve deployed some AI. We’ve produced operational efficiencies through the automation process. We’re on ServiceNow as our primary workflow engine. It’s allowed us to maintain the data across [our cloud, MSP and MSSP] platforms. Being on one platform ensures the data is accurate, that you’re not working three different data structures, and that the architecture is aligned across the business.
We’re not bleeding edge because you don’t necessarily want to be over your skis when it comes to AI. The biggest thing that we’ve seen is that we need to get the policy, governance and management right. You have to have that in your AI strategy to make sure that you can implement it the right way and get the outcomes.
What are Thrive’s main uses for AI in the contact center?
Steele: Process management is the most significant piece you need to do, and process engineering around your environment is needed to understand where those pieces are. Where do you have bottlenecks? How does it actually operate?
So we looked across the ecosystem. We found areas that were what we would call “too many hops.” We might have an agent pick up the phone when a user needed a password reset, and that call might end up going to a different engineer. We were world-class at picking up the phone — 20 seconds — but not world-class with handle time. We were at about six minutes because it took so long to get through the process and ensure we were getting into the right area of the organization.
We needed to compress that time. We were not necessarily deploying AI to save money, but to improve our customer experience. So we deployed automated work assignment through ServiceNow Task Intelligence. We’ve integrated that with our own GenAI capabilities, such as summarization, and some other agents. Now we’ve moved the engineers closer to the customer, and they can resolve problems 36% to 38% [without forwarding a call]. So we’ve reduced the time we’re on the phone with the customers, and we’re getting it to the right location.
So engineers now answer the phone. What happened to the agents who didn’t want to be upskilled to engineers?
Steele: Several really wanted to remain call center agents [and not become engineers]; we helped move them to other call centers in the area.
AI is helping Thrive route calls better. From your perch in the industry, when do you see autonomous AI agents taking on more decision-making responsibility?
Steele: Agentic AI still has to be managed. It’s one of the reasons why ServiceNow put its Control Tower in place, because it’s a great way to get control of your agentic AI, ensuring that your agents are all doing what they’re supposed to be doing. You still have to manage, just like you have to manage people, and you have to make sure that the quality coming out of the agents is what you expect.
What will all this look like in five years? Do you think we’ll still have that human in the loop, and that agentic AI will be gathering information for the humans?
Steele: The AI agent was not a thing 18 months ago, and now it’s everything everybody’s talking about. When we look at where we’re headed in five years, you’ll see the menial tasks that are very much repeatable, rote, will go away. Then you’ll start to see AI moving into some of the decision-making or, like we’re seeing, the information gathering today.
Because we’re a growing business, we’re going to upskill our individuals so that we still have a great workplace and workforce that we will always need. I think you’re going to see AI almost become the exoskeleton for the individual. The individual still may be at the center, but you’re going to make them bigger, faster and stronger.
The vendors I talk with would have us believe we’ll have fully autonomous contact centers soon.
Steele: I think about our roadmap — where we have deployed chat, as an example. That allows for the contact center to be more digital-oriented, versus voice-oriented. Chat will then improve because as we load AI capabilities into that environment, it allows us to automate more. That would reduce the need for the voice side — listening to what people are saying, how they’re saying it, and making sure their call gets to the right place.
You could argue that as the customers go through some voice prompts, they’ll get to the right engineer three layers up, potentially — a senior system engineer — if they are experiencing an event that requires one, such as their network is down.
We will be looking at those from a contact center perspective. Will you ever get away from humans 100%? I think that will be a difficult process.
We’ve seen the automation go through hundreds of thousands and save 30,000-plus hours. Agentic AI is one of those opportunities where you can improve your customer experience and drive your efficiency and cost structure at the same time. Things like that don’t come around that often. That’s one of the reasons why we are driving on this so hard.
Don Fluckinger is a senior news writer for Informa TechTarget. He covers customer experience, digital experience management and end-user computing. Got a tip? Email him.