Reprinted with permission of Energy Central
The customer call centers at utilities across the country all deal with the same pain points: long queues, frustrated customers, and limited budgets. When dealing with this most human-facing part of the power sector, utilities have historically had only one lever to pull: hire and train more customer service staff.
What if, however, that esteemed group of employees at utilities had a force multiplier to expand their ability to meet the customers where they are? That’s the opportunity the City of Kingsport saw when it “hired” ‘Grace’. But Grace isn’t like the other employees in the call center.
Grace isn’t an employee at all, she’s an AI agent.
For Kingsport, this effort represents a deliberate move beyond the clunky chatbots and rigid systems that have long frustrated customers and utilities alike. As shared in recent conversations that Floyd Bailey, Chief Information Officer at the City of Kingsport, and Bobby Slaton, EVP Americas Energy & Utilities at Hansen Technologies, had with Energy Central, Grace (as City of Kingsport has named the Hansen AI Agent solution) is the next progression in AI-assisted customer service that customers will enjoy speaking to, employees will be happy to take tedious work off their hands, and the utility will benefit from without any of the perceived downfalls.
Kingsport’s Decision
The City of Kingsport, Tennessee operates a municipal water utility serving roughly 40,000 customers. While that footprint may seem modest compared to large, investor-owned utilities, the operational reality tells a different story. Kingsport’s customer service team numbers only about a dozen people responsible for handling billing questions, service requests, and customer concerns across the entire city. For a team of that size, the volume and pace of daily demand can quickly become overwhelming.
That imbalance creates familiar pressure points: limited customer service hours, phone lines tied up with high-frequency transactions, and a constant struggle to keep up with expectations using a lean workforce. As one Kingsport leader put it, “Our biggest struggle is just getting everything done. People think government is slow and imagine it’s because the people are slow. It’s really not—we just have a lot on our plate.”
Kingsport’s experience is far from unique. Across municipalities of all sizes, utilities are being asked to deliver faster service, greater transparency, and more digital convenience—often without the staffing or budget flexibility to match. These constraints fundamentally shape how cities evaluate new technology, especially solutions that sit directly on the front line of customer engagement. What Kingsport faces today reflects a broader reality for municipal utilities everywhere: doing more, with fewer people, under rising expectations.
Introducing ‘Grace’
To help manage that overflowing plate, the City of Kingsport teamed up with Hansen Technologies to introduce to its citizens a conversational AI agent, developed in conjunction with Dial AI who is Hansen’s strategic partner for conversional AI capabilities. Hansen AI’s Agent sits in front of and alongside customer service representatives (CSRs). Appropriately named Grace by the City, given the calm, patient nature of the voice on the other end of the line, the Hansen AI Agent integrates directly with Kingsport’s customer information system (CIS) and customer relationship management (CRM) tools. Integrating with these platforms, Grace can access account context in real time while managing customer interactions via voice, chat, or SMS interactions.
While any new integration of AI stepping in where customers are accustomed to dealing with real humans on the other line can be a testy endeavor, Grace is being introduced in a very managed, deliberate rollout to prevent any sort of drop shock by citizens. Slaton highlights that Grace’s instructions can be tweaked and tuned iteratively in near-real time. A massive overnight transformation “isn’t required,” Slaton explains. “We can allow Grace to take five calls the first day and analyze how she did. Then we can make adjustments and have her take 15 calls the next day.”
From a technical perspective, Grace brings several capabilities that differentiate it from earlier generations of automated customer service tools:
- Integrations: Grace hooks into the customer records so she can read account history and proactively suggest actions (e.g., offer e-billing, payment plan).
- Omnichannel memory: Grace can instantaneously recall prior conversations and apply context (across calls, chat, email) when handling a new interaction.
- Scalability: Hansen can run many AI Agent instances in parallel — the “hold time” metric effectively collapses because many “Graces” can attend simultaneous callers.
- Advanced intent & sentiment recognition: Grace supports real-time language switching between English and Spanish, detects sentiment and emotional cues, and adjusts tone accordingly (so a chipper tone and unempathetic response isn’t included if, say, a customer calls about struggles to pay their bills).
“She can check the CIS and say ‘I see that your water bill is significantly higher this month…’” offers Slaton. “It feels like it’s not just a transaction. It’s a technology that understands your personal situation.”
Mapping Grace’s Debut
‘Grace’ is being slowly rolled out in Kingsport as a part of that thoughtful and intentional deployment strategy. Moving from proof of concept to testing internally to testing with utility friends and family are all steps to make sure Grace is ready for her debut. “We’re going live in spring 2026,” confirms Bailey. “Right now, we’re in testing, user acceptance and training of the staff.” In the meantime, Kingsport is making sure the community is informed and ready for this launch.
In particular, Hansen and the City of Kingsport have deployed a physical phone booth in City Hall where any passersby can pick up the receiver and talk to an early version of Grace. This allows citizens to “try” ‘Grace’, the AI Agent, in a non-threatening way. Further community demo booths and in-person outreach will continue in the coming months as well.
When Grace’s switch is effectively turned on, Bailey and Slaton confirm that she will start with high-volume, low-risk flows such as bill pay, account lookups, and status checks. These types of calls are straightforward to handle but can clog up the human customer service agents who are more urgently needed to deal with unique and unexpected customer requests. Over time, though, these teams expect to add tougher flows as confidence in Grace grows.
Benefitting the Existing Workforce
Beyond concerns about customer acceptance, another key area of focus on rollout of an AI tool is the existing workforce. Both Bailey and Slaton highlight that the goal of Grace is not to replace existing customer service agents or shrink the workforce, but instead to be considered a force multiplier.
Grace can automate repetitive and tedious tasks that clog up the to-do lists of the human agents. That way, Kingsport’s existing customer service agents can focus their time on complex problems and further build relationships with their customers. Those who call in with such more complicated needs will see their wait times drop dramatically, while Grace also enables 24/7 coverage for the basic tasks that was previously unaffordable to the utility. In that way, both customers and employees stand to benefit from an assist by Grace.
Bailey echoed that staff worried at first about job loss, but highlighted that when they were shown the tool (as well as the up-skilling path its presence opens up to them) they responded positively. “We have no intention of getting rid of the staff we have,” Bailey explained. “But we have every intention of being able to provide more time and efficiency for our existing staff.” And on a bigger picture, Bailey notes that “AI needs to take away the time wasters so that humans can dream.” Slaton agreed, saying “Let’s offload the repetitive tasks to the AI and allow the customer service reps to focus on the more complex cases. They’re excited about that.”
Measuring Results
Kingsport has much that they will be tracking and aiming to learn as Grace rolls out Bailey reported that Kingsport expects to reach a 75% acceptance rate over time with messaging and demonstrations like the phone booth being key. A common consumer question, as expected, was whether they could still get to a human agent, so Kingsport ensures an easy human escalation path and emphasizes transparency in outreach. Beyond adoption rates, Kingsport will also be tracking metrics like call deflection, average handle time for human agents, and customer satisfaction as Grace scales.
In terms of expected acceptance, Slaton highlighted there will be differences between younger and older demographics. The younger customers are more likely to prefer non-human channels of communication, making Grace perfect. Meanwhile, the older demographics may be more apprehensive, making demos for them and assurances of the ability to reach a human when needed all the more critical.
Lessons Learned: Building Your Own ‘Grace’
In all likelihood, Grace-like tools will be on their way to your utility in the future, so in terms of lessons learned via the Kingsport experience, Bailey and Slaton emphasized the following:
- Start with high-volume, low-risk use cases (bill pay, status checks) and grow from there.
- Run user acceptance testing and friends-and-family pilots and use demos to build community buy-in.
- Ensure easy, transparent human escalation (“press zero” equivalent) so no one feels forced into this.
- Pick a vendor partner you trust, define success metrics together, and plan iterative rollouts.
The future of AI in customer service increasingly looks like a hybrid model – automation handling the transactions, and humans focusing on what requires judgment, empathy, and trust.
