Keeping the Lights On: How AI Helps Utilities Support Customers During Outages

It’s a moment many of us can relate to: the lights flicker out, the tv cuts and we instinctively reach for our phones to seek updates, answers and reassurance. Per the latest U.S Energy Information Administration data, US electricity customers experienced an average of 6 hours and 11 minutes without power due to service disruptions in 2023. During peak outage seasons, when disruptions grow more frequent in certain regions, utility providers’ contact centers are put to the test. A spike in customer inquiries, rising frustration, and the expectation for fast, transparent communication all converge, demanding swift action under considerable pressure.

Utilities have long operated according to this reactive model. When service disruptions strike, the default response has been to enlist the help of all (human) hands available in the contact center. The data indicates this model isn’t sustainable. 75% of agents report they are already overwhelmed by systems and information, leading to long call queues and poor outcomes on both ends of the phone. A high volume of disconnected calls - especially during service disruptions - exemplifies this problem.

An investigation by the Colorado Public Utilities Commission found that roughly 10% of customer service calls made by residents to Xcel Energy in 2024 were disconnected before any resolution was provided. This same year, the Commission found that Xcel Energy’s average Colorado customer experienced more than double the total amount of outage time compared to years prior.

To handle surges in customer inquiries during service disruptions, utility companies must adopt AI as a complement to their human agents. AI technologies enable companies to automate many customer service interactions by engaging customers in a natural conversation with an AI agent that’s equipped with utilities-specific knowledge.

AI solutions can triage high volumes of customer requests in real time, instantly resolving repetitive queries and seamlessly escalating calls, text, and digital messages that human agents previously handled. When a handoff from AI agent to human agent is the necessary next step, the AI generates a full summary of the interaction and call history so there’s no need for a customer to repeat themselves. This isn’t just automation. It’s intelligent, human-like support that helps utility providers scale their customer service during outages to remain compliant with state regulations and improve outcomes for both customers and employees.

Customers no longer need to navigate the notoriously frustrating rigid menu hierarchies of Dual-Tone Multi-Frequency (DTMF) or hear the monotonous “press 1 for billing”. They simply say in their own words what they need to do, for example “check my statement,” and the virtual AI agent understands and responds to complete the task.

Cyber threats continue to rise and the risk of voice fraud is a growing challenge for utilities. The emergence of Generative AI has attracted fraudsters that exploit the technology for nefarious means - taking advantage of easily accessible voice spoofing and deepfake technologies to attack contact centers. A single line of defense is no longer enough, making a multi-layered approach essential. This includes voice biometrics to analyze the unique vocal characteristics of a caller to authenticate customers, spoof detection, a dynamic block list of known deepfake bot offenders, the detection of synthetic voices and replay attacks, and logic-and-coherence checks. 

This enables utility companies to streamline customer authentication while maximizing self-service and keeping their customers’ data and their contact center safe.

Moving customers to natural language self-service applications requires changing the perceptions of those who have historically only used, and tried to circumvent, touch-tone systems. Utilities providers can increase adoption of automated self-service by prioritizing what matters: fast, accurate information leading to an immediate resolution, and precision in routing to the appropriate person when required.

Finding solutions that result in a quick and efficient resolution is especially critical when utility companies are trying to find ways to reduce spending. Xcel Energy’s contact center budget in 2024 was cut by 5% compared to 2020, the PUC discovered when conducting an investigation on service and sales complaints.

For human agents facing peak call volumes, AI streamlines their workload to prioritize those customers in need of a human touchpoint: in moments of true urgency or emotional distress, AI knows when to step aside. Sentiment analysis features help detect when a customer issue needs to be steered carefully and advise the human agent during the call. AI agents can handle back-to-back calls about payment due dates or outage maps so the staff get time and space to focus on the customers who need extra care. Many utility companies serve a diverse customer base, so AI agents are equipped to deliver seamless voice and digital experiences in multiple languages. Bringing in an AI customer service solution that reduces burnout and better supports agents is a critical win in an industry plagued by staffing instability.

It’s also a financial imperative. Utility companies today must navigate a growing set of cost pressures, ranging from aging infrastructure to increasing scrutiny from regulators, often raised by frustrated customers. Not only do poor service outcomes damage customer trust, but they also carry real financial penalties.

Failing to meet state-mandated customer service benchmarks can result in millions of dollars in fines. In June 2025, New York State Electric & Gas Corporation (NYSEG) was fined $5.6 million for missing the 2024 customer satisfaction target. Because NYSEG had also failed to meet the metric in 2023, regulators imposed an additional penalty, bringing the total to $11.2 million. These fines underscore the cost of underperformance and the need for more scalable, resilient customer service strategies, particularly during periods of disruption when call volumes spike and a resolution is delayed or dropped entirely.

Investing in AI-powered contact center solutions is no longer a matter of convenience; it provides operational agility and is a regulatory and financial safeguard. By improving first-contact resolution, reducing wait times, and delivering more consistent service quality, AI tools help utility providers stay ahead of compliance thresholds and avoid costly consequences. Looking ahead, AI will play a central role in building customer-focused communication channels that are not only faster and more efficient, but also resilient and responsive, making it an essential part of how utilities prepare for and respond to service disruptions.