This browser is not actively supported anymore. For the best passle experience, we strongly recommend you upgrade your browser.
| 5 minutes read

Have you hugged your chatbot? How utilities leverage AI to boost operational excellence and customer satisfaction

Utilities face many challenges in 2024, from continuing to deliver safe, reliable, yet affordable energy, to gathering and harnessing the sea of data collected each day.

Improving customer experience and operational efficiency remain priorities. Recent customer satisfaction studies published by J.D. Power suggest that satisfaction with electric and gas utilities remains in decline. Investors, meanwhile, show concern with utilities’ ability to manage inflationary pressures, while meeting increasing investment needs driven by an accelerating energy transition. 

Artificial intelligence can play a role in addressing these issues. As AI use cases expand at an explosive rate, we are only beginning to come to terms with how these new technologies will impact future energy-demand requirements, and utilities’ interaction with this technology is still in very early days. We believe that utilities who seize on practical applications focused on driving rapid results will gain a tremendous advantage over slower-moving peers. 

Understanding AI’s use cases, scope, and impact

AlixPartners’ recently-published 2024 Disruption Index identified AI as among the hottest topics on the minds of corporate leaders. More than two-thirds of 3,100 global executives said AI and automation are the technologies with the biggest disruptive opportunities, but most businesses aren’t sure what to do with them. As highly regulated entities, utilities face even greater legal and compliance frictions to deploying new tools than other organizations in the broader economy.

Let’s start with definitions. There are two categories to understand: Analytic AI and Generative AI.

Analytic AI is quantitative in nature, and preceded generative AI. Analytic AI finds relationships in numeric data to make forecasts, including predicting optimal values in business processes. Examples of analytic AI include recommending products and forecasting the performance or failure of machines.

Generative AI is subjective. Generative AI associates a prompt with related text, concepts, or themes in an underlying knowledge base. While generative AI started with large language models, the technology has evolved to multi-modal foundation models that work with images, videos, and audio. For a deeper dive, look at Generative AI’s second act.

Achieving business results generally requires both approaches. Analytic AI quantifies cause and effect to optimize steps in a business process, while generative AI finds company knowledge and connects with people. Some examples:

Edison International monitors conditions in California that predict potential outages. Large data sets need to be gathered and analyzed by analytic AI and machine learning algorithms. For example, smart meter data is used to detect unusual voltage fluctuations, while drones capture 360-degree images around electric poles to identify areas of concern. 

Similarly, First Energy is utilizing analytic AI via edge computing to process and prioritize pictures of transmission assets for Asset Management. 

Iberdrola Brazil utilizes analytic AI to handle customer requests through WhatsApp. This allows the local utility to meet the customer where they prefer to interact. As it turns out, that means many customers prefer skipping phone calls in favor of texting. This digital channel enables several capabilities, including meter setup, energy assessments, bill downloads, and payments.

On the generative AI front, many utilities are deploying AI as virtual conversational agents, while others are using it to craft content for consumption by customers. 

U.K.-based Octopus Energy drafts customer emails with generative AI. Human supervisors check for quality and send. These emails enable faster throughput than purely handcrafted or form-style correspondence created by human support staff. Generative AI now replies to one-third of customer emails, handling the work of 250 people while earning customer satisfaction scores of 80%, compared to a 65% baseline. 

Another large U.S.-based utility uses generative AI for several tasks, including: 

  • Improving organizational communication: Generative AI drafts written communications, streamlines meeting minutes, writes memos, crafts business cases, and composes emails—as well as optimizes clarity and format.
  • Enabling regulatory knowledge: Generative AI summarizes public utility commission filings and documents. It enables new employees to find and learn critical information much more quickly, without requiring support from scarce resources (more seasoned peers).

Italy’s Iren, meanwhile, uses a combination of analytic and generative AI to create personalized texts for individual electricity or gas customers. Analyzing the requisite usage data, customer by customer, in the absence of modern, AI-driven digital tools, is data and labor intensive. Instead, Iren crafts personalized texts for offers to appropriate customers based on their individual usage history.

 

To further understand the use cases, it helps to review the opportunity from the customer and employee journey point of view. We believe that AI has the potential to positively impact the utility customer experience (CX) and employee experience. 

One practical CX use case involves using AI to proactively monitor, report, and potentially respond (within pre-defined limits) to customer comments and sentiment on social media. Consider a situation in which someone posts a picture of an accident scene that involved a car striking a distribution pole on social media. AI can be trained to scan for and quickly highlight this information for review. Imagine the SVP of customer experience responding in moments on that same social media platform with a note saying, “Thanks for sharing this info. We have notified emergency officials and have a crew directed to that area to fix it.”  
 

Organize appropriately and manage the risks

While the business case for practical, business-case driven AI applications is already strong for utilities, as stewards of the public trust, it is critical for organizations to manage the risks, and to make an explicit choice about the levels of intended coordination and control, which may be at odds with the speed of innovation and adoption.

 

Utilities seem to be favoring a more centralized and controlled approach to the rollout of AI tools in most organizations. Our observations suggest that the most successful rollouts are focused on addressing a specific business need, with a clear and compelling business case that may include AI as one part of a broader effort; and that might also include simple rules-based automation, robotic process automation, lean principles, etc.

Organizations adopting the use of AI are confronted with a set of risks, including:

  • Privacy, including how consumer data is stored and used to train models
  • Fairness in outcomes across groups of people and scenarios
  • Transparency, to readily explain how a model functions
  • Third party data, which can introduce copyright, accuracy, and bias risks
  • Accuracy, including “hallucinations” in which Generative AI may convincingly construct an answer that is not factual
  • Investing in pilots without a clear and compelling business case to address an existing need

Mitigating these risks starts with ensuring that AI efforts are designed and implemented with oversight across all functional areas. Compliance, risk, legal, IT, environmental health and safety, finance, and operations should understand the process to build models so they can confidently guide decisions. Leaders and project team members should understand the key role that “training data” plays in shaping these models.

In addition to aligning horizontally across functions, project teams should connect vertically from leadership to front line workers (e.g., cascading goals). Contact center representatives and the frontline field workforce are often best positioned to understand what data mean from a day-to-day use perspective, and to identify data and data quality needs. Importantly, frontline workers can often best identify how to act on a model’s forecasts.

Find a trusted partner

AI offers high-value use cases to the utility industry. If you’d like to learn more or discuss how we can help your organization drive more value, please reach out.

Tags

article, artificial intelligence, disruption, americas, english us