Beyond the Pilot: How Utilities Are Operationalizing Gen AI
UAI Generative AI Community Conversation | August 13, 2026
Generative AI is quickly moving from experimentation toward real-world utility applications, but getting from a successful proof of concept to a sustainable enterprise capability remains a significant challenge.
That was the focus of the August 13, 2026, UAI Generative AI Community Conversation, “Beyond the Pilot: Operationalizing Generative AI for Enterprise Deployment.”
The panel discussion was moderated by Chat Hull, Ph.D., Head of Data Science & Analytics – AES US Utilities, and featured the following panelists:
- Sreenivas Badri, Sr. Director, Grid and Market Solutions – ERCOT
- David Robertson, Sr. Mgr Advanced Analytics – Duke Energy
- Mike Turner, Chief Data Officer – Austin Energy
Together with participating UAI community members, the panel explored what utilities need to move Generative AI from proof of concept into production, including governance, data readiness, operational ownership, workforce adoption, trust, scalability, and the growing range of AI use cases emerging across the industry.
Most Utilities Are Still Piloting Gen AI
A live poll conducted during the session provided a snapshot of where participating organizations currently stand in their Generative AI journeys.
Of 11 respondents (out of approximately 40 attendees):
- 82% (9) are running pilots and proofs of concept.
- 9% (1) are deploying production use cases.
- 9% (1) are scaling AI across multiple business areas.
- No respondents indicated that they were still only exploring possibilities.
The results suggest that experimentation is well underway across the UAI community. The next challenge is turning those experiments into sustainable production capabilities.
That transition was reflected in another poll question asking participants which operationalization challenge they were most interested in discussing. Deploying AI into production was the clear leader, selected by 5 of 11 respondents, followed by measuring business value and ROI and change management/workforce adoption, with two responses each.
The results point to an important shift: the question is increasingly becoming less about whether utilities should experiment with Generative AI and more about how they can operationalize it responsibly.
Operationalizing Gen AI: Strategy Comes Before Scale
One of the strongest themes from the discussion was that successful AI deployment should begin with a clear strategy rather than with the technology itself.
Organizations need to understand what they are trying to accomplish, which business problems are worth solving, who will use the solution, and what success should look like. From there, utilities can determine the governance, technology, data, security, and operating structures necessary to support those objectives.
As the panel discussed, the AI strategy does not necessarily need to exist independently. Depending on the organization, it may be incorporated into a broader enterprise data or technology strategy. What matters is having a clear direction and connecting individual AI use cases to meaningful organizational priorities.
The discussion also raised an important question utilities are actively working through: Should AI governance exist as its own discipline, or should AI ultimately be governed as another component of an organization’s broader data and technology environment?
There was no single answer. The appropriate model will depend on the organization, its maturity, risk profile, regulatory environment, and intended AI applications.
Gen AI Governance and Risk Remain Major Barriers
When participants were asked about the biggest obstacle preventing broader Generative AI adoption, governance and risk ranked first, selected by four of 11 respondents. Change management and user adoption followed with three responses, while data readiness received two. Cybersecurity/privacy and demonstrating business value each received one response.
The panel emphasized that organizations moving toward production need many of the same fundamentals required for strong data and analytics programs: clear data ownership, stewardship, access controls, security, data quality standards, and defined accountability.
AI adds additional considerations.
Utilities are developing governance processes that evaluate proposed AI use cases before deployment and bring together perspectives from areas such as technology, cybersecurity, legal, regulatory, data, and the business. These reviews help organizations identify potential risks, establish appropriate guardrails, and consider what could go wrong before an application reaches production.
The conversation reinforced a simple principle: AI governance should enable responsible adoption, not exist separately from the business problems AI is intended to solve.
Gen AI Starts with a Strong Data Foundation
Even the most sophisticated Generative AI application cannot overcome poor underlying data.
Panelists repeatedly returned to the importance of data quality, ownership, lineage, integration, and governance. Utilities often have operational, market, customer, asset, engineering, and other information distributed across numerous systems. Creating trusted, governed data environments that allow information to be connected across those systems is therefore a critical prerequisite for more advanced AI capabilities.
The discussion also highlighted the relationship between modern data platforms and AI. As utilities consolidate and govern data within modern analytics environments, they create a foundation that can support not only traditional reporting and analytics but also conversational analytics, Generative AI, and eventually more advanced agentic workflows.
In other words, AI readiness and data readiness are increasingly inseparable.
Building Trust Is Critical to Gen AI Adoption
Technology was only part of the operationalization conversation. Much of the discussion focused on people.
Utilities operate critical infrastructure, and employees responsible for operations, engineering, customer service, finance, and other functions need confidence in the information AI provides before they will incorporate it into their work.
That makes trust a prerequisite for adoption.
One approach discussed extensively during the session was keeping a human in the loop, particularly during early deployments. AI can generate a recommendation, response, analysis, or proposed action, but a subject matter expert reviews and approves the output before it is used.
This model allows organizations to introduce automation gradually while employees develop confidence in the technology.
Explainability is particularly important for operational applications. Users need to understand where an answer came from, which information was used, and whether the output can be trusted before relying on it in higher-risk environments.
The broader lesson from the discussion was clear: do not treat adoption as a technology rollout. Treat it as an organizational change initiative.
Start With Friction, Not Full AI Autonomy
Several examples shared during the conversation demonstrated the value of beginning with AI applications that make employees’ existing work easier rather than immediately attempting to replace entire processes.
One panelist described an AI-assisted field workflow in which employees initially retained the ability to accept or reject AI-generated work recommendations. Resistance eventually declined as users discovered that the technology removed a frustrating, repetitive part of their work and allowed them to concentrate on higher-value activities.
The progression discussed during the session could be summarized as:
Assist → Build Trust → Automate → Scale
That approach can be especially valuable in utilities, where safety, reliability, compliance, and operational consequences make immediate full autonomy impractical for many applications.
Education is part of that progression as well. Community discussion highlighted the value of introducing employees to practical AI applications through demonstrations, learning sessions, examples, and repeatable workflows. As employees see AI solve recognizable problems, adoption can become less abstract and more connected to their daily work.
Where Utilities Are Focusing Their Gen AI Efforts
The community poll also provided a useful picture of where organizations are currently concentrating their Generative AI efforts.
Among 10 respondents:
- Employee productivity — 7
- Knowledge search and document intelligence — 5
- AI agents/workflow automation — 5
- Grid/operations/engineering use cases — 2
- Customer-facing solutions — 1
Respondents could select more than one area.
Employee productivity remains the most common entry point. Copilots and assistants generally offer lower-risk opportunities for organizations and employees to learn how to work with Generative AI.
At the same time, the strong interest in knowledge search, document intelligence, and workflow automation suggests that utilities are beginning to move toward applications that integrate AI more deeply into business processes.
Customer-facing applications remain less common, reflecting the additional risk, governance, accuracy, and trust considerations associated with putting AI-generated information directly in front of customers.
Gen AI Use Cases Emerging Across the Utility Industry
One of the most valuable outcomes of the conversation was the breadth of use cases discussed. Some are already in production, while others remain in development, testing, or conceptual stages.
Use cases mentioned during the session included:
- Employee copilots and productivity assistants
- Enterprise knowledge search and document intelligence
- Operations support assistants using operating procedures, protocols, design documentation, application documentation, and test procedures
- Engineering and operational knowledge assistants
- AI-assisted contingency analysis and operational support
- Generation and resource interconnection inquiry response
- Large-load inquiry support
- Automated email intake, knowledge retrieval, and draft response generation
- Customer bill explanation, including identifying factors contributing to month-over-month bill changes
- Automated API creation and publishing
- Automated IT/service ticket classification and routing
- Call center transcript diagnostics
- Customer experience and process issue detection from call transcripts
- AI-assisted field work sequencing
- Risk registry analysis and forecasting
- Budget forecasting and scenario analysis
- Natural-language access to enterprise data and analytics
- Conversational analytics for executives and business users
- Self-service analytics enhanced by Generative AI
- Forecasting and Monte Carlo analysis
- AI-assisted software development
- AI agents and workflow automation
- Automated anomaly detection and auditing
- Grid, engineering, and operational decision support
Several examples demonstrated how Generative AI can serve as an interface between employees and increasingly complex data environments. Instead of waiting for an analyst to generate every new scenario, future applications could allow trusted users to ask questions in natural language and explore governed data within established parameters.
Other examples showed AI moving beyond information retrieval into workflow execution, including creating APIs, routing tickets, analyzing customer interactions, drafting responses to inquiries, and supporting operational processes.
From Self-Service Analytics to Self-Service AI
One particularly interesting idea that emerged during the conversation was the progression from traditional reporting to self-service analytics and, eventually, self-service AI.
Many utilities have spent years developing centralized data environments and self-service analytics capabilities so business users can answer more questions without submitting individual requests to data teams.
Generative AI could extend that concept considerably.
Rather than selecting predetermined filters from a dashboard, authorized users could eventually interact with governed enterprise data conversationally, asking questions, exploring scenarios, requesting forecasts, or evaluating different assumptions through natural language.
That vision carries tremendous potential, particularly for organizations whose analytics teams are supporting large numbers of internal stakeholders. But realizing it depends on the same foundations discussed throughout the session: trusted data, clear governance, strong security, defined use cases, appropriate guardrails, and user confidence.
Five Takeaways for Operationalizing Gen AI in Utilities
The discussion surfaced five principles for organizations preparing to move Generative AI into production:
1. Start with strategy and business value.
Define what the organization is trying to accomplish and identify use cases connected to meaningful business needs before selecting technology or building infrastructure.
2. Build on a trusted data foundation.
Data governance, quality, ownership, lineage, access, and integration remain essential. AI does not eliminate these requirements; it makes them more important.
3. Establish governance and guardrails appropriate to the use case.
AI governance should include the right combination of business, technology, cybersecurity, legal, regulatory, data, and risk perspectives.
4. Keep humans involved while trust develops.
Human review can provide an important bridge between experimentation and automation, particularly for operational or higher-risk applications.
5. Treat workforce adoption as part of deployment.
Education, communication, demonstrations, early wins, and employee involvement are critical. Sustainable AI adoption requires changing how people work, not simply giving them new technology.
The Next Phase of Gen AI in Utilities Is Organizational
The August 2026 UAI Generative AI Community Conversation demonstrated that the utility industry’s AI journey is advancing.
Experimentation is no longer the primary hurdle for many organizations. Utilities have pilots. They have use cases. They have access to increasingly powerful technology.
The harder question is how to turn those capabilities into something the enterprise can trust, govern, operate, maintain, and scale.
The answer emerging from the UAI community is not a single technology platform or AI model. It is a combination of strategy, strong data foundations, governance, thoughtful use-case selection, human oversight, workforce education, and incremental trust-building.
Moving beyond the pilot is ultimately about more than putting Generative AI into production.
It is about preparing the organization to operate AI as a sustainable enterprise capability.
What Happens Inside a Community Conversation
This session exemplifies how UAI Community Conversations create value beyond what can be captured in a written summary. Members challenge assumptions, compare approaches, and share lessons learned from real implementations.
While this article highlights discussion themes, the detailed methodologies and peer-driven tradeoffs remain part of the member-only experience.
What Makes UAI Community Conversations Different
UAI Community Conversations are member-led, interactive, and grounded in real-world application. They are designed to shorten learning curves and help analytics teams make better decisions faster—together.
Join the Conversation
UAI brings together analytics professionals across utility ownership models, service types, and regions to exchange applied knowledge and advance the practice of utility analytics.
To learn more about UAI membership and how to participate in Community Conversations, visit: https://utilityanalytics.com/about-utility-analytics/.
About the Utility Analytics Institute (UAI)
The Utility Analytics Institute is a community within Endeavor Business Media that supports utility analytics professionals through member-led conversations, events, recognition programs, and industry research.
About this summary
This summary was produced using generative AI, with human editorial oversight to ensure alignment with UAI content guidelines.
About the Speakers
Chat Hull
Chat Hull is the Head of Data Science & Analytics for the AES US Utilities, where he leads teams focusing on data science, AI, and advanced analytics; data engineering and architecture; data governance; and reporting. At AES, he and his team focus on using data, analytics, and AI to craft practical, high-impact solutions that support a cleaner, more reliable energy future. Before entering the utilities industry, Chat worked as a professional astrophysicist, bringing a unique scientific perspective to solving complex business challenges.
Sreenivas Badri
Sreenivas Badri is Sr. Director, Grid and Market Solutions under digital organization at ERCOT, responsible for design, development and support of Grid and Market systems and applications and control room situational awareness tools. With over 21 years of experience of designing and building Energy Management Systems (EMS) and Market Management systems (MMS) applications and integrating with other operational IT systems, Sreenivas has successfully delivered numerous grid and markets technology and functionality upgrade projects across his career.
Sreenivas holds a master’s degree in electrical and computer engineering with a specialization in Power Systems Engineering from Auburn University, Alabama.
David Robertson
David Robertson is Senior Manager of Advanced Analytics at Duke Energy, where he leads teams delivering AI, data science, and advanced analytics solutions that enhance customer experiences and drive business value. With expertise in AI strategy, customer analytics, and enterprise transformation, David focuses on helping utilities apply artificial intelligence responsibly and effectively to improve decision-making, operational performance, and customer service outcomes.
Mike Turner
Mike Turner is the Chief Data Officer (CDO) at Austin Energy, with over 20 years of experience in Analytics and Strategy. In his role, he oversees data assets and promotes innovation within the organization.
As CDO, Mike is responsible for data analysis, governance, compliance, quality assurance, and integration efforts. He also leads the development of business intelligence strategies to improve Austin Energy’s performance. His extensive experience enables data-driven decision-making throughout the utility.
Mike’s career includes serving as a Strategic Planner in the Air Force and establishing the Quality Assurance division for Austin Resource Recovery (ARR). He has managed quality and data analytics programs for organizations such as the Department of Defense (DoD), the North Atlantic Treaty Organization (NATO), and the H-E-B grocery chain, among others.
He holds an MBA from the Forbes School of Business and certifications in safety management, Lean Six Sigma, and business analysis from Harvard Business School. Mike’s dedication to analytics and data-driven decision-making benefits both the City of Austin and its citizens.
About the Author
Leslie Cook, Manager, Membership, Engagement & Training Utility Analytics Institute (UAI)
Leslie Cook is a strategic program director, dynamic business leader, and expert facilitator with 30 years of professional experience. Leslie’s peers describe her as a passionate, authentic, action-oriented leader with a goal driven attitude and enthusiasm for creating connections and collaborations. She excels at developing and managing frontline programs that position companies and their products for sustainable revenue growth and delivering customer-focused solutions that demonstrate product value, build long-term loyalty, and foster client engagement and retention. Cook’s background includes extensive work with associations and institutes in program development and management, marketing communications, event management, branding, and sales/new business development. She holds a BA in public relations from Texas Tech University and served in various program director and marketing roles for the Utility Analytics Institute (An Endeavor Business Media company), Financial Planning Association® (FPA®), HDI (an Informa company), and the Global Semiconductor Alliance (formerly Fabless Semiconductor Association). When she is not out facilitating cutting-edge programs or creating unforgettable member experiences, Leslie can be found traveling the world and also enjoying all that Colorado has to offer, from hiking, swimming, and outdoor excursions with her fur babies, to antiquing for the best mid-century modern finds and blown glass of all kinds. She also leverages her high attention to detail and project management skills when dabbling in home remodeling and interior design, two of her favorite pastimes.
