How power and utilities companies can unlock AI’s true potential

A one-dimensional AI approach fails in this context. AI’s transformational potential is fundamentally about humans and their ways of working. To help achieve desired outcomes, a deliberate integration of three interdependent pillars is required to create a dynamic AI ecosystem that generates performance value rather than theoretical data points:

• Human-centered design and engagement 

• AI-integrated business process reinvention

• Outcome-driven collaboration and value governance

These pillars help drive real bottom-line performance by offering a practical perspective and actionable insights. They draw directly from EY experience in realizing value from AI projects through end-to-end collaboration with key stakeholders.

Human-centered design and engagement

According to studies by EY teams and the University of Oxford’s Saïd Business School, organizations that put humans at the center of their transformation journey are 2.6 times more likely to succeed than those that do not. For instance, AI models can flag potential equipment failures or identify optimization opportunities, but the value might only materialize when humans trust these insights and integrate them into their decision-making and daily workflows. Otherwise, AI risks becoming an underutilized dashboard that is disconnected from reality in the frontline.

For example, a leading utilities company in Malaysia recently won multiple awards for its AI smart grid and asset predictive maintenance initiatives. It emerged from an inclusive journey of building common understanding, trust and transparency and contextualizing AI to frontline workflows, turning users into proactive value contributors.

To turn rhetoric into action, companies can take several steps.

Involve process owners in AI initiatives

Process owners — from operational leaders to frontline supervisors — must be involved early and continuously in AI initiatives. This is key to codesigning workflows where AI is embedded natively into day-to-day operations and not implemented as an afterthought.

Leverage multidisciplinary teams

Shift the focus from siloed experts and AI-skilled teams to multidisciplinary teams. By establishing ongoing forums like communities of practice, operational leaders, subject matter experts, data scientists and AI engineers can cocreate AI use cases with greater ease. This serves as a crucible for mutual learning, trust building, challenging commonly accepted practices and contextualizing AI insights within operational realities.

Communicate AI benefits that resonate with stakeholders

Tailor stakeholder communication to focus on the benefits of AI adoption for each user persona to cultivate ownership and dispel fears. Provide practical, jargon-free training that reveals how AI can simplify tasks and augment capabilities rather than replace roles.

Create iterative feedback channels and support loops

Establish iterative feedback channels and support loops to help build user confidence and ease transitions to new operating models. This involves addressing questions promptly and integrating operator feedback to help refine AI tools and improve earlier outcomes.

AI-integrated business process reinvention

Introducing AI without re-engineering business processes is akin to adding horsepower to a clogged engine. For example, instead of layering analytics over existing workflows to integrate predictive AI into asset maintenance, a major American utilities company completely revamped maintenance scheduling, dispatching and inventory management by integrating AI insights in real time.

The company built feedback-rich ecosystems where AI tools, data streams and human workflows dynamically adapt, much like a digital twin. Its leadership also transcended outdated top-down, IT-driven approaches. As a result, the company not only improved maintenance cycles but also greatly reduced unplanned outages and costs, resulting in clearly tracked operational improvements.

The following are essential steps for embedding AI deeply into business processes to help achieve a sustainable operational impact and continuous improvement.

Conduct a comprehensive current-state diagnostic

Before deployment of technology, conduct a thorough assessment of existing processes, workflows, data maturity, technology applications and cultural readiness toward AI to understand holistically where the opportunities lie. Department heads should engage with frontline staff and managers and IT representatives in this exercise so that projects are grounded in reality before insights are shared with operational and executive leaders.

Identify AI-impacted processes

Visualize value chains by mapping operational workflows from end to end, then identifying touch points where AI insights can enable automation, prediction or decision support. This helps break down traditional silos to create ecosystems where AI tools, data streams and human workflows dynamically adapt.

Build AI into core IT and operational technology architecture

Move AI from isolated dashboards into integrated and intuitive process platforms where real-time data, AI algorithms, operational systems and tools communicate seamlessly. This would help enhance the speed and accuracy of decision-making as insights are not only observed but also acted upon.

Expand AI use cases iteratively

Once AI is integrated, instill a mindset of continuously turning data into insights and action swiftly and repeatedly. Go beyond obvious use cases to drive efficiency and agility, with AI initiatives treated as ongoing pathways of change that evolve with the business rather than one-off projects. This practice helps avoid “pilot purgatory”, a continuous state of experimentation that doesn’t result in tangible outcomes.