By taking agentic AI seriously, leaders today must reassess and reimagine processes for synergies across the enterprise, setting the foundation for scale. Major oil and gas companies have been around for over 100 years, and employees do things because they always have, for reasons that may have become untethered from reality. The mantra should be to do new things with AI, not to do old things differently, and oftentimes that begins with finding and eliminating unnecessary churn in systems.
Executives also find themselves in an age-old quandary with technology: how to push for productivity gains while reassuring employees concerned about job security, which inhibit AI adoption and experimentation.
For oil and gas companies, responsible AI involves:
- Know where AI is being used and tier models according to risk levels because limited resources require prioritization. Executives should be asking questions like: What data is feeding the model? What is the output?
- See the organization in full to understand its patterns and behaviors, leveraging more tools than ever to address safety and risk without minimizing the upside of AI. Modernize infrastructure enough so that AI can be leveraged from end to end. Without cohesiveness from cross-functional data, for example, AI can be ineffective or even destructive, highlighting the perils of “garbage in, garbage out.”
- Understand that “responsible AI” is what enables organizations to do more with AI, especially as regulation takes shape — much like brakes on a car provide the confidence to accelerate.
Power and utilities
In this sector, AI has been spoken about more in aspirational terms, as silos and different operating companies make it harder to achieve the end-to-end thinking that AI needs across enterprises. And power and utilities (P&U) is the only industry with three charts of accounts/ledgers: for Federal Energy Regulatory Commission reporting, tax reporting and SEC reporting. That’s a separate layer of systems and processes that a lot of other industries don’t have.
Now the conversation is beginning to shift as agentic AI becomes more of a reality and as more P&U organizations complete enterprise resource planning system upgrades or add more advanced smart meters. Companies are asking: Can I leverage AI to make our transformation faster and cheaper? And how can I make sure the future state has AI embedded in it?
Using AI to assist customer service representatives is one thing; automating it from end to end is more powerful, but it’s a huge risk if, say, customers follow improper advice from an agent and have their lights shut off. Responsible AI provides the solution for risks such as these and allows agentic AI to potentially flourish.
This is borne out by our work with clients, who are typically avoiding high-risk or high-reward use cases. “When I go to the executive committee, data and AI governance projects get signed immediately, but for the use case in operations or customer service, they ask more questions,” one EY client executive said. “When I’m bringing in the guardrails under responsible AI, those have a faster path to get approved.”
Today, productivity and efficiency gains are not expected to translate into headcount reductions. P&U companies are more concerned with reducing operations and maintenance costs while achieving public-facing metrics such as reliability, safety and customer satisfaction — goals that must also be affirmed in their rate-making cases. Yet the other side of AI is the demands it is expected to place on the grid, forcing executives to think in terms of long-term plays, not quick wins on labor costs.
Tactics for P&U executives to move the needle on AI include:
- Capitalize AI as an asset up front and then deploy more operations and management use cases to lower costs. Insight readiness and preparedness in operations will be focus areas more than in other sectors, including workforce management and optimization and predictive maintenance, as a way to avoid outages and boost impressions with customers and commissions.
- Get ahead of the curve on data maturity and controls. Invest in data as one would with any other enterprise asset. P&U organizations are good at protecting large data silos. But integrating disparate data is hard, as a common data model is often missing across fragmented systems and applications. The old mentality for P&U was, “You don’t want to be last but second-to-last.” The industry is shaking off that mindset, with more prodding from regulators, but it remains behind other sectors. Most P&U organizations also don’t have a high-ranking lead for data, showing how it’s not prioritized as much as it should be.
- Appeal to talent with a disruptive mindset. The future workforce sees P&U as a bit stodgy as compared to pure tech companies, financial services or retail and consumer products. As a result, natural, tech-led cultures are not prevalent in the subsector, and people need more training. Pockets of progress are more likely to develop than functional disruption or enterprise transformation. Create a culture that embraces change, strengthened with responsible AI.
