Creating AI-Ready Organizations

Artificial intelligence has moved from a buzzword to a strategic imperative for all industries including the global energy sector. While the technology itself is advancing at break‑neck speed, deeper obstacles to progress are cultural and organizational: legacy structures, siloed data, and a workforce trained for a pre‑AI world often outpace the pace of technical adoption.

If energy firms are to unlock AI's promise—lower‑cost operations, sharper emissions reductions, and new business models—they must redesign the very foundations of how they work. The Creating AI‑Ready Organizations panel at CERAWeek 2026 explored what that redesign looks like, who is leading the charge, and what we can expect to see on the horizon.

Panelists included: 

• Geoff Parker, Faculty Director, Irving Institute for Energy & Society, Dartmouth College
• Gwenaelle Avice‑Huet, Executive Vice President, Industrial Automation, Schneider Electric
• Debbie Pickle, Chief Human Resources Officer, Williams Companies
• David Rabley, Global Energy Strategy Lead, Accenture
• Rob Schapiro, Senior Director, Energy Partnerships, Microsoft
• Antonia Bullar, Moderator, S&P Global Climate & Energy Consultant

The panel's diversity—spanning technology, a leading integrator, a major upstream/downstream operator, a consultancy, and an academic thought leader—mirrored the ecosystemic reality that all panelists agreed is essential for an AI‑ready energy firm.

From "AI Strategy" to "AI‑Embedded"

Schapiro defined "frontier firms" as companies that have moved beyond an AI strategy to having AI woven throughout operations. In these firms, agentic AI—systems that can act autonomously, suggest actions, and even self‑optimize—is an embedded capability rather than an after‑thought. ADNOC's ambition to make every asset "self‑aware" by 2030 served as an illustration, with goals including cost savings, emissions reductions, and fewer safety incidents.

The Guardrails of Risk and Asset Context

inline-ceraweek-2026-ai-ready-geoff-parker.jpg

Irving Institute Faculty Director Geoff Parker.

Parker emphasized that not all energy assets carry the same risk profile. Lower-risk infrastructure such as solar installations, for example, presents fundamentally different safety and environmental considerations than higher-risk assets like production wells, where failures can have significant operational and environmental consequences. Accordingly, risk-specific guardrails must be built into AI strategies from the outset. When firms take a piecemeal approach, they can create additional vulnerabilities such as accelerating technical debt and introducing new security vulnerabilities in their software code bases. An integrated, risk-aware approach, Parker argued, helps mitigate these downstream challenges while enhancing long-term value creation.

Transformational vs. Innovative Leadership

Two leadership styles surfaced as essential. Transformational leadership provides the top‑down vision and resources needed to overhaul legacy processes. Innovative leadership, on the other hand, empowers middle managers and frontline teams to experiment, fail, and iterate. Schapiro emphasized that both styles must coexist: executives set ambitious AI targets, while managers create the "safe‑to‑fail" environments where those targets become reality.

Humility and the "Human‑in‑the‑Loop"

Pickle highlighted a subtle but critical point: leaders must model humility. When senior leaders openly adopt AI tools, admit missteps, and actively seek feedback, they normalize a culture where AI is a teammate, not a replacement. This cultural shift facilitates broad AI adoption.

From Raw Data to Orchestrated Intelligence

Rabley broke down AI readiness into a five‑layer architecture:

1. Data Collection: Capturing structured and unstructured streams.
2. Data Organization: Turning raw feeds into a coherent, searchable repository.
   1. Parker noted this is a common hurdle for incumbents.
3. Ontology Layer: Defining the meaning of data elements so AI can interpret them correctly.
4. Semantic Layer: Mapping business questions to the right data sets and models.
5. Orchestration:  The top‑level layer where LLMs, reinforcement‑learning agents, and digital twins collaborate to make autonomous recommendations.

Only a handful of firms have successfully baked all five layers together, which explains why less than 10 % of companies can point to a clear AI‑generated ROI today.

Continuous, Curious Learning as a Competitive Edge

Both Parker and Pickle framed the HR challenge as a "nimble‑learner" problem. The answer is not a one‑off training program but a continuous learning loop where employees experiment with AI tools, share successes (and failures), and collectively raise the organization's AI literacy. Williams Companies, for example, measured a $7 million bottom‑line uplift from a pilot team that adopted AI‑augmented forecasting, demonstrating that modest, well‑tracked experiments can drive real financial impact.

Universal AI Fundamentals

Both Avice-Huet and Schapiro stressed the need for baseline AI competence across the entire workforce. Schneider's rollout of an AI curriculum illustrates how scale can be achieved when training is framed as a universal competence rather than a niche skill. The curriculum helps employees speak the same language.

Above the Line vs. Below the Line

Parker introduced a useful mental model: "above‑the‑line" roles have agency to shape outcomes and direct AI (e.g., process engineers using AI to redesign a workflow), while "below‑the‑line" roles are largely execution‑focused (e.g., a ride-sharing driver following AI‑generated schedules). The goal for employees is to upskill to stay above the line.

Collaborative Advantage

The panelists agreed that creating a shared learning ecosystem is essential for firms because AI breakthroughs are too complex, costly, and data‑intensive for any single company to achieve in isolation. Firms should pinpoint the narrow slice of the value chain where they hold a genuine competitive edge and join industry‑wide networks that pool data, standards, and best‑practices. In this collaborative, orchestrated ecosystem, firms contribute the capabilities they excel at while relying on the network for the remaining functions. By doing so, they turn isolated pilots into scalable solutions, accelerate the collective learning curve, and ultimately lift all boats.

From Celebration to Business Impact

Pickle made it clear that AI success must be tied to business outcomes, not just technology milestones. Rather than gamifying adoption through points or badges, Williams tracks AI initiatives against concrete KPIs such as cost savings, emissions reductions, safety incident rates.

Change is the Only Constant

Schapiro cautioned against the "perfection‑is‑the‑enemy" trap. While building a robust data foundation is essential, companies must move forward iteratively—launching AI solutions, learning from real‑world data, and expanding capabilities in rapid cycles. This approach builds "muscle memory" for the organization, enabling it to keep pace with the accelerating AI landscape.

For the next generation of energy leaders—whether you're graduating this spring, guiding a team, or launching a startup—understanding the layered architecture of AI, championing a culture of curiosity, and building cross‑sector partnerships will be the cornerstone of lasting impact.

dpine-250x250.jpg

Dartmouth at CERAWeek 2026

CERAWeek is regarded as one of the most influential annual conferences in the energy sector, drawing more than 11,000 energy leaders, researchers, technologists, and entrepreneurs from around the globe.

At the 2026 conference, Dartmouth faculty members appeared in panels spanning organizational transformation, industrial data infrastructure, grid modernization, electricity market design, and wildfire risk management. 

Explore Dartmouth's engagement at CERAWeek 2026.