Having an AI strategy is one thing. Turning it into measurable business value is another.
Many organisations have moved beyond experimenting with AI, but scaling it across the business remains difficult. The challenge is rarely technology alone. Real progress depends on whether strategy, data, technology, skills, governance and culture are ready to move together.
Based on our work with clients and Cegal’s own AI transformation, we have identified six dimensions that determine whether an AI strategy stays on paper or delivers value in practice.
1. Strategy & Leadership
Execution starts at the top. Without a named executive owner, a dedicated budget and a funded roadmap, AI tends to remain a collection of experiments. Mature organisations integrate AI into the corporate plan and tie it to business outcomes, whether that is improving production uptime, accelerating subsurface interpretation or reducing emissions. Clear sponsorship also tends to unlock progress in every other dimension.
2. Data Foundation
Energy companies are rich in data, from seismic and well logs to sensor streams and production records. Too often it sits in siloed systems, in inconsistent formats and behind manual access requests. AI is only as good as the data beneath it. The goal is to treat data as a product: governed, high quality and readily accessible. A simple test is whether a priority dataset can be obtained in analysis-ready form without asking another team. A word of warning on this dimension; it’s essential the right controls are in place, e.g. if an agent or LLM can access data it should not have access to, that creates a different problem altogether (see dimension 5 below)
3. Technology & Tooling
Early AI work often runs on isolated tools, such as a single notebook used by a single team. Scaling requires a common platform to build, deploy and monitor models, with reusable components that speed up delivery. For energy organisations, that platform must work across cloud, edge and on-premises environments, and keep pace with rapidly evolving foundation models. Standardising on one platform, rather than letting each team build its own, is often the step that moves an organisation up a level.
4. Talent & Skills
Successful execution needs people on two levels. The first is dedicated specialists such as ML engineers, data scientists and AI product managers, often starting in a central Centre of Excellence. The second is broader AI fluency across the workforce. In energy, the real value comes from pairing technical skills with deep domain expertise, so that geoscientists, engineers and operators help shape the solutions they will use. This is typically achieved through a mix of hiring and upskilling.
5. Governance & Responsible AI
In safety-critical and heavily regulated industries, trust is non-negotiable. Governance should begin with foundational policies on acceptable use and data privacy, then mature into a risk-tiered review process in which higher-risk use cases pass a documented sign-off before launch. Done well, governance does not slow innovation down. It gives the business the confidence to move faster, because everyone knows the guardrails.
6. Culture & Adoption
A technically brilliant solution that nobody uses delivers no value. Culture is often the most underestimated dimension. Organisations build momentum by communicating early wins, embedding AI tools into daily workflows and helping employees see AI as an accelerant to their work rather than a threat to it. One practical measure is the share of employees who have used an AI tool in their work in the last 30 days.
It is tempting to treat these six dimensions as a checklist. In practice, AI strategies rarely fail because one box was left unticked. They stall at the seams between dimensions.
Consider a predictive maintenance model that performs brilliantly in testing. If the sensor data feeding it is fragmented, it cannot be trusted at scale. If the operators who must act on its alerts were never involved in building it, they will quietly ignore it. If nobody has defined who is accountable when it gets a call wrong, it will never be approved for the field. The technology worked. The organisation was not ready for it.
This points to an uncomfortable truth. Progress is set by your weakest dimension, not your strongest, and most organisations naturally invest in what is visible, such as tools and pilots, over what is hard, such as data discipline, governance and culture change. The leaders who make AI stick are usually the ones willing to be honest about where they are really standing, and to fund the unglamorous work first. A single successful pilot is not the same as an AI-ready organisation.
Executing an AI strategy is not a single project. It is a coordinated effort across six dimensions: strategy and leadership, data, technology, talent, governance and culture. Weakness in any one of them can hold back the rest, however strong the others may be.
For energy companies in particular, where the stakes and the complexity are high, the question is less “what can AI do for us?” than “what must be true in our organisation for it to deliver?”
Chris Cartwright, Managing Director UK and Africa, Cegal
At Cegal, we have asked ourselves that question in our own transformation and alongside our clients.