No Sector I Encounter Has More Operational Data Than Energy. Few Act on Less.

I do not work in oil and gas. I train commercial and technology teams across sectors, and energy companies keep appearing in those rooms, which has given me a particular view of this one. Here is what I notice. No sector I encounter generates more operational data, and few act on less of it. A single producing facility produces continuous measurement from thousands of points. Pressure, temperature, vibration, flow, power draw. Most of it is recorded faithfully, stored properly, and consulted only after something has already gone wrong. The historian becomes an archive rather than an instrument. That gap is not a technology problem. The technology has existed for years and the sector was an early adopter of the underlying mathematics. It is an organisational one. ## Why adoption stalls here specifically **Safety culture cuts both ways.** A sector where mistakes kill people has built strong instincts about unproven methods, and those instincts are correct. They also mean any new approach carries a burden of proof that a retail business would never impose. That is defensible on the process side. It gets misapplied when it blocks a maintenance scheduling model that touches nobody's safety case. **The operational and commercial sides run separately.** Engineering has the sensor data. Commercial has the market and contract data. In many organisations these are different systems, different reporting lines and different vocabularies. Most of the interesting questions need both. **Capital cycles are long, software cycles are not.** A sector accustomed to evaluating decisions over decades struggles with a tool that will be materially different in eighteen months. ## Where it earns its keep **Predictive maintenance on rotating equipment.** The most established application in the sector globally. Vibration and thermal signatures shift before failure. Catching that converts an unplanned shutdown into a scheduled one, and the difference between those two numbers is large enough that the business case rarely needs arguing. **Pipeline and asset integrity.** In Nigeria this carries weight beyond efficiency. Losses from interference and theft along the pipeline network are a documented and persistent problem. Pattern detection across pressure and flow data is one of the few tools that scales to the length of the network. **Energy optimisation.** Applies as much to the diesel and gas generation running industrial sites as to production itself. Consumption that does not match output is a detectable pattern. **Document and compliance load.** Regulatory submissions, Nigerian content reporting under the NOGICD Act, permit documentation, tender responses. Unglamorous, enormous, and well suited to the tools. ## The part nobody has governed Meanwhile the commercial teams have already adopted. Somebody is pasting tender documents into a free chatbot to summarise them. Somebody in planning is dropping production figures into a tool to build a chart faster. Somebody in legal is running a joint venture agreement through a summariser. I call it BROAI. Bring Your Own AI. In this sector the material can include production data, commercial terms with partners and governments, and reserve information. That is not ordinary corporate confidentiality. The engineering side is governed to a high standard. The commercial side, in most organisations I see, is not governed at all. ## What good looks like One approved tool set covering the whole organisation rather than the operational side only. A written rule about production, commercial and partner data. A named owner per use case carrying a number. Start where safety cases are not involved, prove the method, then earn the right to go closer to operations. SWITCH, the framework I use with clients, opens with Scan and Score because in this sector the honest answer to what data do we actually have is usually more complicated than the systems diagram suggests. A question for the next executive meeting. Your engineering data is governed carefully. Who is governing what your commercial team pastes into a chatbot on a Friday afternoon? ## Frequently asked questions **Where should an energy company start with AI?** Predictive maintenance on rotating equipment, in most cases. It is the best understood application in the sector and the business case is straightforward. **Is AI safe to use around operational technology?** Advisory and scheduling applications sit well away from the safety case. Anything touching control systems belongs inside existing process safety governance, not alongside it. **What is the biggest risk for energy companies adopting AI?** Ungoverned use on the commercial side. Production data, partner terms and reserve information going into free tools while attention stays on the operational estate. **Does AI help with losses along the pipeline network?** Pattern detection across pressure and flow data can flag anomalies faster than periodic inspection across a network of that length. It supports the response rather than replacing it. **How does Nigerian content regulation interact with this?** Reporting and documentation obligations under the NOGICD Act are a substantial administrative load, and that load is well suited to automation. *Stanislaus Martins advises and trains enterprise teams across Sub Saharan Africa on practical AI adoption and governance. Formats are on the [speaking and training page](/speaking), or start a conversation on the [work with me page](/work-with-me).*