AI results directly impact maintenance KPIs

The IBM research paper Oil and gas in the AI era shows that AI is gaining ground quickly in the oil and gas sector. Applications range from exploration and drilling to refining, pipeline monitoring, predictive maintenance, and emissions management.

For maintenance, the operational results are particularly promising. Oil and gas companies report 27% better production uptime, thanks in part to AI-driven predictive maintenance on installations. There’s also 26% better asset utilisation, 18% lower operational costs, and 18% higher employee productivity.

These are exactly the KPIs that maintenance organisations actively manage. AI helps by making better use of existing data, adding an extra layer to the knowledge already present in your maintenance history and operational processes.

Fewer data silos means more context

As an asset or maintenance manager in oil and gas, you usually have plenty of relevant data – on maintenance, incidents, emissions, and safety, for example. The value increases when you connect that information more effectively.

For AI, context is crucial. A deviation in a measurement becomes far more meaningful when you can also take into account the maintenance history, outstanding work orders, inspection results, criticality, and compliance obligations. This creates a much more complete picture of the asset and the risks around it.

That’s why the first step in the research paper’s plan is: make use of existing functionality, break down silos, and build an organisation-wide approach. IBM also highlights data quality, integration of legacy systems, metadata, ownership, and access policies as essential conditions for scaling AI.

One clear view of your assets

In oil and gas, you want to see the condition of your assets quickly – not just from a maintenance perspective, but also from integrity, reliability, and safety. Think of inspections, corrosion or wall thickness measurements, open work orders, incidents, and equipment that’s out of service.

When this information is spread across different systems, it takes more time to see the full picture. It becomes harder to determine priorities, assess risks, and decide what action is needed.

For AI, this overview is vital. The better the data on assets, locations, maintenance, inspections, incidents, and work orders are connected, the better AI can help identify patterns and anomalies. Not as a replacement for expert knowledge, but as valuable support for maintenance decisions.

Well integrity shows how concrete AI can be

The research paper gives the example of a European oil and gas organisation that has rolled out AI widely for operational improvement and innovation. The company trained more than 100 employees in AI and data science and established a Centre of Competence.

What’s interesting is the approach: they work with smaller “fireflies” – fast, scalable solutions for simple problems such as automatically extracting data from PDF documents. At the same time, they run larger AI projects, including an application to monitor well integrity and detect leaks more effectively.

More than 80 AI use cases have now been identified, with active work underway on 20 new ones. The well integrity application is already in production. Here, AI sits close to the operation: at the level of data, risks, and maintenance decisions that directly affect safety, reliability, and continuity.

AI connects maintenance with incidents, emissions, and compliance

Maintenance in oil and gas also touches on safety, environmental performance, and accountability. According to the research paper, oil and gas companies report 29% fewer operational incidents, 22% better energy efficiency, and 15% lower CO₂ emissions.

Concrete AI applications for environmental and emissions monitoring include detection of diffuse methane emissions and real-time monitoring of flaring activity using satellite imagery, IoT sensors, and image recognition. Another example is agentic AI for pipeline monitoring: systems that independently flag deviations based on sensor data and satellite images and support follow-up actions.

The real value again lies in the connection to operations. An emission incident, safety report, or deviation gains much more meaning when linked to assets, locations, inspections, and follow-up actions. This enables faster decision-making, better recording of measures, and clear demonstration of what happened.

Start small enough to measure

The research paper advises oil and gas companies to choose one or two AI pilots with clear value. Not a broad programme straight away, but an application where you can assess the impact on maintenance, uptime, or production optimisation.

Such a pilot could, for example, involve predicting equipment failure in drilling installations, or creating a digital twin of a single asset such as a pumping station or compressor unit, based on real-time sensor data. You then focus on measurable outcomes like reduced unplanned downtime, higher uptime, or better production optimisation.

It’s important not to approach such a use case from a purely technical angle. With AI-driven simulations and digital twins, collaboration between subsurface engineers, data scientists, and IT is essential. This clarifies what data is needed, which systems must connect, and how the results are used in day-to-day operations.

Where data silos disappear, AI gains more value

AI needs a framework rooted in the operation: asset, location, maintenance history, inspections, work orders, and follow-up actions. This is where the link with IBM Maximo Oil & Gas comes in. It helps bring maintenance, reliability, asset integrity, safety, and emissions management closer together.

Recent updates in IBM Maximo Oil & Gas align well with this, offering better support for emissions management, compliance, document management, safety, inspections, and mobile incident management with mapping functionality in IBM Maximo Mobile.

The gain doesn’t come from yet more data, but from better connections between what you already know, measure, and record.

Would you like to know what this means for your maintenance organisation and how you can put AI to practical use for maintenance, reliability, asset integrity, and emissions management? Get in touch with Wouter Schouten on +31 (0)6 52 68 37 43 or w.schouten@gemba.nl.