Artificial intelligence in energy is not only about chatbots, copilots, or office productivity. At Woodside Energy, the most important AI work is tied to physical assets, operational data, safety, reliability, and the people responsible for running complex industrial systems.
Andrew Melouney, vice president for digital at Woodside Energy, describes an AI strategy built over years, not rushed into place after the arrival of generative AI. The company’s current direction points to a broader future for industrial AI: systems that rely on strong data foundations, fit into real workflows, and help experts make better decisions under demanding conditions.
Industrial AI starts with the nature of the work
Woodside Energy operates in an environment where digital tools are connected to large physical systems. The company works across exploration, drilling, subsurface work, project development, operating assets, and global energy portfolio marketing and trading.
That matters because the energy sector’s AI journey has not followed the same path as consumer technology. The work is asset intensive, safety critical, and highly physical. Many operations also take place in harsh and remote locations, where continuity and reliability are central concerns.
For Woodside, AI adoption began with the data already being produced by equipment, plants, and assets. Melouney says the company has had “very large volumes of operational data” from the systems it runs. Those data streams created clear use cases around reliability, safety, and efficiency.
The company has used traditional AI techniques since around 2015, including analytics, optimization, and predictive models. Generative AI is newer, but it is being added on top of an existing base rather than treated as the starting point.
Trusted data is the operating layer
The source of Woodside’s AI work is its operational data. Melouney describes data as foundational and as an asset. In facilities with sensors everywhere and decisions happening in real time, the quality, structure, and governance of that data become essential.
Woodside has invested in an enterprise scale data platform designed to be secure, structured, and governed. The aim is to make data usable in data science applications and AI agents with enough trust that it can support responsible decisions.
The company has also built platforms that continuously ingest high frequency data from assets and enterprise systems. That makes it possible to connect data sets that would otherwise remain separate.
Maintenance intelligence is one example. Woodside uses it to analyze historical maintenance records alongside equipment performance. Because those data sets are governed and available in one place, the company can correlate maintenance records from SAP with equipment and performance data from its time series data lake.
The practical goal is simple: do the right work at the right time. On one piloted asset, Woodside sees an opportunity to reduce maintenance hours by up to 15% over five years. The same foundation now allows the company to put agentic AI over the analytical model to provide better insights and improve the solution further.
AI supports operators rather than replacing them
Woodside’s strategy is built around decision support. The company is not presenting AI as a substitute for operational expertise. Instead, it is designing systems that help workers use their judgment and experience with better information.
Melouney frames the goal as empowering people to make better and faster decisions. That is especially important in high-stakes settings, where operators remain accountable for the work being done.
The clearest example is Woodside’s Startup Advisor, an agentic AI solution that supports operators starting up LNG plants. These facilities are highly technical and require specialist skills to start up reliably, consistently, and safely.
The Startup Advisor works like a copilot beside operators who sit in front of a panel and manage the startup process. Its role is to optimize, support, and empower rather than take accountability away from the people running the plant.
This approach also changes how AI is evaluated. In an industrial setting, a useful AI tool is not merely one that can generate a response. It must fit into a workflow, use trusted data, and support the decisions that matter to operations.
From isolated tools to enterprise capability
Woodside’s AI program has moved from individual solutions toward a more coordinated enterprise-wide capability. Melouney describes the company’s innovation philosophy as: “think big, prototype small, and scale fast.”
That means identifying large opportunities, testing them on a smaller part of an asset or subsystem, learning from the deployment, and then scaling what works. Maintenance intelligence and Startup Advisor both reflect this pattern.
Woodside also learned from a broader early approach to generative AI. Around 18 months, two years into that journey, the company had deployed AI broadly for personal productivity and allowed the organization to become familiar with the technology. That helped build understanding and trust.
Over time, the company shifted toward a tighter focus on higher value priorities. The challenge became deciding where to invest central time and resources, while still enabling teams across the organization to solve problems in their own domains.
That shift reflects a common issue for industrial AI. Experiments can prove what is possible, but scaling requires repeatable patterns, governed platforms, clear ownership, and alignment between people, process, and technology.
The next step is agentic AI in core workflows
Woodside’s direction is toward more autonomous and connected AI systems, but within a framework of governance and human accountability. Melouney says the ambition is “an autonomous enterprise,” with agents that can interact deeply with core workflows.
That ambition depends on the groundwork already described: data quality, secure platforms, governed data assets, trust between digital teams and the organization, and workers trained in agile ways of working, design thinking, and problem solving.
The lesson from Woodside is that industrial AI is not a quick layer added to existing operations. The company is rethinking work itself, asking where AI belongs in the process and how it can create better outcomes.
For energy companies, the future of AI may be less visible than consumer tools but more consequential inside the systems that keep assets running. The companies best positioned to use agentic AI are likely to be those that have already done the quieter work: collecting trusted data, governing it well, and designing technology around the experts who run the operation.