Teaching AI to Optimize Wind Turbine Performance

For some time now, Woodside has operated under a clear innovation philosophy: think big, prototype small, and scale fast. We identify major opportunities, test them on a limited basis, extract the right insights, and then roll them out across the organization. Maintenance intelligence and our Startup Advisor are prime examples. With multiple plants to start up and countless assets requiring maintenance, we set a bold ambition to optimize these processes. The approach begins with a narrow prototype,perhaps a single subsystem or a portion of an asset,and then expands through learning and iteration.

A critical lesson from our AI journey has been the shift from isolated AI solutions toward a more coordinated enterprise-wide capability. About 18 to 24 months ago, during our generative AI exploration, we rarely deployed AI broadly for personal productivity. We were open about the business problems we aimed to solve, which helped the organization become familiar with AI, understand its capabilities, and build trust. However, we soon realized the need to pivot. We moved from a wide-open approach to a tighter focus on where to invest time and resources,targeting higher-value solutions. This allows us to empower the rest of the organization to solve problems with technology in their own domains without always relying on a central team.

The “think big, prototype small, scale fast” mindset has remained crucial. The transition from broad use case development to a narrower focus on high-value priorities has paid off. It enabled us to pursue opportunities like the Startup Advisor, an agentic AI solution designed to optimize and support operators who manage the startup of LNG plants,facilities that are incredibly technical and require specialized skills. This copilot-like tool allows operators to review previous startups, monitor current progress, and gain insights to optimize operations. Starting an LNG facility is immensely complex, and the Startup Advisor makes a junior operator feel as though they have an experienced colleague sitting beside them.

Scaling has naturally led us toward more agent-based solutions. Today, we have about 50 AI agents in production, supporting both operating assets and enterprise workflows. These tools have been proven in live environments, and we have moved from point solutions that solve small, specific problems to AI with agency that works across entire workflows. This success is built on standardized platforms and repeatable patterns. We don’t want 50 solutions built in 50 different ways. Instead, we empower our technical teams and users to roll out solutions quickly and safely using a patternized, platform-based approach.

A final essential lesson is that strong governance is critical for speed, not a hindrance. Traditional methods of governing digital systems won’t scale to the breadth required for AI. A clear philosophy on innovation, the shift from isolated to enterprise-wide capability, and robust platforms with solid patterns and governance are the three pillars that have guided us.

Our partnership with Infosys has been instrumental in accelerating scaling and embedding AI. As our managed service provider, Infosys handles the operations of our core business. Our license to innovate depends on our license to operate. For my team to reimagine how work gets done at an operating asset, we need core platforms running reliably and safely every day. An experienced partner like Infosys, working alongside our internal teams, provides that foundation. As we move from pilots to enterprise-wide deployment, their support allows us to scale. In Perth and Western Australia, we have a strong local team, but like everyone, we struggle to find AI talent. Partnering with Infosys gives us access to different operating models and co-mingled teams that bring diversity of thought and experience. Fundamentally, this partnership lets us operate and innovate with more confidence. Woodside retains ownership of strategy, governance, and accountability, but we cannot achieve our goals without strong partners.

Governance is especially important in our well-regulated environment. Every AI use case goes through a structured assessment covering privacy, cyber controls, and the question of not just “could we do this?” but “should we do this?” We bring together safety, ethics, transparency, and accountability. If concerns arise, the case goes to an AI council of senior leaders who oversee prioritization and risk management. Finally, lifecycle management is key. With 50 agents today,and potentially 500 or 5,000 in the future,we need to track usage, efficacy, model drift, and when retuning is required. This is an area many organizations are still figuring out. Partnering with others, like Infosys, to co-create solutions for this challenge presents a significant opportunity.

Looking ahead, our long-term vision for AI at Woodside is to continue scaling these agentic solutions, amplifying the experience and decision-making of our staff. As we move from 50 to 500 to 5,000 agents, we aim to unlock greater value creation across the sector, transforming how complex industrial operations are managed.