Australian energy giant Woodside has deployed approximately 50 AI agents in production environments, supporting both operational assets and enterprise workflows, marking a significant shift from isolated pilot projects to coordinated, enterprise-wide AI capability. The company's approach, articulated in a detailed walkthrough of its AI strategy, offers a rare look at how a heavily regulated industrial operator is scaling artificial intelligence from experimental prototypes to mission-critical deployment.
The driving philosophy behind Woodside's AI expansion is captured in a simple three-part mantra: think big, prototype small, scale fast. This framework has guided the company through a deliberate transition from broad, experimental use-case development to a narrower focus on high-value priorities. The result is a growing portfolio of AI agents that do not merely assist with isolated tasks but actively participate in complex industrial workflows.
From Broad Experimentation to Focused Industrial AI
Woodside's generative AI journey began roughly 18 months to two years ago with a relatively open-ended approach. The company deployed AI broadly for personal productivity, allowing the organization to become familiar with the technology, understand its capabilities, and build trust. This phase was intentionally permissive, designed to let employees across the business experiment with AI tools and learn what worked.
However, the strategy has since pivoted. The company now applies a tighter lens to where it invests time and resources, concentrating on higher-value solutions. The critical insight was that a central AI team could not scale effectively if it remained the sole source of AI innovation. The goal shifted to empowering domain experts across the organization to solve problems with AI in their own areas, using repeatable patterns and a standardized platform rather than building every solution from scratch.
This transition from a broad, exploratory phase to a focused, value-driven approach has been central to Woodside's ability to move AI into production environments where reliability and safety are non-negotiable.
The Startup Advisor: A Copilot for LNG Plant Operators
The flagship example of Woodside's targeted AI strategy is the Startup Advisor, an agentic AI solution designed to support panel operators during the startup of LNG plants. These facilities are among the most technically complex in the industrial world, and starting them up requires highly specialized skills and years of experience.
The Startup Advisor functions as a copilot for the operator. It can replay previous startups, compare them with the current process in real time, and provide actionable insights to optimize the startup sequence. For a junior operator, the system effectively acts as an experienced colleague sitting beside them, offering guidance drawn from historical data and established best practices. The solution was first prototyped on a single subsystem before being scaled across multiple assets, exactly as the “think big, prototype small, scale fast” model prescribes.
This use case is particularly instructive for organisations considering AI in industrial settings. It is not about replacing human expertise but about augmenting it, compressing years of experiential knowledge into an accessible, interactive tool that improves both safety and efficiency.
How Many AI Agents Does Woodside Have in Production?
Woodside currently has around 50 AI agents in production. These agents support both operating assets and enterprise workflows, and they have been proven in live environments. The company has intentionally moved away from point solutions that solve narrow problems in specific areas toward agentic solutions that can work across entire workflows. This shift from isolated tools to coordinated agents is a direct result of standardising on a single platform and developing repeatable deployment patterns.
Standardisation and Platform Thinking at Scale
A key lesson from Woodside's AI journey is that scaling quickly requires not building 50 solutions in 50 different ways. The company has invested in a standardised platform and a set of repeatable patterns that allow technical teams and end users to roll out AI solutions quickly and safely. This platform approach enables consistency in governance, security, and performance monitoring, which is essential when moving from dozens to potentially hundreds or thousands of agents.
The emphasis on platform thinking also addresses a common enterprise challenge: the tension between speed and control. By providing pre-approved building blocks and clear patterns, Woodside empowers teams to deploy AI solutions without requiring deep involvement from a central AI team for every initiative.
Governance as a Scaling Enabler
Governance is often perceived as a bottleneck to innovation, but Woodside frames it as a critical enabler of speed. Operating in a well-regulated environment, the company has implemented a structured assessment process that every AI use case must pass. This process evaluates privacy controls, cyber security, and ethical considerations, asking not only whether the company could deploy a particular AI solution but whether it should</
If an AI solution raises concerns during the structured assessment, it is escalated to an AI council composed of senior leaders from across the organisation. This council oversees prioritisation and risk management, providing a forum for robust debate about the trade-offs involved in each deployment.
Perhaps the most forward-looking aspect of Woodside's governance approach is its focus on lifecycle management. The company recognises that managing 50 agents is relatively straightforward but that 500, 5,000, or 50,000 agents will require entirely new capabilities. Questions about usage rates, model drift, retraining needs, and outcome efficacy will become increasingly complex at scale. This is an area where Woodside acknowledges it is still learning, and it sees partnering with experienced providers as essential to solving these challenges.
Partnering for Scale: The Infosys Relationship
Woodside's partnership with Infosys, its managed service provider, plays a central role in the company's AI scaling strategy. The logic is straightforward: a licence to innovate depends on a licence to operate. For Woodside's AI team to innovate on operating assets and corporate functions, the core platforms, systems, and applications must run reliably, safely, and consistently every day. Infosys manages those core operations in partnership with Woodside's internal teams.
As Woodside moves from pilots to enterprise-wide deployment, the partnership also provides access to talent that is difficult to hire directly. With a strong local team in Western Australia but facing the same AI talent shortages as every other organisation, the ability to draw on Infosys's broader workforce through co-mingled teams brings diversity of thought and experience. Woodside retains ownership of the strategy, governance, and accountability, but it relies on partnerships to execute at scale.
What This Means for the Industrial AI Sector
Woodside's approach offers a pragmatic blueprint for industrial organisations looking to deploy AI at scale. The emphasis on starting broad to build organisational familiarity, then narrowing to high-value priorities, is a pattern that many enterprises could adapt. The investment in platform standardisation and repeatable patterns addresses a common failure point where early pilots never translate into production systems. And the focus on governance as a scaling enabler rather than a compliance burden reflects a mature understanding of what it takes to operate AI responsibly in regulated environments.
The most significant implication for the broader industry is Woodside's explicit goal of moving from dozens to hundreds or thousands of agents. This vision of a future where AI agents are as ubiquitous as software applications in an enterprise environment will require new approaches to lifecycle management, monitoring, and governance. Organisations that invest in these capabilities now will be better positioned to scale their AI initiatives when the next wave of agent-based solutions arrives.
For now, the company's 50 production agents represent a concrete milestone in the journey from AI experimentation to industrial reality. The next phase, scaling that number by an order of magnitude or more, will test whether the foundations Woodside has built are strong enough to support truly enterprise-wide intelligence.
