Petronas develops AI tools to drive down costs, improve operational efficiency

Malaysia’s state energy firm Petronas is building in-house artificial intelligence (AI) capabilities to drive further improvements in its upstream operations, according to Wan Zul-atfi Mat Nawi, senior manager for upstream technology and innovation.

Petronas has long prioritised cost reduction and operational efficiency, Mat Nawi explained at an investment forum in Kuala Lumpur on April 23. “We have been successful, but now we have almost run out of opportunities to make further improvements. This is where AI comes in,” he said at an investment forum in Kuala Lumpur on April 23.

The process of leveraging AI involves five key stages, he explained. First, data must be explored — identifying and removing irrelevant information while recognising that complete data capture is unrealistic. “There are always gaps,” Mat Nawi noted. The next steps are to build an AI model, establish a data pipeline, scale the model, and finally, operate it in a sustainable manner.

He emphasised that no AI model is flawless and continuous refinement is essential. AI, he said, is not intended to replace humans but to enhance their capabilities and address complex challenges.

Petronas is opting to develop its AI solutions internally rather than relying on third-party providers, which Mat Nawi said is currently the more efficient approach.

Petronas, Malaysia’s largest natural gas producer and one of the world’s top LNG exporters, launched its Artificial Intelligence Centre of Excellence in 2022 in collaboration with international partners. The initiative aims to accelerate the development of AI-driven solutions that enhance efficiency across energy sector operations. Since then, Petronas has signed cooperation agreements with several global technology firms, including US-based Microsoft and Databricks, as well as Abu Dhabi-based AIQ.

The company has developed a range of AI-powered tools to support operational optimisation. Among them is STELLAR, a model that captures tacit knowledge from experienced operators and engineers, offering actionable insights to optimise plant operations in real time. Another is ARIES, a suite of machine learning tools designed to improve plant data performance. The Engineering Wizard platform unlocks domain-specific data and expertise across engineering disciplines, transforming it into digestible knowledge that enhances engineers’ productivity.

Autopilot System, an automated operations tool, streamlines production by eliminating routine manual adjustments and improving plant ramp-up efficiency. Meanwhile, PIVOT DA provides real-time visibility into operational performance, enabling more informed business decisions and cutting report generation time from weeks to just hours.

Mat Nawi was speaking at the International Forum to Attract Foreign Investments in Turkmenistan's Economy (TEIF 2025).