The Edge Malaysia (25-/05/2026) - The recent and inaugural International Palm Oil Millers Conference (IPOMC) 2026, organised by the Incorporated Society of Planters, carried a message too clear to miss: palm oil milling — and indeed the wider plantation value chain — can no longer afford to stand beside the mill, wiping its brow, waiting for the next compliance circular to arrive. 

This is no small corner of industry. Malaysia alone has about 446 palm oil mills, while Indonesia has well over 1,000 mills, with its milling footprint still expanding alongside the scale and complexity of its plantation sector.

The conversation is shifting. Not always loudly, not always evenly, but unmistakably. It was moving beyond mere processing and paperwork towards value-driven modernisation, environmental, social and governance (ESG) discipline, regulatory readiness, business resilience, talent renewal and technological innovation. In short, the mill is no longer just where fruit becomes oil. It is increasingly where efficiency, credibility, sustainability and future competitiveness are won — or quietly lost. 

Amid rising costs, labour pressure and sharper scrutiny from buyers and regulators, the old idea of plantation and milling excellence is being steadily reinvented. It is no longer enough for an estate merely to harvest fruit well, or for a mill merely to process it efficiently. The industry must now measure better, prove more, waste less, respond faster, trace further, train differently and attract the next generation into work that still runs on palms, people, steel, steam and increasingly, data.

When everything wants to be smart 

It must be said, with a small smile, that “smart” has become one of the most hardworking words in modern industry. We now have smart cities, smart homes, smart farms, smart dashboards, smart sensors, smart phones and, naturally, smart mills. The word is used so freely that even a kettle with a blinking light may begin to feel overqualified. 

In palm oil, “smart” usually refers to the use of digitalisation, automation, IoT, data analytics and AI to improve monitoring, control, traceability and decision-making. IoT — the Internet of Things — is especially important because it allows sensors, cameras, machines and control systems to be connected, sending real-time information from the estate or mill floor to dashboards and decision-makers. 

That is fair enough, provided we remember that smartness is not in the label but in the outcome. A smart estate must improve field discipline. A smart mill must reduce losses, improve efficiency and quicken intervention. Otherwise, “smart” risks becoming another shiny adjective in a brochure — clever in spelling, but not yet wise in practice. 

The old mill gets a new brain 

One particularly interesting development is the emergence of the AI-driven palm oil mill, now with about two years of operating experience for review. The early signs appear encouraging. 

This is not yet the final answer to every milling challenge, and it should not be treated as a magic button with artificial intelligence (AI) sprinkled on top. But as a working example of where palm oil may be heading — blending engineering, digitalisation, IoT, AI and human judgement — it is looking good enough for the industry to pay serious attention. 

There is something wonderfully Malaysian about the world’s first AI palm oil mill being launched not in Silicon Valley, Shenzhen or some glass-and-steel laboratory, but in Kuala Kangsar, Perak. 

A palm oil mill that had been operating for decades was not retired, demolished or sent quietly into industrial memory. Instead, it was taught a new language: sensors, cameras, algorithms, predictive tools, dashboards and data.

In short, the old mill was given a digital ear, a sharper eye and perhaps, if we are brave enough to say it, a small electronic brain. 

For decades, palm oil mills have been the hardworking middle child of the supply chain. The estates get romance: sunrise over palms, planters in boots, loose fruits on the ground, the theatre of harvesting. The refineries get sophistication: fractions, formulations, oleochemicals, consumer products and export markets. The mills sit in between, noisy, steamy, oily and essential — the place where biology meets engineering and where fresh fruit bunches must be converted quickly, efficiently and cleanly into crude palm oil and kernels. 

It is not glamorous work. But without the mills, the plantations are only fields with a delivery problem.

At the conference, technology proponent AIREI — which I later discovered stands, rather grandly, for Artificial Intelligence Robotics Engineering Industries — described an AI-based smart palm oil mill as one that uses AI to optimise production, reduce labour dependency, cut inefficiencies and improve process control. 

I recall being exposed to SCADA (Supervisory Control and Data Acquisition) decades ago in an aeroponic farm and later in a biomass plant. At that time, SCADA was already a useful step forward — helping us monitor equipment, track readings and control processes without running around with clipboards like overworked scouts. 

But SCADA was largely about seeing, monitoring and responding to preset parameters. AI-driven systems go further. They can analyse patterns, detect anomalies, predict faults and recommend adjustments in real time. Put simply, SCADA helped the system become more visible. AI now promises to make it more predictive — though common sense must still remain very much human. 

In practical terms, this means a mill fitted with advanced sensors that monitor temperature, pressure, amperage and machine performance in real time; AI-enabled CCTV cameras that track fresh fruit bunch (FFB) volume, quality and anomalies; and AI-driven control systems that regulate equipment and resource use based on live data. In short, the mill is no longer merely listening to the groan of machinery and the instinct of seasoned operators. It is beginning to see, sense, learn and respond — hopefully with fewer coffee breaks than the rest of us. 

The system covers the FFB ramp and weighbridge, steriliser, digester and press, clarification, kernel plant, boiler, crude palm oil (CPO) storage tank and data room. Features include online FFB analysis, automated steriliser recipe selection based on FFB quality, real-time prediction of mesocarp fibre oil loss, online VCT underflow oil-loss monitoring, auto boiler fuel control, instant oil extraction rate (OER) calculation using flowmeters and integration between operations, maintenance, store, procurement finance, human resource and weighbridge data.

If we are now familiar with ChatGPT, then going forward, AIREI is pursuing MyPalmGPT features to assist production and maintenance teams. One can almost imagine the seasoned mill engineer raising an eyebrow: “GPT also knows about choked conveyors, kah?” 

Perhaps not fully — at least not yet. But give it enough good data, operational discipline and a few real-life bruises from the mill floor, and it may begin to know where to look before everyone performs that old industry ritual of blaming the night shift. 

AI must earn its keep 

AI must earn its keep. In palm oil milling, it is not enough for technology to look impressive on a dashboard or sound fashionable in a conference hall. It must deliver where millers feel the pressure most: labour, losses, throughput, cost, reliability and return on investment. 

The AIREI’s AI-based smart mill was implemented from August 2024, with the achievement section showing the post-implementation performance trend. The strongest message from the graphs is that the mill became more stable, less labour-dependent and more efficient in extraction.

The last few performance slides shared are where the story begins to move from digital promise to commercial substance. After implementation from August 2024, the smart mill appears to show a meaningful reduction in mesocarp oil loss, a stronger and more stable OER, a sharp reduction in manpower and a major decline in breakdown hours. 

It was shared that oil loss trended down. OER reportedly improved from about 17.85% to 19.02%. That is significant, because OER remains one of the biggest revenue levers in milling. Even a 1% gain in OER can translate into substantial additional oil recovery and revenue for a mill. 

The operational gains were equally noteworthy. Total manpower was reduced from 66 to 31 workers, representing a 53% labour reduction. This directly lowers recurring operating costs and reduces dependence on foreign labour. Breakdown hours also fell sharply, from an average of about 12.21 hours to 3.08 hours, suggesting steadier throughput, fewer disruptions and better process discipline. 

Taken together, these are not cosmetic dashboard improvements. They go straight to the heart of mill economics: lower losses, higher extraction, reduced labour reliance, less downtime and more stable operations. That is why the investment case deserves serious attention. 

The technology provider’s claim of a favourable payback within two years. This becomes credible not because AI sounds fashionable, but because the graphs point to improvements that accountants, engineers and millers can all understand without needing artificial persuasion. That, I must admit, deserves serious attention and follow-through.

Nationally, incentives are welcome, including policy measures to encourage AI training and readiness. But even without leaning too heavily on incentives, the reported returns suggest that a well-implemented AI smart mill can offer a favourable return on investment — provided the engineering practices are sound, the data is reliable, and the people using it are properly trained. 

Still, one caution remains. No mill becomes smart by software alone. Before AI can perform its digital magic, the old-fashioned fundamentals must first be in place: sound engineering, disciplined maintenance, reliable instruments, proper calibration, clean steam lines, healthy boilers, sensible layouts, trained operators and management that truly understands what the mill is doing. 

AI cannot rescue poor engineering. It will merely report the chaos faster, perhaps with nicer graphics. A smart mill, therefore, is not a workerless mill. Even with automation, it still needs people — fewer perhaps, but better trained, more alert and more technically capable.

AI can analyse patterns, surface insights and recommend actions in a structured and timely manner. In some cases, certain responses can be executed automatically through the control system. But ultimate accountability must remain with the management team, who must be able to validate, adjust or override recommendations based on operational judgement and real-world conditions. 

Sensors may detect, cameras may watch and algorithms may suggest. But people must still interpret, repair, verify and take responsibility when steel, steam, oil and fruit refuse to behave politely. 

The crop supply test 

There is also another practical point. A mill that sources nearly 90% of its crop from outside suppliers lives with a different kind of uncertainty. It is dependent on third-party fruit, and therefore exposed to the familiar challenges of supply consistency, bunch quality, crop assurance, timing, supplier loyalty and market competition for FFB. 

AI can improve monitoring, control and decision-making inside the mill gate, but it cannot fully command what arrives at the ramp. That is why smart mill technology should hopefully also be tested and implemented in an integrated mill linked to its own estates, where crop supply, harvesting intervals, bunch quality and traceability are more constant and better assured. Given a longer operating period under such conditions, the technology can be evaluated more fairly, refined more deeply and proven more convincingly. 

Good technology needs good data, and good data begins with a more reliable crop stream. There is also every reason to expect that such technology will improve with time. A smart mill is not a one-off installation where everything becomes perfect the moment the ribbon is cut and the minister leaves. It will require fine-tuning, calibration, operator learning, software refinement, maintenance discipline and the slow accumulation of operational data. 

Sensors, cameras, control parts and digital components may also become cheaper and better as adoption widens and suppliers compete.

In that sense, the Kuala Kangsar mill should be seen as an important working platform, not the final textbook answer. Give it time, measure it honestly, improve it continuously and let the results speak. If the numbers hold, reliability improves and practical benefits become clearer, more millers may (and should) be enticed to follow — not because AI sounds fashionable, but because it finally proves useful where it matters most. 

Let someone else go first 

There is, of course, that familiar industry reflex: let someone else try it first. 

Let another mill carry the first risk, suffer the first teething problem, answer the first awkward board question, and find out whether the “smart system” is truly smart — or merely expensive with better lighting. 

There is also the understandable hesitation of ageing mill owners, quietly wondering who will take over, who will sustain the business, and who will have the appetite, patience and competence to manage the next generation of technology. Machines, after all, may be upgraded more easily than succession plans. 

Such caution is understandable. Millers are practical people. They know steel, steam, fruit and workers do not always behave kindly towards fashionable promises. The mill floor has a way of humbling PowerPoint slides. 

Yet someone must still lead the way. 

Early adoption creates the learning ground for everyone else. It allows mistakes to be understood, systems to be improved, people to be trained and confidence to be built. More importantly, it prepares the next generation of millers, engineers, technicians and operators to work naturally with digitalisation, IoT and AI — not as outsiders staring at a dashboard, but as practitioners who understand both the machine and the meaning behind the data. 

The future workforce cannot be trained only on yesterday’s tools. They must learn to combine field sense, mill discipline and data intelligence. For palm oil milling, the real challenge is not whether technology will arrive. It already has. The question is whether we shall meet it early, learn from it wisely, and shape it to serve the industry — or wait until the future arrives at the mill gate, honks loudly, and asks why nobody prepared the security pass.

Joseph Tek Choon Yee is a former president of the Malaysian Estate Owners’ Association and past chief executive of the Malaysian Palm Oil Association.

[To read Part 2, click here]

Read more at https://theedgemalaysia.com/node/804801