ACROSS Malaysia’s mid- to large-scale manufacturing facilities, the imprint of Industry 4.0 is increasingly visible on the factory floor. From sensors on production lines to AI cameras at entry points to dashboards glowing in control rooms.
On paper, the transformation is well under way and aligned with national initiatives such as the New Industrial Master Plan 2030 (NIMP 2030), the Malaysia Digital Economy Blueprint, and DoSH ((Department of Occupational Safety and Health)’s Vision Zero.
However, discussions with plant managers point to a more complex reality. Data related to near-miss incidents often remains fragmented—camera footage is stored in one system, maintenance records in another, while safety reports continue to be compiled manually at the end of the week.
The visibility exists, but it isn’t connected.
At nearly 4% of global GDP, the cost of workplace incidents, as estimated by the International Labour Organisation (ILO), points to a structural issue, one that raises the question: Is integrated safety the missing piece in Malaysia’s Industry 4.0 ambitions?
A digitised factory floor is not the same as an intelligent one
Malaysia’s workplace safety record reflects a system under pressure. According to a report by Bernama, Malaysia recorded 4,409 workplace accidents between January and May 2025, with the manufacturing sector being the highest contributor. Mohd Hatta Zakaria, director-general of DoSH, has highlighted the disproportionately high incidence of workplace accidents originating from industrial corridors in Johor, Perak, and Penang.
This indicates a pattern that points not to a lack of safety awareness, but to a structural gap in how safety is being monitored and managed.
The irony is that many of these facilities have already made significant digital investments. The problem here is not a shortage of data, but a shortage of connected workplace safety data. In one example cited by viAct, a large construction operator continued to face recurring safety challenges. Near-miss incidents involving heavy machinery, risks around open edges and confined spaces, and PPE non-compliance were not consistently identified through manual audit processes.
At the same time, digital tools in use operated independently: a computer vision system monitored PPE compliance, while incident reporting was managed through a separate platform. Following the deployment of an integrated AI safety system, the same site reported a 10x improvement in safety scores, alongside stronger alignment with regulatory compliance requirements.
Each of these systems performed its intended function, but operated in isolation, without a unified view of risk.
As a result, safety and productivity continued to be managed as parallel priorities handled by different teams, across disconnected platforms, even within facilities that had already considered themselves digitally mature.
Why Vision Zero in Malaysia actually needs Integrated AI Safety Systems
DoSH’s Vision Zero initiative is Malaysia’s national commitment to eliminate workplace fatalities and permanent disabilities, and is frequently cited as a policy target. But operationally, it sets a far higher bar than most safety systems on the ground are built to meet.
What Vision Zero demands is a live operational awareness of risk, which an integrated AI safety platform like viAct brings in 2026. These systems do not simply detect violations; they identify patterns across thousands of micro-events like near misses, repeated unsafe behaviours, and shifts in operational conditions and translate them into early warnings.
For EHS leaders, this represents a shift from managing compliance to managing risk dynamically.
The broader implication for Industry 4.0 is increasingly clear. Digital investments in automation, IoT, and video analytics have improved efficiency and visibility across Malaysian industries. However, without integration, these systems operate in silos, limiting their ability to support real-time safety outcomes.
As Vision Zero targets become more central to national safety strategy, the effectiveness of these investments will depend on whether they are structured to support connected, intelligent safety systems or whether they still remain in a parallel process, disconnected from how operations actually run.
One platform, two mandates – how integrated AI safety systems connect the dots
Most AI platforms in the market are built to answer one question very well. But the right one today must answer two simultaneously: Is this facility safe right now, and is it running as efficiently as it should be?
The architecture that makes this possible is not revolutionary in its individual components—computer vision, edge AI, and scenario-based detection have all been available for several years. What is different, is how those components are assembled and what they are pointed at.
At its core, a unified safety platform deploys AI modules in existing CCTVs across a shop floor’s critical zones like production lines, loading bays, chemical storage areas, or machinery perimeters. It runs continuous scene analysis against a library of pre-trained safety and operational scenarios. It monitors every shift, every hour, flagging deviations in real time and feeding a centralised dashboard that both the safety manager and the operations director can act on.
The practical significance of that shared dashboard is harder to overstate than it first appears. When a safety event and an operational anomaly are visible in the same place, at the same time, correlations that would previously have been buried in separate reports become immediately obvious.
A spike in PPE violations on Line 3 that coincides with a production push on a Friday afternoon is no longer a coincidence waiting to be discovered in a monthly safety audit. It is a live operational signal, surfaced in real time, addressable before the shift ends.
Gary Ng, CEO of viAct, puts the premise plainly.
“The manufacturers who will define Malaysia’s next industrial chapter are not asking whether to invest in safety AI or productivity AI. They are asking why those should ever have been two separate decisions. When you bring safety and operational data onto a single intelligence layer, you stop managing incidents and start preventing them.”
What this AI integration looks like on a real factory floor
Consider a scenario familiar to any plant manager in Malaysia’s food and beverage sector. A worker on a high-speed assembly line begins skipping a hand hygiene checkpoint to maintain pace during a peak production window. In a fragmented system, this surfaces, if it surfaces at all, as a compliance violation in the safety log at the end of the shift.
In viAct’s integrated AI environment, real-time alerts do more than flag violations. They reveal recurring breaches on specific production lines, during specific shifts and under specific supervisors.
That pattern is the intelligence. It tells the operations team that the process design is creating pressure that the safety procedure was not built to absorb, and it gives them the data to redesign the workflow rather than simply discipline the behaviour.
This article is contributed by viAct CEO and co-founder Gary Ng (pix).
