Unilever plans to build more than 40 AI-enabled digital twins across its manufacturing network over the next 18 months. The rollout expands an existing factory technology program after the company reported double-digit waste reductions at two sites, a 30% defect reduction at one site and capacity gains at another.
“Scaling AI across our operations isn’t just a technological shift, it’s a commitment to superior products, sustainability and empowering our teams across our factories,” said Adam Raeburn-James, global vice president for digital business operations at Unilever.
“Through our partnership with Accenture to accelerate digital twins, we are turning innovation into measurable impact to create desirable brands for our 3.7 billion consumers worldwide.”
The consumer goods group announced the multi-year rollout with Accenture on June 16, describing the work as a move from existing factory deployments to a “scalable blueprint” for global adoption. The expansion covers digital models of factory equipment and production lines, fed by live shop-floor data to monitor and predict how machines and processes perform.
Moving from prediction to automated adjustments
A core manufacturing use case for digital twins is prediction. NIST describes a digital twin as a computer model of a physical system that can support simulation, monitoring, optimization and decision support. Unilever’s version applies the model to production variables such as process flow, dosing, viscosity, defects and energy settings.
Unilever and Accenture are also positioning the twins as decision-support systems, not only monitoring tools. Unilever and Accenture said the program combines digital twins with AI-enabled insights and agentic capabilities, allowing factory teams to identify issues earlier, simulate scenarios faster and make decisions across the production cycle.
Accenture said its industrial AI work uses advanced analytics and AI agents to predict maintenance needs, improve performance and help teams act faster.
The announcement also points beyond dashboards and alerts. As the system learns and employees gain confidence in its accuracy, Accenture said it can “progressively take on certain adjustments automatically, with human oversight.” The wording keeps humans in the loop, but it also moves the twin closer to controlled factory intervention.
Early factory results drive the business case
Unilever’s business case is built on existing sites. In Raeford, North Carolina, a digital twin supporting deodorant stick production for brands including Dove, Degree and Axe predicts 95% of process-flow restrictions, with Unilever citing a 20% waste reduction and a 10% capacity increase.
In Poznan, Poland, a twin stabilizing mayonnaise viscosity for Knorr and Hellmann’s products has reduced minor stoppages by up to 20% and cut waste by nearly 30%.
Unilever also cited an intelligent mixer in Cu Chi, Vietnam, where an AI digital twin optimizes raw-material dosing for liquid home care products such as OMO detergent, delivering 1% to 2% savings in premium ingredients.
At Gandhidham, India, a digital twin helped reduce Dove soap quality defects by 30% over four years, measured in distribution centers before delivery to customers. At Haldia, India, an energy twin optimizes fan speeds, temperature setpoints and moisture controls for powder detergents, though the announcement did not quantify the thermal-energy reduction.
The figures remain company-reported results. The June 16 announcement did not disclose baseline production volumes, measurement windows for each percentage, total program cost or the sites selected for the planned builds.
Managing the risks of AI in operational technology
The scale-up also puts manufacturing AI inside an operational technology risk debate. In December, the NSA, CISA and international partners warned that adopting AI into OT systems can introduce risks to the safety and security of the environments and critical functions those systems support.
The guidance urges operators to establish AI governance and assurance, test and monitor deployments, keep humans in critical decisions and implement fail-safe mechanisms.
Those controls are relevant because Unilever’s rollout describes a path from prediction to adjustment. NIST says digital twins can monitor status, detect anomalies, predict system behavior and prescribe future operations in manufacturing.
If recommendations become automated adjustments, factory operators need evidence that the model remains accurate, the control boundary is clear, the rollback path works and the change is visible in operational logs.
Integrating into a wider supply chain strategy
Unilever positioned the rollout as part of a wider supply-chain technology program rather than a stand-alone AI experiment. The company said in April that World Economic Forum Global Lighthouse Network recognitions brought its total to 11 designations across 10 sites, the most held by any FMCG business.
The Forum’s 2026 Lighthouse analysis drew on more than 220 industrial sites and examined manufacturers embedding AI across production and value chains.
Nicole van Det, CEO of Accenture Netherlands and Nordics and global account lead for Unilever, said the expansion pairs advanced tools with “smart process design and disciplined execution on the shop floor.”