MC4 Smart Factory: AI-Operations and Automated Quality Assurance

Replace manual logbooks and shift reports with automatic, real-time measurement of the company's milling and packing operations. Objectives:

• calculate OEE (availability, performance, quality) automatically every shift.

• track yield and extraction against every active production Job Order.

• inspect every moving bag on the packing lines inline by computer vision and reject defective bags without stopping the line.

• alert the responsible teams in real time to prolonged stoppages, abnormal extraction rates and quality failures.

• link operational data to the corporate ERP environment so production performance is measured, attributed and reported without human transcription.

2026

Measured on the live platform (81.56% availability, 77.91% performance, 100% quality), zero scrap with 880 units diverted to rework equal to 0.16% of output, 98.92% yield with a 75.28% extraction rate and milling loss of 1.08% against the 1.5% usual in commercial milling, and computer vision inspection live on five packing lines rejecting defective bags inline at production speed. Every loss is now attributed by cause; including 130 unclassified stops, had never been visible before. Nine workstations are monitored on one live screen, replacing manual logbooks, and the health and safety extension of the same platform is implemented and running. The OEE figure is the operation's first continuously measured baseline, so no before and after comparison is claimed.

Smart factory platform (Meta Smart Factory, v1.389.4): automatic OEE engine, order-based yield tracking, analytics database, real-time alerting and notification module, floor layout live monitoring. Inline vision sensors and cameras (five live feeds) on five packing lines, with mechanical bag pusher, holder and line stop for automatic rejection of defective bags. Integration with the corporate ERP environment (SAP S/4HANA).

The project does not use solutions from the Global CIO catalog

Nine monitored workstations (milling lines A, B and C, the feed mill and the packing lines), each with its own controls and data characteristics, were connected into one measurement platform. Vision inspection had to work at production speed: analysing a continuous stream on moving bags and triggering mechanical rejection (bag pusher, holder, line stop) without halting the line. Live material balancing had to be linked to active Job Order IDs in the corporate ERP with component-level accuracy. Operator discipline changed as well: order and target entry before production and real-time stop-reason tagging are now mandatory. The result is measured performance data with zero manual transcription.

MC4 Smart Factory is an operations intelligence programme built on one principle: every machine, every kilogram and every bag in the company's milling and packing operations should be measured automatically, in real time, without human transcription. It covers the milling and packing lines and links operational data to the corporate ERP environment (SAP S/4HANA).

The capability set went live in the first quarter of 2026, and the operational teams have been working from the platform since 1 July 2026.

Automatic OEE: the platform captures availability, performance and quality per shift across the mills, the feed mill and the packing lines, replacing manual logbooks as the source of shift performance.

Every loss is now attributed rather than estimated: availability loss was 18.44% of planned production time, and the largest categories were 130 unclassified stops, process stops and cleaning. These losses had never been visible by cause before.Order-based yield tracking maps live material balancing against active Job Order IDs, measuring produced kilograms against target with component-level accuracy: 98.92% yield with a 75.28% extraction rate, inside the 72-76% range normal for efficient mills, and milling loss of 1.08% against the 1.5% usual in commercial milling.

Inline quality inspection by computer vision is implemented and running in production: five packing lines carry live camera inspection with five live camera feeds, inspecting every moving bag for packaging anomalies, printing errors, missing or empty date codes and structural defects. A defect triggers an inline rejection that isolates the bag through a bag pusher and holder with line stop, while the reason is logged in the analytics database: inspection at production speed, not sampling after the fact. Real-time alerts on prolonged stoppages, abnormal extraction rates and quality failures reach the responsible teams instead of waiting for shift-end reports. The programme was designed as a platform, not a point solution: the same platform has since extended computer vision into shop floor health and safety monitoring, detecting missing PPE and restricted-zone entry, and that extension is implemented and running. Warehouse IoT environmental monitoring and automated ERP stock updates are the next phases, with the same architecture planned for the company's other milling sites.

Saudi Arabia.

The platform covers the company's milling and packing operations; the same architecture is planned to extend to the company's other milling sites.