Why AI matters at this scale

DRT Metal Packaging is a established, mid-market manufacturer of metal cans and containers, operating with a workforce of 501-1000 employees. For a company of this size and vintage (founded 1949), competing requires maximizing the efficiency and intelligence of existing assets rather than competing on pure scale. AI presents a critical lever to do just that—transforming data from production floors and supply chains into actionable insights that reduce costs, improve quality, and enhance agility. In the capital-intensive, thin-margin world of metal packaging, even single-digit percentage gains in operational efficiency translate directly to significant bottom-line impact and competitive advantage.

Concrete AI Opportunities with ROI Framing

1. Predictive Maintenance for Legacy Equipment: Many of DRT's production lines, including stamping and forming presses, are critical and costly to repair. Unplanned downtime halts revenue. An AI model trained on vibration, temperature, and power draw data can predict component failures weeks in advance. The ROI is clear: a 20-30% reduction in unplanned downtime can save hundreds of thousands annually in lost production and emergency repairs, paying for the sensor and software investment within a year.

2. Computer Vision for Quality Assurance: Manual inspection of millions of cans is slow, inconsistent, and prone to human fatigue. A computer vision system deployed at key points on the line can inspect every can at high speed for defects like dents, seam flaws, or coating inconsistencies. This directly reduces scrap rates, improves customer satisfaction by catching defects earlier, and frees skilled workers for more value-added tasks. The ROI manifests in lower material waste and reduced liability from shipping defective products.

3. AI-Optimized Production Scheduling: Fluctuating demand for different can sizes and materials (steel vs. aluminum) makes production planning complex. Machine learning algorithms can analyze historical order patterns, raw material supply chain data, and even customer forecasts to generate optimized production schedules. This minimizes changeover times, reduces excess inventory of finished goods, and ensures raw materials are purchased at optimal prices. The ROI is seen in lower carrying costs, reduced waste from obsolete stock, and improved on-time delivery rates.

Deployment Risks Specific to This Size Band

For a company in the 501-1000 employee band, the risks are less about technological feasibility and more about organizational readiness. First, the skills gap is pronounced. There is likely no in-house data science team, requiring either strategic hiring or reliance on vendor-managed solutions and consultants, which can create dependency. Second, data infrastructure is often fragmented. Legacy machines may not be networked, and critical data might live in isolated systems (e.g., ERP, MES, spreadsheets), requiring upfront investment in data integration before AI models can be built. Finally, change management is paramount. Success depends on frontline supervisors and machine operators trusting and acting on AI-driven insights. A top-down mandate without involving these key users in the process will lead to rejection. A phased, pilot-based approach that demonstrates quick wins on a single line is essential to build organizational buy-in for broader rollout.