What AI already does well in supply chain management

Supply chains perform a series of actions starting with product design and proceeding to procurement, manufacturing, distribution, delivery, and customer service. “At each of these points lie big opportunities for AI and ML,” says Devavrat Bapat, Head of AI/ML data products at Cisco. That’s because the current generation of AI is already very good at two things needed in supply chain management. The first is forecasting, where AI is used to make predictions about downstream demand or upstream shortages. Moreover, algorithms can detect one or more events they recognize as precursory to failure, and then warn assembly line operators before production quality falls short.

The second is inspection, where AI is used to spot problems in manufacturing. It can also be used to certify materials and components, and track them through the entire supply chain.

Ultimately, AI will optimize supply chains to meet specific customer needs for any given situation. The enabling technology exists but the remaining challenge is it requires a level of data sharing that can’t be found in supply chains today. In the meantime, many companies continue to reap the benefits of improved forecasting and inspection.

Forecasting

Take for example, Amcor, the biggest packaging company in the world, with $15 billion in revenue, 41,000 employees, and over 200 plants globally. Most of their market is in food and healthcare packaging.

“We make the packaging for about one third of the products in your fridge,” says Joel Ranchin, the company’s global CIO. Some of the challenges Amcor faces in manufacturing have to do with accurate forecasting and adapting to changing demand. Orders are often modified in the food supply chain space as needs change. In hot weather, for instance, people drink more Gatorade, which can create a sudden explosion in demand, so there could be a 10 to 15% spike in demand for bottles. The same is true for other kinds of products. There could be more fish in the ocean suddenly, which increases the demand for packaging to accommodate additional tons of fish. “Even though we try to forecast, it’s very difficult because we don’t always know our customers’ needs ahead of time,” says Ranchin.  

The challenges are similar on the other side of the supply chain. If Amcor can’t accurately predict shortages, it can’t stock up ahead of time on raw materials. More importantly, the company needs to predict price changes, so it can buy more at lower prices before a hike, or less if it looks like a drop is on the horizon.