For years, manufacturing companies operated under a mantra of “lean.” Manufacturing relied on reduced supplies in warehouses, based on just-in-time delivery from around the world. Plastics could come from China; chips from Taiwan; steel from Germany.
Then COVID brought about significant changes. The pandemic demonstrated the fragility of “lean,” and measures meant to ensure efficiency became the same impasses that halted production. Shipping containers were stuck at sea. Manufacturing plants for basic goods, such as packaging, were shuttered and caused global shortages of simple items we take for granted — even toilet paper.
In response, today, companies have sought flexibility in their supply chains. They have begun to re-shore and near-shore a variety of manufacturing lines, moving away from a single source to a much more resilient set up.
The construction of new manufacturing facilities in the U.S. is up 116% over the past year, according to Dodge Construction Network. Meanwhile, large manufacturers such as Apple, have moved manufacturing out of China and into India and Southeast Asia.
This massive shift in where manufacturing takes place is coupled with significant investments in new manufacturing sites. This includes massive investments, such as the $11.8 billion from Ford and SK Innovations, and far smaller, but much more pervasive investments from pharmaceutical plants, craft breweries and ice-cream makers. According to the New York Times, “As of August this year, manufacturers had added back about 1.43 million jobs, a net gain of 67,000 workers above pre-pandemic levels.”
Supply chains are already changing, and companies are seeking to change them. Yet as part of those changes, AI can deliver far more resilient and far more powerful supply chains.
For example, one large manufacturer found its suppliers would miss delivery dates and key timelines, which resulted in a scramble as the company approached a delivery milestone for its own clients.
An AI application sifts through tens of thousands of schedule lines and predicts with 85% accuracy if a supplier will miss their commit date. The data science team at the large manufacturing company can then focus on the most important parts, ensuring they’ll arrive on time, and have confidence that the others won’t need intervention.
These AI-enabled tools have fostered much more visibility, and very importantly, they have led to businesspeople — not just those in highly technical jobs — to make very intelligent choices about their inventory. They can work with suppliers who are consistently lagging behind, and plan ahead for disruptions in the supply chain.
This is just the start. There are many applications for large enterprises that would help build a far more resilient supply chain — including early warning systems, fraud detection and demand forecasting. All of these enable businesses to put their own models into production and monitor their performance simply.
This is a human-centered approach to AI. The goal is never to remove a human from a process; rather it is to simplify the process of monitoring and managing models — enabling large enterprises to deal with complexity and small-to-medium sized enterprises to achieve AI at scale with fewer employees.
We envision a future where warehouse decision-makers have access to real-time shifts in buying patterns, and can proactively make changes in orders, reduce forecasting errors, and virtually eliminate under and overstocking.
In this instance, a sample feature set may include a vendor’s current inventory level, delivery routes that pass through high traffic ports, or weather patterns that are related to delivery delays. In evaluating outcomes of a predictive delivery use case, researchers at MIT learned that container positions on a ship have different levels of priority for loading and unloading containers, and a significant delay in unloading a container could impact a delivery — while all other factors remained the same.
AI/ML solutions can process vast amounts of data from multiple sources — IoT devices, inventory systems, databases, transportation systems and other smart technologies — far more data than humans are capable of processing in a short time frame.
Yet, companies that foster human-machine collaboration will enable both to perform better, gaining deeper insights and ultimately driving better results. As companies accelerate their adoption of AI/ML to streamline production planning and warehouse management, they can take advantage of how machine learning helps deliver actionable insights to humans who can then solve problems and drive continual improvements.
The future of AI and humans is bright because the two will work together to achieve humans’ goals. This is the idea of human-centered AI, and the thought behind lowering the barriers to AI so that many can understand and work with the techniques.
Tao Liu is head of operations for Vianai Systems, Inc.