Joseph Tsidulko | Content Strategist | January 11, 2024
Recent years have focused public attention on the fragility of global supply chains. These expansive logistics networks, vital to manufacturers in every country, have been upended by transportation delays, strained by labor stoppages, and plagued by increasing complexity and interconnectedness that exacerbates their long-standing inefficiencies.
Supply chain planners looking to untangle these knotted networks are getting a boost from a cutting-edge technology that offers tremendous, and still largely untapped, potential. They’re putting artificial intelligence to work to make supply chains more efficient and resilient as we head into an increasingly globalized future.
Businesses use AI to manage and optimize supply chain activities—such as monitoring product quality, balancing inventory levels, and identifying fuel-efficient delivery routes—with more efficiency than traditional software.
Artificial intelligence (AI) is a general term for applications that simulate human intelligence and perform complex tasks. Its subfields include machine learning (ML), in which systems learn from consuming vast amounts of data rather than being programmed with step-by-step instructions. Thanks to this learning process, AI systems can outperform traditional software in functions such as deciphering information from video feeds, interpreting spoken and written text, predicting future market behavior, making decisions in complex scenarios, and surfacing insights buried in large data sets.
These kinds of capabilities are proving extremely useful in managing and optimizing workflows across almost every leg of the supply chain. For example, supply chain systems powered by ML algorithms can discover patterns and relationships within data sets that are often imperceptible to humans or non-AI systems, so they can more accurately forecast customer demand—which leads to more economically efficient inventory management. AI can also analyze factors such as traffic and weather conditions to recommend alternative shipping routes, reducing the risk of unplanned delays and improving delivery times. It can monitor workspaces to spot poor quality control procedures and health and safety violations. And new use cases are constantly emerging as supply chain professionals continue to experiment with the technology.
Companies are employing AI systems in their supply chains to help optimize distribution routes, boost warehouse productivity, streamline factory workflows, and more.
Manufacturers of finished goods often rely on hundreds, if not thousands, of components shipped from partners around the globe to arrive in their assembly facilities on a coordinated schedule. AI is proving it can find patterns and relationships buried within large data sets that help optimize these logistics networks, which span cargo freighters, delivery trucks, warehouses, and distribution centers. Supply chain optimization also requires tracking physical goods every time they switch hands. Here, AI can automate documentation with its ability to intelligently enter, extract, and classify data embedded in text files to help ensure the integrity of multiparty transactions.
Some manufacturers are taking advantage of AI in forecasting, using it to predict production capacity and optimize warehouse capacity based on customer demand. Some are enlisting AI to flag potential delays and equipment malfunctions before they cause production problems. Others are using AI to derive operational insights from large streams of data that flow from proliferating Internet of Things (IoT) devices and sensors installed across their storage and transportation infrastructure.
While AI offers many potential benefits to the supply chain, implementing the technology can be difficult and expensive. Running intelligent applications in production requires powerful computing systems—either on-premises edge servers or cloud-based instances—that typically need to receive data from integrated sensors and devices deployed in the field as part of an Industry 4.0 approach. Businesses typically realize the greatest benefits when they train machine learning models on their own data sets, an even more compute-intensive and data-dependent process.
Modern supply chains have become so intricate, entangled, and expansive that manufacturers struggle to maintain end-to-end oversight of the flow of materials and goods arriving in their facilities. AI's unique ability to rapidly analyze large data sets can illuminate the inner workings of even the most complex logistics networks.
When ingesting massive streams of logged data and other logistics signals, intelligent algorithms trained through machine learning often surface valuable insights, such as causes of variability or ways to improve capacity for processes with fixed and variable time elements that lead to bottlenecks. And AI-powered supply chain management (SCM) tools are better than traditional systems at tracking vast quantities of supplies in real time as they pass through intermediary manufacturing and distribution partners on the way to becoming finished products. This enhanced visibility and traceability can help manufacturers identify suppliers who are potentially violating quality or ethical sourcing practices.
By upgrading supply chain transparency, AI use can create time and cost savings, which we’ll describe more later. It can also help manufacturers ensure that the components they use to make their products are sourced in accordance with ethical, quality, and sustainability standards, a responsibility that regulators and many consumers expect them to fulfill. Organizations simply can’t afford to work with suppliers—even those based overseas—that infringe on labor, good governance, or environmental rules, and analytics tools embedded in AI-enabled supply chain applications can identify patterns that reveal fraudulent or unethical sourcing.
Manufacturers have been at the forefront of AI innovation, experimenting with and deploying various forms of the technology across the many production facilities, storage and distribution centers, and transport vehicles in modern supply chains. This can yield a number of benefits.
AI can make warehouses more efficient by helping organize their racking and design their layouts. By evaluating the quantities of materials transported through warehouse aisles, ML models can suggest floor layouts that speed access to and the travel time of inventory—from receiving to racks to packing and shipping stations. They can also plan optimal routes for workers and robots to shuttle inventory faster, further boosting fulfillment rates. And by analyzing demand signals from marketing, production line, and point-of-sale systems, AI-enabled forecasting systems help manufacturers balance inventory against carrying costs, further optimizing warehouse capacity.
With AI’s ability to learn complex behaviors and work under unpredictable conditions, repetitive tasks, such as counting, tracking, and documenting inventory, can be completed with greater accuracy and less labor; bottlenecks are identified and mitigated. By identifying inefficiencies and learning from repetitive tasks, AI can reduce the cost of operating a complex supply chain.
AI can also save manufacturers and distribution managers money by reducing the downtime of vital equipment. Intelligent systems, especially those processing data from IoT devices in smart factories, can identify malfunctions and breakdowns in their early stages or predict them before they happen, limiting disruptions and the associated financial losses.
AI can usually spot anomalous behavior from both humans and machines much sooner than people can. That’s why manufacturers, warehouse operators, and shipping companies are training algorithms to expose flaws in their workflows, employee errors, and product defects. Cameras installed in logistics hubs, assembly lines, and delivery vehicles feed into computer vision systems that use AI to inspect work to reduce recalls, returns, and rework. The system can catch worker and machine mistakes before products are misassembled or sent to the wrong destinations, saving time and material waste. Intelligent systems can also conduct root cause analysis, assessing large volumes of data to find correlations that explain failures and equip teams to make better fixes sooner.
AI is also directly embedded in ERP systems used to manage financial transactions as goods flow through the supply chain, helping companies avoid costly billing and payment errors.
Manufacturers are taking advantage of AI’s capabilities to manage their inventory levels with greater precision and efficiency. For example, AI-powered forecasting systems can use inventory information shared from a downstream customer to gauge that customer’s demand. If the system determines that the customer’s demand is decreasing, then it will adjust the manufacturer’s demand forecasts accordingly.
Manufacturers and supply chain managers are also increasingly deploying computer vision systems—installing cameras on supply chain infrastructure, racks, vehicles, and even drones—to tabulate goods in real time and monitor warehouse storage capacity. AI also records these workflows in inventory ledgers and automates the process of creating, updating, and extracting information from inventory documentation.
Supply chain managers can run AI-powered simulations to gain more insights into the operations of complex, global logistics networks and recognize ways to improve them.
They’re increasingly using AI in conjunction with digital twins—graphical 3D representations of physical objects and processes, such as assembled goods or factory production lines. Operations planners can simulate various methods and approaches on digital twins—how much would output increase if they added capacity at point A versus point B?—and gauge results without disrupting real-world operations. When AI selects the models and controls the workflows, these simulations become more accurate than those run with traditional computing methods. This application of AI can help engineers and production managers assess the impacts of redesigning products, swapping out parts, or installing new machines on the factory floor.
In addition to 3D digital twins, AI and ML can also help create 2D visual models of external processes so planners and operations managers can evaluate the potential impact of changing suppliers, redirecting shipping and distribution routes, or relocating storage and distribution hubs, for example.
AI systems can monitor work environments throughout the supply chain, such as assembly lines, storage facilities, and shipping vehicles, and flag conditions that jeopardize the safety of workers and the public. That might mean using computer vision to enforce the use of personal protective equipment (PPE) or verify that workers follow other company safety protocols and Occupational Safety and Health Administration standards. Or it could mean processing data from systems aboard vehicles such as trucks and forklifts to monitor whether drivers are operating them safely and soberly. When monitoring factory equipment, AI can help predict malfunctions and other potentially dangerous situations. And AI-powered wearable safety devices can increase protection: Consider sensor-enabled vests that connect to AI systems, analyzing warehouse workers’ movements and alerting them to the risk of injury based on their posture, movements, or location in the warehouse.
AI systems informed by sensors throughout distribution facilities and vehicles also help ensure that hazardous materials are properly handled and disposed of, protecting those who live and work nearby. AI can automate hazardous tasks, allowing workers to avoid situations that pose risks. For example, smart robots might use AI algorithms along with cameras and sensors to plot the most efficient route through a warehouse, then transport hazardous materials while avoiding objects in their path and relaying results to a warehouse management system. If accidents and failures occur, AI can perform root cause analysis to discover their exact causes and prevent repeats.
Manufacturers that assemble products via complex supply chains are especially dependent on timely and well-coordinated deliveries; delayed arrival of a single component can set back an entire production schedule. AI is taking on the task of lessening these delivery holdups.
Logistics companies use machine learning to train models that optimize and manage the delivery routes by which components move along the supply chain. These models can prioritize shipments based on order volumes, delivery promises, contractual deadlines, customer importance, or product availability. And they can provide all nodes in the distribution network with more-accurate estimated times of arrival, identifying shipments that, if delayed, risk creating larger problems.
By driving operational efficiencies, AI can make supply chains more sustainable and lessen their harmful environmental impact. For example, ML-trained models can help organizations reduce energy consumption by optimizing truckloads and delivery routes so trucks burn less fuel while delivering supplies. AI can also help decrease the amount of wasted product at various stages of the supply chain. Consider AI-driven production planning that analyzes past inventory levels, current demand forecasts, and real-time machine maintenance statuses to help ensure a manufacturer doesn’t overproduce.
AI is also used to analyze the lifecycles of finished products and deliver insights that contribute to a circular economy, where materials are reused and recycled. And supply chain planning and sourcing systems with built-in AI can help increase transparency across suppliers, and enable them to adhere to both environmental and social sustainability standards, such as paying workers fairly.
AI has become the gold standard for predicting demand based on both internal data signals, such as sales pipelines and marketing leads, and external signals, such as broader market trends, economic outlooks, and seasonal sales trends. Supply chain planners can use AI embedded in demand planning software to estimate not only demand but also the potential impact of scenarios such as economic downturns or severe weather events on demand, as well as on their own costs, production capabilities, and ability to make deliveries.
Putting AI to work in planning and managing supply chains can’t be done overnight. While the technology offers tremendous potential to reduce costs and simplify processes, it can sometimes be expensive and difficult to deploy. There are some common challenges companies face when infusing intelligence into their supply chain operations.
Let’s consider a hypothetical American automobile manufacturer that assembles three popular models at its plant in Michigan. The tens of thousands of parts and components—such as steel, tires, spark plugs, and needles for gauges—are mainly sourced from mills and manufacturing centers in a dozen US states, as well as Canada, China, Germany, Japan, and Mexico. Some components are produced at facilities the company owns and operates, and others come from third-party distributors.
Our hypothetical car company frequently receives massive deliveries, some from overseas on cargo freighters and others trucked from out of state or across North American borders. These supplies must eventually merge at the Michigan plant for final assembly into an SUV, truck, or sedan. But first, they need to be ordered, paid for, tracked, received, and stored in large warehouses with limited capacity that the company maintains in the vicinity of the plant.
As if operating a supply chain this large and complex wasn’t challenging enough, the car company must contend with inflation making supplies more expensive to procure and rising energy costs eating into their profit margins. Raising the prices of finished vehicles could help, but their sales leaders believe that would tank customer demand. And, in the aftermath of the pandemic, the company must satisfy new regulations that govern factory work environments, including enforcement of the use of PPE.
Concerned executives ask technology consultants whether they can benefit from AI, and where in the supply chain. Their answer is yes—and almost everywhere.
For starters, AI can outperform the company’s basic software when forecasting sales for each type of vehicle based on trends. It can also more accurately model how sales may be affected by scenarios such as gas price hikes or unexpected market penetration of electric vehicles. Those intelligent forecasts are a godsend to supply chain planners—they help them procure the right amount of supplies to fulfill demand without incurring additional ordering costs, overstocking their warehouses, or carrying excess inventory. The forecasts also give planners confidence to invest in opening, or save money by shuttering, various production lines and help ensure those lines are appropriately staffed.
Cameras connected to AI-powered visual models can monitor the car company’s production lines and the distribution facilities to make sure workers follow safety and environmental protocols. Other models trained by machine learning can analyze logistics data to help optimize shipping routes, cargo loads, and warehouse operations, boosting timely deliveries. Finally, AI and decision-making models can automate repetitive processes involved not only in handling physical supplies but also in maintaining the inventory and transaction records necessary to ensure all parties in the supply chain are paid fairly and on time.
Real automobile companies are improving efficiency, reducing errors, increasing accounting accuracy, and redeploying employees to better support business needs—saving them money in almost every area of their supply chain operations. Consider Mazda Motor Logistics, which uses Oracle Transportation Management to help identify the optimal carrier, route, and service level when distributing cars and car parts throughout Europe, increasing on-time deliveries.
Businesses often find it challenging and expensive to get AI fully running in production environments. They can take these steps—even before identifying a specific project, in some cases—to prepare a legacy supply chain planning and management system for a boost of intelligence.
Before deciding on a specific node in their supply chain to augment with AI, manufacturers may find it useful to audit their entire logistics network to identify bottlenecks, productivity drains, and error-prone processes. These audits help business planners identify where AI and other technology investments can yield the most value.
A supply chain modernization initiative usually involves multiple problems to solve, benefits to attain, and executive leaders to appease. But most manufacturers can’t afford the expense and downtime of upgrading everything at once. Before outlining specific projects, decide on priorities. Then, come up with a strategy for a far-reaching transformation that addresses the most pressing concerns in its early stages. Create a roadmap that ensures each project along the way will enable the next—and have adequate funding.
Having identified the specific facet of supply chain operations that will benefit most from an AI infusion, the work of designing the solution begins. Consider the types of systems needed—such as cloud-based applications, edge servers, data science platforms, and internet-connected devices and sensors—and how they will need to integrate with each other and existing IT resources. This is the point at which most companies, if they haven’t already, opt to engage a systems integrator or other type of consultancy with industry expertise.
Numerous technology vendors serve up supply chain solutions, and most of them claim some form of AI is built into their products. But because AI is a broad term that describes a diverse set of capabilities, there are major differences between offerings. Selecting a technology vendor is like committing to a long-term relationship—one that hopefully will endure well beyond the current project. Manufacturers, advised by their systems integrators, should carefully assess each bidder’s technological capabilities, price, and support models, as well as their corporate culture, to find a fit.
Once a company has selected a technology vendor, it begins the process of implementation and integration. Typically, a systems integrator works closely with internal IT teams and the vendor to install systems, integrate them with existing ones, and conduct testing before deploying them into production. The implementation phase usually requires some downtime, as well as a period of employee training once it’s complete. However, if cautiously scheduled and effectively executed, the switch from staging to production can be accomplished with minimal disruption.
Change can be disconcerting for employees who have done their job in the same way for a long time, even if it was labor-intensive and inefficient. Before implementing a new AI-enabled solution, create a strategy for preparing the organization to embrace it. The plan should involve communicating with workers about the problems or goals that motivated AI adoption, the productivity benefits the organization hopes to achieve, and the benchmarks leaders will use to evaluate the project’s success.
In some ways, an AI project is never fully complete. AI is a dynamic technology that constantly improves through a feedback loop of monitoring and adjustment. And even when AI-enabled systems seem to be working well, teams should experiment with modifications and collect data that tracks the results to inform further performance refinements.
A manufacturer’s supply chain spans geographically dispersed and operationally isolated facilities—often managed by multiple independent partners—and the distribution routes connecting them. Each phase of the journey from raw material or subcomponent to finished product requires distinct technology solutions. These solutions handle functions including procurement, planning, transportation, inventory, maintenance, and analytics—and they can all benefit from AI.
While these multifaceted systems do very different jobs, they cannot be siloed; data must travel alongside supplies through the entire logistics network. Oracle Fusion Cloud Supply Chain & Manufacturing (SCM) is a comprehensive suite of applications that handles and seamlessly connects each individual phase of the supply chain. These SCM applications use built-in machine learning to help improve automation, forecasts, and insights. The cloud-based software also fuels collaboration within an enterprise, as well as with external subcontractors and partners.
Does AI get better with time?
AI is a unique technology in that it’s able to improve with use. For example, the more data run through a machine learning model, the better that model gets at providing supply chain planners with useful functionality and insights.
How does artificial intelligence save time and effort in manufacturing?
Manufacturers often use AI to surface insights from large amounts of data that help them make their assembly processes, logistics networks, and workflows more efficient. The technology can also help automate repetitive tasks, reducing the need for manual labor.
Is AI the future of supply chain?
AI has proven remarkably adept at improving supply chain planning, management, and operations. The technology is already being embedded in almost every facet of supply chain operations, and new use cases continue to emerge. AI will certainly be an integral component of all supply chain management systems in the future.
Why is AI important in supply chain management?
Supply chains have become increasingly complex, interconnected, and expansive over recent years, challenging manufacturers’ ability to manage them. AI can assist by analyzing the proliferating amount of data generated by modern supply chains and using that data to develop remarkably accurate forecasts, reveal operational insights, and improve the efficiency of storage and transportation processes across vast logistics networks involving multiple independent partners.
How AI can be used in supply chain
AI can assist with almost every function of a modern supply chain, including planning, inventory and warehouse management, transaction processing, transportation, monitoring, and inspection. And new use cases for the versatile technology continue to be developed.