Momentum is building around digital technologies including artificial intelligence, machine learning, Big Data analytics infrastructure, and soon, for blockchain deployments. But digitally transforming supply chains can’t be accomplished simply by sticking to a list of “hot” technologies it requires matching technologies to strategy and a commitment to change.
Ever been in a diner where there are 200-plus items on the menu and about 50 of them seem pretty good? Sifting through digital supply chain technologies can pose a similar problem. There are many intriguing choices—and they seem never ending.
A couple of years ago, Internet of Things (IoT) was top of mind, feeding real-time data into digital control towers. Now artificial intelligence (AI), machine learning, and robotics powered by AI are all the rage. And blockchain is expected to soon revolutionize the way goods are tracked from field to table. And, there are longer-term game changers like driverless trucks and aerial drones—if and when they materialize.
With so many technologies—which ones should you bet on, which ones are worth monitoring, and which ones are over-hyped? According to analysts and consultants who assess these technologies, it’s less about latching onto a technology “hot list” and more about knowing your unique priorities.
The digital supply chain trend encompasses many technologies, from more established categories of software that “digitize” a process, to newer tech like AI that can more dramatically transform supply chains. “The broadest way to look at digital transformation as it applies to the supply chain is to consider end-to-end processes, looking at digital technologies that can help us do a better job all the way from the supplier’s suppliers to the end customer,” explains Steve Banker, vice president of supply chain services at analyst company ARC Advisory Group.
Amongst the overfull menu of digital technologies, there are some that are running really hot, such as AI and autonomous mobile robots (AMRs). Analysts say some of them hold the potential to build off of each other so that data from IoT- connected things, AI-driven apps, blockchain constructs, or analytics platforms add up in a synergistic way. Emerging is an era when supply chains can be more self-correcting, leveraging analytics Platform as a Service (PaaS) foundations.
Even though there are technologies to watch right now, Dwight Klappich, vice president of supply chain execution research at Gartner, advises companies to assess their pain points first, then map back to appropriate technology. “It’s good to educate yourself on the art of what’s possible, but you have to flip the discussion around and start with what your needs are, and from there explore the technologies that will help you address them,” Klappich says.
Because supply chains are under intense pressure to do more with less, and labor resources including warehouse associates and truck drivers are scarce, it’s no surprise that AI and robotics are piquing interest. AI can be used in many forms: AMRs and other types of robotics such as collaborative piece picking arms use AI for tasks like how to navigate to a pick location or how to pick up a carton or pill bottle.
“Mobile robots are just taking off,” Klappich says. “They are seen as a way to achieve some level of automation without spending tens of millions of dollars to put in wall-to-wall automation. With AMRs, you can get some of the efficiency benefits of a highly automated facility along with benefits like adaptability and flexibility, and at a lower upfront cost.”
Because operational labor is becoming a highly constrained asset, companies are showing strong interest in using predictive analytics in areas like labor forecasting, planning and scheduling, says Klappich. AI and predictive analytics have been applied to high-level disciplines like supply and demand planning, but Klappich says that though traditional labor reporting systems are mature, warehouse operators now have interest in using machine learning to improve their intra-day and medium- to long-range labor planning.
“Today, companies can’t find enough warehouse labor, so they can’t wait until the last minute to figure out what their labor requirements are going to be. That is why labor forecasting is becoming such a huge issue,” says Klappich.
Machine learning within supply chain planning and forecasting software is continuing to advance, according to Banker, though machine learning has been around in demand planning for several years. The technology is well suited for demand management because it can constantly assess how actual sales and shipments stack up against forecasts and suggest changes to forecasts or the best type of algorithm to use, says Banker.
“It’s not new, but the level of attention and investment by supply chain planning and transportation management software providers is increasing and more [machine learning] capabilities are being brought to market,” says Banker.
Broader and deeper use of AI and machine learning within supply chains is being aided by the maturation of Big Data infrastructure that allows for quicker, less costly projects compared to previous eras of analytics, says Joe Vernon, analytics supply chain transformation leader with consulting company Capgemini.
While in some cases, providers of mobile robotics or supply chain management software embed machine learning in their solutions, in other cases, end user organizations and consultants can leverage Big Data and PaaS analytics infrastructure for supply chain objectives, Vernon explains.
PaaS are Cloud-based and may feature machine learning libraries that can be adapted to different applications, as well as Cloud storage and compute infrastructure needed to store and analyze data gathered into what’s commonly called a “data lake.” Major PaaS providers, such as Microsoft with its Azure platform, also offer business intelligence tools to visualize what machine learning has uncovered. Using application programing interface (API) calls, needed data from multiple sources can be tapped quickly and analyzed.
Using machine learning and PaaS, Vernon says, enterprises can quickly apply machine learning to issues such as how to optimize inventory levels across a distribution network or demand forecasting. “AI and machine learning have many potential applications, but one of the early success stories is inventory optimization—those classic questions of what to put where and when,” Vernon says. “The key thing with machine learning is that it learns over time. The technology sees what has happened and figures out how it can do better the next time around.”
A blockchain world?
While it has been talked about in supply chain circles for some time, blockchain use is expected to ramp up later this year with pilot projects in multiple industries, including tracking of green leafy produce in Walmart’s supply chain and control over returnable pharmaceuticals involving companies such as AmerisourceBergen.
Blockchain use in the supply chain is still new, says Vernon, but will grow, and once widely adopted, could overlap nicely with AI to support an era of self-adjusting digital supply chains. Capgemini’s 2018 research on blockchain found that 87% of surveyed companies were at least in the early stages of an initiative, though only 3% were actively using it. “We expect that use rate to grow this year,” says Vernon.
Others believe that blockchain will indeed see use in supply chains, but it might end up being more of an incremental improvement. “I have a hard time getting too excited about it because it’s just going to be a new form of messaging,” says ARC’s Banker. “It’s going to have some advantages, like better security around the messaging, but I just don’t see it as being something deeply transformative.”
However, notes Vernon, because blockchain involves a “construct” around supply chain relationships and contracts, this will tend to be useful in the long run. Once there is critical mass for blockchain, and as more companies leverage advanced analytics and machine learning in areas like inventory and demand management, Vernon believes supply chains can become more “autonomic” or self-driving.
This will be further aided by robotic process automation (RPA) of routine supply chain tasks like the creation of purchase orders within enterprise systems, and real-time data on shipments or the condition of goods coming from Internet of Things (IoT) solutions.
The vision is that these technologies will play off of each other to the point where supply chains can in many respects be self-adjusting, sort of like a digital nervous system, and much less like the traditional sequential steps that need to be planned and executed by people using distinct software solutions.
“There are various technologies that need to gain adoption for autonomic supply chains to take shape, but underneath it all, and driving the autonomic aspect, will be machine learning,” says Vernon.
Vernon believes other digital supply chain technologies such as control towers and “bots” or intelligent agents within applications will help create this new era of self-driving supply chains. To date, he adds, only part of this vision has become reality, such as bots within logistics solutions that can dynamically re-route shipments to save time and cost.
The full benefits of a digital supply chain will come from the interplay of technologies, says Nick Vyas, executive director of the University of Southern California (USC) Marshall Center for Global Supply Chain Management.
“If we can integrate artificial intelligence, machine learning and then blockchain applications, I think this combination is going to be the trinity of the digital supply chain space,” says Vyas. “The fusion of these three trends can become a huge force enabling us to live in a digital world.”
Vyas believes the main hurdle toward further adoption of these technologies is organizational rather than technical. Many companies, their Boards, and even some supply chain practitioners are used to legacy approaches to supply chain management centered on incremental gains within existing processes. Some people are also fearful that AI will cut out the human element and human expertise, adds Vyas.
“AI can be harnessed to start to make many supply chain decisions—replenishment orders, buy-or-make decisions, transportation choices—in an intelligent way where we can rapidly begin to see the benefits,” Vyas shares. “But harnessing this power requires letting go of the some of the control we currently possess over reaching decisions.”
These fears are largely unfounded, Vyas believes, because AI is often deployed as “augmented intelligence” in which the AI or machine learning algorithms are configured by human experts based on factors and thresholds set by humans.
“We want people to make some decisions—but make them on an exceptional basis,” Vyas says. “We don’t want to waste human resources on making repeatable, replicable decisions that can be digitized with AI.”
The “transformation” portion of digital chains should not be forgotten in examining what’s possible, points out Lora Cecere, founder of research firm Supply Chain Insights. In short, don’t just focus on digitizing what already exists, but on how technology can change the way you do business.
“We’ve been focused on improvements within the four walls of a company, and stuck on driving balance sheet improvements,” Cecere says. “To break out of that thinking, you have to ask yourself: What capabilities does your supply chain need to drive the business vision in digital transformation? It may start with questions like: What products do I sell, and how do I sell them? It might be that your company offers something under Cloud-based delivery, or you digitally print products, or reuse waste streams to create products. So, the first questions should be those fundamental questions.”
Cecere also believes that more adoption needs to build around standards for master data quality so that digital technologies like AI and machine learning can realize fuller potential. Fortunately, Cecere adds, the ISO 8000 standard establishes better data cleanliness for supply chains. The standard will create consistency around factors such as company/partner identifiers, using a mechanism known as the Authoritative Legal Entity Identifier or ALEI. Adoption ISO 80000 will establish data cleanliness so that supply chain partners can begin in more intelligent transportation networks, Cecere says.
“ALEI cleans up concerns like company data and location data,” says Cecere. “That’s the start of building outside-in transportation systems that allow us to better connect with third parties and sense what is happening in the supply chain. Today, enterprises operate from an inside-out perspective trying to optimize for yesterday’s problems, whereas if we are ‘outside-in’ and can connect with other parties seamlessly, we can start to be much more proactive and use predictive analytics. The first step is knowing who the parties are.”
When it comes down to it, technology like blockchain has potential in that it should improve security and trust through “immutable records,” says Klappich, but new technology alone cannot solve issues that require a heavy dose of inter-company discipline.
“In supply chain management, regardless of the technology, ecosystem enablement is very difficult,” says Klappich. “A new unproven technology alone, blockchain or any other, is not going to get suppliers to do what they are supposed to do, when they are supposed to do it, how they are supposed to do it, in the way you want it done. The issue isn’t technology, it’s ecosystem enablement.”
How to get there
With broader digital projects, adds Banker, it’s typically best to break them down into proof of concept experiments, pilot a solution, and eventually scale up use in the working supply chain. AMRs fit well with this methodology, says Banker, because they can be tested at relatively low cost, perhaps using a “robot as a service” model, and then rolled out for wider use if they meet expectations.
Other digital technologies, such as using new forms of supply chain planning that leverage AI or machine learning, are typically costlier to test and scale up than AMRs, but may do more to transform a supply chain.
“[AMRs] are a perfect example of a technology for which you can easily do proof of concepts, put a few robots in, and if it works, it works, and if it doesn’t, it hasn’t cost you a lot of money,” says Banker. “On the other hand, if you want to transform your end-to-end supply chain, and you don’t have supply chain planning solution and want to pilot that, this tends to be a costlier solution for any decent size company. But if you want to pursue that more mature view of supply chain transformation, it may be the place you need to get to.”
For some companies not as advanced in using the latest digital technologies, smaller scale “digitization” of an existing process—like putting in parcel shipping software—can add value, adds Banker.
“You may have an organization just starting to get on a digital bandwagon, and there is something they want to do, so they call it a digital transformation project to get everyone on board, but it is more of a digitization,” says Banker. “That can still add value, because you are moving from a manual process to a digital one.”
Ultimately, companies can benefit from a more modest project, or a more ambitious one. The bigger projects are well served by proofs of concept and piloting, Banker advises. “There is no such thing as a failed proof of concept, as long as you learn something,” he says.