Supply chains are some of the most complex and dynamic systems for businesses today and are seen by the boardroom as essential for success. However, the ongoing risk of an organization’s supply chain is often a focus point.
Take a midsize company in the life science industry; You would be amazed about the global footprint and myriads of internal and external partners/stakeholders. On average, maintaining 100-plus partners is considered average. Execution of a synchronized, end-to-end supply chain requires on average 40-plus touch points from Demand to Supply and back to Sales Order execution. Maintaining one integrated strategy to track a supply chain, maintain partnerships, and keep technology scalable as the business dynamically changes is an enormous undertaking. We no longer see companies embarking on multiple year implementations of strategic systems (with exception maybe of ERP and APS), but rather focusing on using quick return-on-investment (ROI), fast implementations and directly solving business needs. I call this the “Crawl, Walk, Run” deployment method.
While technology systems have advanced remarkably to meet the level of complexity that companies require, AI has been a massive topic of conversation in its readiness and applicability to global supply chains. Unfortunately, the downside of innovation like Artificial Intelligence (AI) is that we don’t always know how to evaluate the products being introduced. Have you not asked yourself the question: “How do I define AI?” and “Have I double-clicked enough, and do I truly understand what I am investing in and solving for?”
The right implementation of AI in this industry is absolutely essential to what supply chain experts are striving for: Global health, increased speed and access for patients to life-saving medicine, and the betterment of the world at large. To do that right, we need to focus on two things: Decoupling data volume and data actionability, and actually thinking through the “What? So What? Now What?” approach.
First– in the AI race, many vendors are quick to highlight that they have the “most” signals and the “top software that gets you all the data you need.” How many pictures have you seen that show fancy dashboards and people sitting in front of 5-7 screens with a futuristic looking tool set, with the latest lingo also talking about AI and Machine Learning (ML) enhancements?
It always ignites the “So What? and Now What?” question as fragmented, metrics-based Control Towers with massive signal loads become the “MO” for interpretation and decision making. This is the “Digital Bullwhip effect” as people have a tendency to respond to exceptions and signals without proper holistic correlations of signals and proper use of historic pattern analysis. It can actually backfire on companies as they see only increased dynamics and Digital Bullwhip as a result of their employees struggling. This results in adopting a management by decibel and “death by data” attitude.
So, what are people supposed to do with that information? The almost humorous gap here is that we prize innovation so much that we’re willing to purchase tools that our people can’t even use. What I propose we focus on instead is holistic actionability– not the volume– of the signals a given solution can provide. What technologies are they introducing on top of your data that can help your people filter down to the most actionable insights? And how is a solution capable of looking across all these partner systems and data platforms?
Let’s look at the universal problem of disruptions: There’s a discrepancy in a log, or a shipment tracker presents an alert message. The natural human instinct is to escalate! Do something about it. Bring a manager in. If you’re that manager, you go into crisis control.
But that’s not always the right call.
Sometimes a discrepancy is just an anomaly, and a moment of human instinct actually creates a noisy escalation stream that detracts from operations. Why? Because it was informed by an “individual miss” instead of a holistic understanding of the landscape. That’s a real problem for many solutions that are available on the market. Especially when that tool is being sold as AI… but is it really?
AI and ML are tools that help analyze large, historic, multi-system, end-to-end data. They help assess correlations between each system and function and solve for an outcome that is defined by the company. Using AI and ML, you can actually use all these control towers and digital twins to model and provide valuable insights to your employees. Then they can make informed actions, and more importantly– know when to ignore an anomalous or non-additive signal. That brings me to my second point– understanding the ways that our company cultures, functional setup, and general human instinct informs implementation of AI technologies. The years have seen the introduction of all kinds of new solutions like Enterprise Resource Planning, Advanced Planning Systems, and Transport & Last Mile. These tools have been instrumental in the efficiencies and responsiveness inherent to today’s supply chain. However, they also came with a flood of new signals and information no human is capable of analyzing and worse, they are fragmented and incapable of prioritizing business objectives. How can a single plant operator or logistics specialist decide what data is valuable versus driving a bullwhip? AI/ML applied across fragmented supply chains is the answer.
Because what happened with prior advancements? Desktop-stored spreadsheets and siloed decision-making we can’t seem to upgrade. Why? Because such rapid change in systems have trained people to “tune out.” We’ve created such an overload of information, that our people understandably turn to their own experience to solve problems. What we need to do is use AI to empower humans with better– not more– information, so they can make decisions informed by reality vs. instinct.
This is where the true white space lies for AI/ML. We can unify the end-to-end ecosystems of supply chains and determine through probability and deterministic modeling what should be acted upon and what can be ignored. Now we’re building models that harness that ocean of data points– and critically– deliver the right ones so our teams can make the best decisions for the business and global health at large. It's time to embrace high-quality data and build a human ecosystem that will allow that information and technology to be applied on end-to-end supply chains and let data drive.
