Getting started with AI in supply chain might not start where you think. Some try delving into deep learning or a crash course in generative AI (GenAI), but I don’t recommend starting there. Instead start with the foundation of your AI strategy, which should be an understanding of your company’s supply chain and your data. Second, avoid islands of isolation and limited impact by integrated AI into your supply chain so the results are aligned end to end. And third, lead the change needed, which in most cases is for your team to use technology more effectively, not build it themselves.
One supply chain executive told me AI makes him feel a bit like a deer in the headlights, which is why it’s worth stating that you don’t have to embark on this journey alone. Find a solution provider you trust, who will help shine the light to start where you are and evolve. Work that takes a whole team of people deliver, for the production and deployment of AI at scale, can instead be accomplished by a planner right in your company’s supply chain workflow as part of an end-to-end supply chain solution, putting the power of AI at their fingertips. Since this advice on getting started with AI in supply chain may seem counterintuitive, let me explain.
Why AI in supply chain?
First, why get started with AI in supply chain? The unquenchable appetite for adoption is driven by many factors. IBM’s Institute for Business Value found that 75% of CEOs believe generative AI will be key to competitive advantage. MHI and others continue to report that talent is the number supply chain challenge. Workers from picking and packing to planning and procurement are in short supply, stimulating interest in increasing productivity of the existing workforce. Planners are buried in tedious tasks using legacy, fragmented technology, which 48% say doesn’t help them do their job effectively, according to a survey by the boom! Global Network.
What AI has been able to do for years is find patterns and make predictions at a scale far beyond our human cognitive capacity, such as forecasting for a retailer with billions of sales records and millions of items. AI can accomplish tasks humans couldn’t do before, and do them better and faster, saving time and money. It can predict lead times for thousands of parts and automatically update changes, flagging only those for attention that fall outside of parameters a planner sets. It can increase forecast accuracy and provide greater granularity by incorporating additional signals beyond sales history using demand sensing.
What about generative AI?
Generative AI (GenAI) has introduced remarkable possibilities for increasing productivity by enabling humans to asking their system for help in their own words. Consider a planner in Brazil working with the previous lead time prediction example, who has forgotten how to update the parameters. Instead of searching for and reading multiple help documents, the planner can simply ask for help in natural language and receive an answer that is a single, synthesized explanation of those sources in Portuguese (even if the documentation is in English).
No matter how fetching GenAI’s communication abilities seem, it is important to remember that it doesn’t actually understand. Behind the curtain of wonders like ChatGPT is a probabilistic sentence completion machine, not a sentient being. Sometimes hilarious examples of its “hallucinations” illustrate its failure to understand (My Dinners with GPT-4 by Justin Smith-Ruiu is one of my favorites). Because it doesn’t understand, we need humans at the helm.
Understanding your supply chain is the bedrock, and people bring that understanding
For all its prowess, AI lacks the three c’s: context, collaboration and conscience, all uniquely human capabilities. It fails to understand because it cannot derive meaning from context, work together to solve problems, or hold us to a higher standard. Those traits are essential for business, along with understanding of your supply chain and its data. This knowledge is the bedrock upon which AI must be built, because as my favorite expert on the subject says, the three pillars of AI are data, data, and data.
A powerful advanced planning system can save planners time and provide insights, accelerating their intuition to make better decisions, as P&G has found. When a hurricane was bearing down, the instant transparency gained from their concurrent planning system allowed them to reposition inventory, solving all the expected disruptions, save for one gap in supplying bottles to a partner. Someone remembered they still had inventory of the old bottles, a eureka moment built on human expertise and institutional memory, features that a system doesn’t possess, even when enhanced with AI. It doesn’t understand.
Increase productivity by automating the obvious, those tedious tasks that require minimal thinking but much time, and save human understanding for where it counts. As one executive told me, his vision is 80% “touchless planning” – the solution does the planning, but it is guided by people, who use its outputs and insights, freeing up their human understanding for the 20% that can’t be automated, due to complexity, uncertainty, and sensitivity.
Connecting AI islands of isolation
Humans act on system-generated output to make decisions. But if those insights are isolated in functional areas not visible across the network, they will have limited impact. Even if AI provided a 100% accurate demand forecast, if you’re waiting on your production scheduler to return from vacation and correct errors in a spreadsheet or lack capacity to produce that demand, all you’ve done is create a highly-efficient silo. Concurrency combined with AI is the breakthrough, because it connects islands of isolation and aligns the nodes of your supply chain. That AI-enhanced demand forecast is truly powerful when it immediately informs supply, reducing the latency to keep them in concert.
Collaborative decisions are possible when everyone is on the same page at the same time, any time. If AI provides insights that require the supply chain professional to swivel between systems, open a tool to grab insights, download data into a spreadsheet, and then import it into the planning system, productivity goes down the drain, the risk of errors rises, and the likelihood of adoption shrinks. Only when AI is aligned and embedded across the supply chain, integrated right into the planner’s workflow, will the value of its insights truly be achieved. AI combined with concurrency is the true breakthrough in supply chain management.
Lead the change for people to use AI, not build it themselves
As Tom Davenport and Randy Bean report in MIT Sloan Management Review, their research finds that the groundswell to accelerate AI is shifting how it is done. One trend is AI moving from “artisanal to industrial,” as companies adopt platforms with the capabilities they need already built in. The second is the increase of techniques like automated machine learning and growth in “citizen data science,” which enables those with foundational understanding of their business and company data to leverage AI without a data science background.
For example, AI can improve demand forecasts by adding “signals” beyond sales history, but it takes understanding of your business and data to best guide the selection of signals. Asthma rates by region could be relevant for a life sciences company, whereas for a retailer festivals and events near stores might be more predictive. Human expertise grounded in an understanding your supply chain and data guides signal selection for demand sensing but does not depend on technical proficiency. These signals can be ingested and applied automatically and seamlessly, relying only on knowledge the planner already has, the ability to name columns on a spreadsheet.
AI brings a powerful set of techniques to bear on our supply chain problems, so exploiting that power will entail change for supply chain professionals. Upskill them with better awareness of how to leverage AI but based on a solid foundation of supply chain understanding, not an expectation of technical proficiency. Supply chains are never static, nor are the business models to run them, which means systems must be engineered for adaptability. The data feeding AI is dynamic, so data handling must be adaptive, to monitor changes automatically and orchestrate and adjust as needed. And the system itself must be adaptive, so your supply chain isn’t locked into a customized moment in time but can grow and evolve with the business.
AI can add horsepower to your engine, so don’t be afraid to get started with AI in supply chain. Start where you are, with expert knowledge of your business and data, connect your islands of isolation, and lead the change for your team to leverage AI.
Polly Mitchell-Guthrie is the VP of Industry Outreach and Thought Leadership at Kinaxis, the leader in empowering people to make confident supply chain decisions. Previously she served in roles as director of Analytical Consulting Services at the University of North Carolina Health Care System, senior manager of the Advanced Analytics Customer Liaison Group in SAS’ Research and Development Division, and Director of the SAS Global Academic Program.
Mitchell-Guthrie has an MBA from the Kenan-Flagler Business School of the University of North Carolina at Chapel Hill, where she also received her BA in political science as a Morehead Scholar. She has been active in many roles within INFORMS (the Institute for Operations Research and Management Sciences), including serving as the chair and vice chair of the Analytics Certification Board and secretary of the Analytics Society.
The post Getting Started with AI in Supply Chain appeared first on Logistics Viewpoints.