Whether they know it or not, most consumers use artificial intelligence (AI) in their daily lives. They shout out commands to Alexa. Respond to shopping suggestions on Amazon. Interact with chatbots on customer service sites. And preview movie or music recommendations on Netflix, Spotify, and YouTube.
AI has become such an underlying force that most of us just take for granted it will always be there, keep getting better, and eventually reshape how we work, live, and play. Oddly enough though, AI technologies haven’t fully extended to supply chain management where it could really bring benefits.
Indeed, an MHI Industry Report conducted last year, found only 12% of 1,001 supply chain professionals polled worked for organizations using AI, although 60% expect that to change in the next five years.
What’s been deterring them thus far? Supply chain professionals who have experience with AI, and specifically Cognitive Automation, say the biggest holdup has been a misunderstanding of what the technology does, what it doesn’t do, and what’s needed in order to successfully deploy it.
We sat down with one of those experts, Helen Davis, head of supply chain BPC & HC at Unilever, who has piloted Cognitive Automation with various companies within her career. Here’s what she had to say about the value of the technology for the supply chain, the obstacles she has faced, and where she sees adoption headed in the near future.
We have more than a few use cases right now at Unilever involving Cognitive Automation, but within the US we are focusing first on dynamic deployment and dynamic safety stock settings. In my past experiences, the focus was autonomous forecasting.
For raw and packed materials, for example, the AI would recommend we can make changes to improve efficiency based on historical data as well as real-time POS consumption data. Reducing safety stock (extra stock purchases to eliminate the risk of running out of stock on items) is a key benefit for us, of course, but even more beneficial is having the right amount of stock for our customers and consumers. Furthermore, in the future, we can let the system go on autopilot so that it will order our raw and pack materials as needed versus having to rely on alert systems.
Well, for me personally, it was always about improving service while eliminating waste.
I’ll give you an example that I think to which everyone in the supply chain can relate. If a customer pulls too much inventory, and you didn’t forecast it, everybody suddenly has to scramble. You start getting emails asking if your production team can just switch their production plans, which isn’t often feasible. Figuring out what to do about the situation takes time you and your customers don’t always have.
But with Cognitive Automation, the AI system recognizes that there’s been this big pull of inventory and it can auto adjust the production plan. It can auto order your raw and packed materials – accurately and within a nano-second – so your team does not have to spend days or weeks calculating things and doing it manually. This helps increase service and reduce potential business waste while making you far more agile in making supply chain course corrections. Not to mention, freeing your staff up to deal with other issues that might need more of a personal touch … like innovations.
It’s always been my vision to apply it across the entire supply chain. The supply chain is connected, so if a supplier or factory goes down, there must be an adjustment upstream. Whether that means moving inventory, a change in production schedule, or a different material order request, there is a consequence to the interruption that can be handled faster by Cognitive Automation.
I don’t know that many people are looking at applying it in this way across factory floors. There is a lot of opportunity to do that with batch manufacturing in particular or to predict machine failures. For the most part, most companies are currently focused on using Cognitive Automation demand planning and inventory management, but I feel there is much more that can be done with this technology.
No, because it greatly simplifies everything. When I first started using Cognitive Automation, I had many different systems. I had different WMS (warehouse management system) solutions on different versions. I had 15-year-old demand planning systems. I had ERP and SAP and a host of other systems. Many people told me, “you should just wait until they all integrate better with one another. Get a data lake first.” But the beauty of some Cognitive Automation solutions is that you do not have to wait. It sits on top of what you already have and intelligently connects the dots for you.
Well, there are a lot of people who don’t fully trust it. People tend to want to touch the forecasts. They don’t believe the system’s conclusions and recommendations are right. It’s all about change management, in one instance we had to lock demand planners out of the system so it could do its job, but they would be alerted to anomalies. And sure, in the early going, there were a couple of times it wasn’t right – but that was because we did not feed it the right information. Also, it turned out to be more right than a human. As time went on, we improved, and in turn, the Machine Learning kicked in and then the system was right nearly 100% of the time.
You need to embrace change management in five different areas.
First, you must be digitally open. In my experience, people who have been in supply chain management a long time aren’t so open to digital technology. Companies should hire leaders who are digitally open because you can’t drive digital transformation if it will be blocked. Along with this, change comes faster if driven from the bottom up vs top down, so make sure you have front-line leaders that will help with leading this change.
Second, you must have good partnerships between supply chain and IT functions. Things just won’t run properly without that. To implement Cognitive Automation, everyone has to be on the same page with how it’s going to be deployed, used, and extended over time.
The third requirement is to regularly communicate progress with Cognitive Automation and celebrate any successes while sharing hurdles that remain. Really ensure you are transparently reaching out daily and weekly with key metrics and identifying the areas in which you are working to tackle problems.
Fourth, don’t blame the system anytime something goes wrong. People tend to do that when they don’t fully understand the technology. Most of the time, its bad data going in resulting in bad information coming out. And sometimes it’s just about having too many people touching everything. You have to be ready and willing to limit that kind of access and shore up best practices for data entry.
Finally, train people using the systems to understand it and know how to work with it. They don’t need to be data scientists. But they shouldn’t be constantly pummeling your one and only data scientist with questions. Invest in training to give them enough of a knowledge base to be able to answer simple questions about the Cognitive Automation system themselves.
COVID-19 has made all of us say, “we need to be more agile.” If we weren’t agile before this pandemic has ensured we pivot.
I think Cognitive Automation is going to help us achieve that agility we’ve always lacked. In my opinion, those that can take and run with it quickly will win. Those that are apprehensive and ask “is this the right thing for us” will be left behind. Because success in this digital age is all about agility and how quickly we can adjust when things don’t go quite as planned.
Get social with us!
Follow us on LinkedIn: Cognitive Automation Community
Follow us on Twitter: CognitiveAutomation