When a container ship named the Ever Given got stuck in the Suez Canal, it was a human-interest story, especially when normal-sized construction vehicles that looked like Tonka Trucks tried to dig it out. But the Ever Given’s fate created major issues and logistical nightmares for hundreds of companies whose products were delayed indefinitely. For example, there were warnings of supply chain interruptions that could cause another toilet paper shortage. Many of these companies struggled to manage changes to the shipping schedule in real-time because they don’t have a digital infrastructure in place to analyze the signals and change data efficiently. The snowball effect of sudden changes adds significant risk on production and maintenance schedules, sending demand planning across these complex networks in a downward spiral.
Improved efficiency of the supply chain and material management relies on digital transformation. One of the benefits of making a digital transformation is the ability to accurately predict outcomes and become a more sustainable supply chain. To do this, you need to have data you can trust, that is understood, and can be analyzed quickly. The key to an effective digital transformation is AI.
However, AI can be intimidating to the uninitiated who may see platforms like IBM’s Watson and think this brilliant AI will come in and take over. They might not understand that even Watson has to be trained and conditioned for the tasks it performs, and if your AI is trained incorrectly, with wrong verifications, you are going to get poor outcomes. Frequent poor outcomes end up turning people away from AI, which can wipe out the inroads that have been made on the path to digital transformation.
Implementing AI for any purpose starts in small increments. Pick a problem that is well defined, such as an area with Indirect materials management where there is a lot of repetition involved. As AI and ML are trained for specific tasks, the data will eventually be able to provide outcomes that result in better inventory management.
AI isn’t a magic wand. It isn’t going to enrich the data you already have and it isn’t going to turn poor data into good data. AI can, however, understand data more efficiently, eliminating the need for lengthy data cleanses. This way, you know what data you have to work with and how to either use that data more efficiently or shift directions on the type of data collected.
Build a Foundation
To drive the outcomes in specific areas—rather than enriching everything—you need to build a foundation for AI to work with. Build the models and continue to add new signals as the AI is trained for the specific task.
To build the foundation, you need to bring in the experts. Not just the technology experts but experts in that area, the people who are familiar with that specific task and needs. You want someone familiar with the supply chain and demand planning who knows what data is vital, and what information would remove mundane tasks. Supervised verification teaches the technology what to apply the next time a similar situation comes up and what analysis is necessary for the required outcome.
Focused Data Not Perfect Data
Because AI and ML can handle tasks quicker and more efficiently than doing things manually, there might be a tendency to aim for perfect data. But perfect data, if it exists, doesn’t offer focus for the task at hand…and it is only as valuable as the outcomes it can drive. Create a wish list of what you want to accomplish with the data like inventory optimization, and put together a data set around that goal. If you are starting small, look at specific business unit or geography i.e. North America that can work as a training set for the rest of your global network.
Be wary of relying solely on data analyses based on history. If there is a change—say a ship stuck in a canal—the historical data alone is worthless. Agility is important. You need to be adding decision based knowledge and external signals able to keep up with current needs, and your data needs to be able to react in real-time.
Best Practices to Get Started
One of the reasons to turn to AI in your digital transformation process is to bring speed to value. What once took years now takes months or less. But AI can’t help you if you don’t make the investment. Some best practices to get started:
- Start small and win big by first defining a less risky part of your business with a quick outcome to see how AI can work for you.
- Remember that data doesn’t need to be cleansed nor perfect to be efficient for your needs.
- Think bigger from the start. Even though you may want to start the AI digital transformation small, you want to think big so you can build in logical steps.
- Get executive buy-in from your chief procurement officer and your chief supply chain officer.
Enterprise AI is not a futuristic technology that is going to swoop in and instantly do all the jobs that make the supply chain optimized. Instead, it is a pertinent tool that will make your supply chain more efficient so you can work around snags in delivery, like the Ever Given getting stuck in the Suez Canal.