Barcelona, February 27, 2018.- SAP has published an interesting report on the effects of Artificial Intelligence, Machine Learning and other disruptive technologies in the Supply Chain. Its title is: “The Digital Evolution of Supply Chain Management” and we recommend its complete reading. Here we have selected only an analysis of entrepreneurs, consultants and experts.
Vineet Vashishta, Founder andChief Data Scientist V-squared
Predicting demand and matching it with production is never easy. Machine learning can help supply chain management in the following ways:
Optimization:
JIT inventory, route planning, and so many other SCM challenges are optimization problems. Machine learning has powerful applications for solving these types of problems.
Automation:
People are better at solving complex problems under uncertainty than machines. Unfortunately, so much of SCM professionals’ days are consumed by repetitive, menial tasks. Machine learning can automate even those tasks that require unstructured data analysis or human interaction. People are free to solve complex problems and create advantages for the business.
Real-Time Transparency:
No matter how complex the supply chain graph, machine learning can visualize it. That transparency provides decision support for SCM professionals. It also allows the organization to see how the supply chain enables or impacts the rest of the business.
Looking Ahead:
The supply chain graph is enabling the 1st generation of machine learning based predictive systems. That’s the most exciting potential application I see. Businesses with the ability to accurately forecast demand or supply chain disruptions will own a significant advantage in the near term.
Oliver Christie, Artificial Intelligence Strategist Foxy Machine
Prediction is currently limited by our human capability, experience, and comprehension. Thanks to thousands of years of evolution, we are great at basic pattern recognition but still rely too much on past experience of the world we know.
And while analytics can show a picture of data, it cannot understand a changing situation. What was right today will not be right tomorrow… or even in an hour. The ‘rules’ are changing faster than ever before, and to keep up we need a different approach. We need to move beyond human prediction.
AI will become the ultimate tool to both understand and act in every logistics situation. By harnessing all available data companies can gain a competitive advantage. The signal will truly be found in the noise, and acted upon. Demand will be matched to supply in ways unthought of using traditional approaches. The evolution of this approach leads to ‘Just In Time’ supply being replaced by ‘Future Now’ products – the right product, matched in real-time, to fit any need.
To build this future all that is needed is the right machine learning, lots of data and trust in the prediction.
George Vyshnya, Co-Founder and CTO SBC Group
Machine learning (ML) has the capability to forecast product demand changes in advance, providing valuable decision support for supply chain management.
This in turn helps to increase efficacy and profitability of the business. The quality of management decisions will increase, and the profit will grow as a result of efficiencies below:
- ML helps to satisfy the local demand in real-time by planning for the right products (SKUs) to be to the correct logistic hubs and retail stores. This means that less capital will be frozen in stocks, which is not being used for the end customers for a while.
- ML ensures that consumer habits in different geographical locations are accurately captured and learned.
- ML creates a highly accurate mapping of customer habits and preferences to help forecast sales and product demand in real time. Thus, it optimizes expenses and gains efficiencies in all points of the integrated supply chain.
Juan Perez, Chief Engineering and Information UPS
Artificial intelligence (AI) and machine learning (ML) are helping to improve the e ciency and accuracy of the logistics business –– and the changes will only accelerate in coming years.
Today, UPS uses machine learning in our telematics solutions to better predict component failures in our vehicles — and hopefully avoid vehicle breakdowns when our drivers are making their rounds. Our route-optimization software, ORION, already has helped our drivers shave miles off their routes. But in the future, AI will help our dispatchers and drivers make even better decisions and improve our efficiency and our sustainability.
One of the biggest challenges up and down the supply chain has always been forecasting demand. Predicting customer demand ultimately affects everything — from the size of production runs, the space needed in cargo ships, the inventory held in warehouses as well as the number of delivery trucks and drivers needed to cover the proverbial last mile. In the future, UPS will use machine learning not just to forecast demand, but also to predict everything from whether a customer has packages to be picked up to whether a recipient is available to accept delivery. With AI and machine learning, the possibilities are limitless.
Matthew York, Research Analyst Ardent Partners
In a business environment where it is critical for key stakeholders to collaborate across functional areas and business units, machine learning has the potential to more-accurately predict an organization’s product demand, manufacturing, and supply cycles and better serve enterprises and their customers.
When integrated with ERP and supply management technologies, like spend analysis, eSourcing, and eProcurement, business solutions that feature machine learning algorithms will have the ability to ingest raw customer and user data and analyze them in relation to each other to more clearly understand and predict user patterns and behaviors. For example, after analyzing inventory levels and sales, procurement, and manufacturing cycles over a period
of time, ERP, MRP, and supply management tools embedded with robust machine learning capabilities could define and predict peak demand periods as well as periods of high and excess production capacity.
Ideally, machine learning capabilities will be able to help supply chain and procurement teams and the business users they support normalize or balance their production cycles vis-à- vis customer demand to optimize their manufacturing schedules and improve margins. In other words, machine learning-enabled tools have the potential to help business leaders identify the optimal times to begin (or scale up) production to meet anticipated customer demand in a more-timely and cost-effective way.
Richard Wilding, Full Professor & Chair of Supply Chain Strategy Cranfield School of Management
AI and machine learning within the supply chain is increasingly important. There are three areas that any such system requires these are sensing capability, reasoning capability, and finally communicating capability.
Machine learning can be utilized in any of the above three areas, but important applications in the reasoning area can be applied. Machine learning works by “training” algorithms to detect patterns and create generalizable rules — enabling more effective forecasting and improved scheduling. This future state is then re-assessed by the algorithm to further learn and optimize. Thus, an improved prediction of complex demand patterns can be made to enable the supply chain to plan and schedule more effectively.
Within the supply chain context, this can be utilised to detect demand patterns (sensing), create schedules, rules and plans for inventory and delivery (reasoning), and finally communicate across the supply chain to stakeholders (communicating) of events and changes that could impact customer value and costs.
Machine learning has potential applications across all supply chain areas from transport, manufacturing, warehousing, and final delivery to name a few.
Transportation & Logistics Software 