The COVID-19 epidemic has resulted in the breakdown of supply chains globally, the economic impact of which will linger for years or maybe decades to come.
A recent survey by the Institute For Supply Chain Management, nearly 75% of companies reported some sort of supply chain issues dues to coronavirus-related transportation restrictions. What makes it worse is that the full downstream impact of the disruption cannot be known in advance.
For the first time in the last 21 years since we have been running this company, supply chain management has been the top of the mind of everyone - the company CEOs, the heads of nations and the IMF, the world economic forum and even the medical experts.
It is good to see this much attention being paid to the field we have loved and served for the last 21 years in this company and 7 years before that. But at the same time it is amusing to see so many instant experts spring up like desert flowers after the rain. A lot of these instant experts are just plain wrong and peddling wrong notions to people who know even less than they do. Many of them bring good journalistic or academic credentials and that is when their action become alarming.
There is a lot of talk that by using smarter technologies like AI a huge amount of data can be processed and such market anomalies can be predicted and a solution can be found out from data obtained during the crisis. This is a good start.
In this blog I want to focus on where exactly in supply chain AI can have the biggest impact, and how to go about deploying it.
It is important to note at the outset that use of artificial intelligence in supply chain management is nothing new. Almost all the algorithms currently in use in SCM can today be classified as artificial intelligence.
Take for example the simplest form of AI algorithms in SCM - those used for route optimisation using a simple tool such as google maps that everyone is familiar with.
All over the web there are very simple instructions on how to use google maps to get a route from point A to point B.
What happens behind the scene when you do this? Here is a simple sequence of events (sourced from this blog) and the human intervention required for these steps:
ROUTING PROCESS STEP
%AGE OF AI USED
% OF HUMAN INTELLIGENCE REQUIRED
Google looks at the two addresses you gave it, and quickly geocodes (identifies the latitude and longitude coordinates), before plotting two markers on the map.
Google will then single out all of the possible road segments in between your two points.
It scores those road segments based on factors like the shortest distance, the length of connecting road segments, and the traffic conditions at the time of the day.
It returns you the highest scoring route, and some runner-up alternatives.
In this case all the steps are automatic - there is no human intelligence required to carry out the routing. But that is not the case in every SCM process.
If you read the article on Examples of Practical Considerations in Route Optimisation you will discover the real life practical considerations in supply chain routing. Once you appreciate the Critical Functionality Of A Good Transportation Management System (TMS) it is easy to see the room for further application of artificial intelligence where currently human intervention is required because no currently available algorithm is capable of processing the decision automatically.
So, is supply chain routing the only process in which AI can be applied?
Quite the contrary - it is probably the process where most application of AI is currently possible.
Rest of the supply chain processes are even more susceptible to application of AI in future - what remains to be done is that each of these processes needs to be analysed in a great deal of detail - just like out example above.
Below is my estimate of the various supply chain processes that have a room for AI application:
SUPPLY CHAIN PROCESS
ROOM FOR AI APPLICATION - %AGE
Inventory management - planning, optimisation, time phased movement
Demand management - anticipation and fulfilment planning
Sales and operations planning
How do you work out where to apply AI?
Just like the routing process above - break down each key process into microprocesses that make up the entire process. Once you have done that it is easy to see what part of it is currently reliant on human intervention and could be automated using a script. After that it needs a lot of trial and error to get the scripts right and work out all the variations of the microprocesses. It is a tedious task, but immensely rewarding when done correctly. Expect to make a number of small wins over time that will accumulate to make a big win in due course.