Improving Planner Productivity with Gen AI

Gen AI has revolutionized many aspects of life and business. Supply chain planning is no exception. The power of Gen AI is its ability to learn using a vast amount of data in a short time period. Whether it is an essay, a movie, a story, a shape, voice or a piece of art, given the number of examples it is exposed to, it can quickly learn, imitate or create. But not always accurately, even responding with “hallucination” in some cases. Like any other tool one needs to learn how to use it effectively and its relevance to the task at hand.

There are many uses in supply chain planning for the use and application of Gen AI. Many are available today. They are all aimed at improving planner productivity. A fairly straight forward example is to help planners to prioritize, and even respond to their emails and messages received by them. It can also help planners and users by guiding them to find what they are looking for and possible options available to them by learning the use cases and “study” past issues or questions asked before. Gen AI can also help to respond to customers’ requests and changes in their orders when and if demand and/or supply changes or interruptions occur. If properly connected to the optimization engine, it can find causes as well as the impacts of interruptions in the supply chain and inform or even correct the plan accordingly in real-time. Gen AI can also help in performing many scenarios without any involvement of users such that it can recommend what the best strategy should be when and if conditions change.

At this point in time, Gen AI models are incapable of performing numerical optimization or performing heuristics that can lead to optimal plans.

The main reason is that they are not exposed to enough example to learn from. The choices are exponentially large and methods of search to find the answer require specialized mathematical algorithms or heuristics that can perform the optimization. To this end, Gen AI models need to be connected to an accurate digital twin of the supply chain with the intelligence to optimize. This requires a fully connected model of the supply chain to understand the cause and impact of each interruption. A high level model of the supply chain, as in S&OP solutions, is ineffective.

At Adexa we have been deploying AI/ML techniques since the inception due to our founders’ background in the field. We have practiced AI/ML techniques not just on the demand side but also on the supply side since supply side is just as stochastic, if not more, as the demand side. Large Language Models (LLM) have been around for a while but only now we have the luxury of fast processing power that can yield the results in a reasonable time frame. We have already deployed these techniques in a number of areas to improve planner productivity by our self-correcting, self-improving and self-optimizing approach with Adexa Genies^©. Our progress in this area has been mainly because of our accuracy of modeling and digital twin at the S&OE level. Let’s work together to build a more intelligent supply chain.