This article is based on the work of Adrian Kosowski, NavAlgo’s Head of Research, who was invited to talk at Hyperlog’19. It is co authored by Jan Chorowski and Zuzanna Kosowska-Stamirowska.
The AI Revolution in Logistics: the race is on. For the logistics industry, the AI Revolution is mostly about the race between traditional actors in logistics (carriers & freight forwarders) and virtual companies of the new economy (virtual & dematerialized actors, online retailers). A race, which will lead to new, exciting, next-generation services.
Bearing that in mind, how can one bring about an AI revolution? Here are some dos and don’ts with regards to obtaining a substantial benefit from AI in a reasonable time frame and at minimal risk.
Mistake n°1: Artificial Intelligence is really cool, let’s ride the hype!
For AI to make a difference, you need a comprehensive AI strategy, which is effective only as a part of a global digitalization strategy that affects your core operations. Spending money on random and simple use cases won’t cut it – many actors in the logistics industry end up with an intelligent chatbot on their website. Your chatbot can be fantastic, but is it affecting your core operations or is it just a nice-to-have?
Mistake n°2: Let’s do a quick AI roll-out based on the data we have.
A popular use case is to do sales with a Dynamic Pricing platform to increase revenues. If done in an isolated way, it will boost your company locally, without bringing you closer to being part of an AI revolution. Dynamic pricing is easy and quick to implement (as it needs sales data only), it helps ensure availability of service, and leads to a more profitable use of resources with an immediate revenue increase.
However, loyal customers may get priced out at certain periods, and it also means matching supply and demand by playing with demand. In a way, it’s like telling your customers that it’s them who should adapt – not you. Dynamic pricing often has little impact on key aspects of your core business, and effectively moves relationship with your clients towards price, and possibly even away from you, towards price comparison platforms. As such, if you expect your operations to be demand-driven, make sure that your AI vision is aligned with the demand-driven strategy of your business.
If you expect your operations to be demand-driven, make sure that your AI vision is aligned with the demand-driven strategy of your business.
A global approach as we see it:
When it comes to AI, a typical fear that tends to arise is represented in the following quote: “For me it’s too early for AI: we are not even done with digitization yet”.
In fact, nothing can be further from the truth. For almost any business, the digitization strategy and the AI strategy share a common foundation: Data acquisition. Defining this foundation is of the absolute essence. A natural answer to both involves data acquisition from telemetry, or simply: IoT investment.
When placing IoT & data acquisition at the foundation of a complete AI vision, a key question is “how far can we go with IoT data in logistics?”.
In logistics, IoT data has direct commercial value to clients, and makes forecasting and prediction-based optimization possible. In addition, IoT data traversing different supply chain actors helps work around many traditional data integration issues for the supply chain. We argue that the pipeline – or perhaps more accurately, the “stack” – of layers of intelligence based on IoT can be represented like this:
Strategic questions then arise and need to be approached comprehensively:
How much should one invest in IoT? What solutions can be developed in-house? What elements “pool” to collaborative platforms? What should one leave out? The optimal quality of a solution & investment budget can only be achieved as part of a coordinated strategy. Different benefits, costs, and threats are associated with leaving different elements to third parties.
For large operations:
IoT and AI are complementary but “competing” for budget. AI platform budgets typically saturate while IoT budgets grow with the number of devices. The decision then is whether one needs to invest more in AI or not, while for IoT, it is a more subtle decision as to how many devices one needs – what type, and where to put them.
To make it a bit harder, an IoT pilot has to be performed before the AI platform is really working (since IoT is the source of data for AI). It thus is a subtle investment process based on a mix of technical and business intuition, and possibly iterative budgets.
For smaller operations:
The typical needs of a smaller logistics business are a subset of the IoT pipeline (see above). Smaller players have to make a bigger, collaborative effort, and have less luxury in acting sub-optimally. Good IoT coverage & data pooling (which is different then data sharing!) among similar businesses, using AI in a 3rd party platform, may be the path to success.
For small-to-medium operations:
Aggregating data from multiple businesses to develop AI for forecasting & optimization typically benefits the operations of these businesses. Having an external provider of AI methods is perhaps a comparatively harmless decision.
Defining your AI strategy impacts not only how your entire business will operate, but also what you will mean to your clients after the AI Revolution.
With these elements in mind, we conclude at NavAlgo that defining your AI strategy impacts not only how your entire business will operate, but also what you will mean to your clients after the AI Revolution.
AI can help you obtain substantial benefits. Technology should help push the boundaries of your core business, and should not be a mere buzzword. Clients will not remember you had AI in process, they may however remember that the service you delivered matched their expectations of a logistics service in the new decade – the logistics of the 2020’s.
Thanks for reading this, if you have questions or remarks, please reach out, we will be happy to answer them to the best of our ability. Cheers!
NavAlgo develops an AI engine for Things in motion. Their aim is to facilitate and automate the extraction of value from data in supply chains. They are working with some of the leaders of the logistics industry.
Bios:
Jan Chorowski – Head of AI at NavAlgo. Jan is a world class AI expert, co-author of the God Fathers of AI. In addition to his PhD in Neural Networks, he holds an MSc in Electronics and Electrical Engineering. An ex-Googler who worked at Google Brain, he specializes in deep learning architecture design, speech processing, and natural language models, and developed a great taste for challenges in the logistics sector.
Adrian Kosowski – Head of Research at NavAlgo. Adrian is a world class algorithmics and AI expert who received his PhD in Computer Science at the age of 20. He received numerous prizes for his scientific achievements. Thanks to his multidisciplinary profile, Adrian naturally tends to take a holistic approach to challenges which he encounters, especially in logistics. He was a researcher at Inria and a professor at Ecole Polytechnique (France). He is also a second time co-founder and a coach for top competitive programmers.
Zuzanna Kosowska-Stamirowska – CEO of NavAlgo. Zuzanna specialises in the extraction of economically valuable information from network data and complex systems analysis. She holds a Master’s degree in Economics and Public Policy from Sciences Po, Ecole Polytechnique, and ENSAE, as well as a PhD degree in forecasting of maritime trade. The results of her PhD were published in the Proceedings of the National Academy of Sciences of the United States of America (PNAS). She also advises policy makers on digitization and port strategies.
Photo credit: piqsels.com