Expert insights

A.I. Is Making Logistics More Predictable

31/3/2023

A discussion with experts from CEVA and IBM

By using the power of computers to analyze large sets of data, Artificial Intelligence, or A.I., is already making an impact on people’s day-to-day lives. A.I.’s effect on companies both large and small will be revolutionary, a real paradigm change.

In particular, A.I.’s ramifications on the logistics business sector will be meaningful. Soon, A.I. will be used across all our activities, extracting value from the billions of data points we generate and enhancing the management of supply chains, logistics platforms, cargo flows, vehicle fleets and more.

One of the first applications of A.I. at CEVA will be to enable predictive logistics.

 

Predictive Logistics & A.I.

 

What is A.I.? 

Guilhaume Leroy-Meline | IBM Consulting France  

Artificial intelligence is a tool that augments the work of humans. Ever since the dawn of the industrial age, humans know that you need the right tool for the work that needs to be done. The economic environment today is increasingly zero-time, increasingly international and increasingly complex. There is a lot of data, a lot of information to ingest and understand -- and every year, the amount gets bigger and bigger. To manage all that, we need the right tool. We need collective intelligence: that’s humans working together. But with so much complex data, we now also need technology, and that’s artificial intelligence. 

 

What are predictive logistics?

Michael Rabaud | CEVA Logistics
Our customers are asking us for more information and more visibility to better face their current business challenges, and predictive logistics will allow us to do that, and much more. There are different species of predictive logistics, and each with its specific use. A.I.-powered predictive logistics will help us identify the best transportation routes, boost the accuracy of demand forecasting, manage inbound-outbound issues, optimize warehouse operations and staffing, control inventory, estimate shipment ETAs, and just generally smooth and improve overall logistics management. A.I.-powered predictive logistics will first simply enhance the logistics industry, but very soon, these new technologies will completely transform our work.

Guilhaume Leroy-Meline | IBM Consulting France

A.I. will let CEVA move from being just a “goods carrier” to focusing on customer relationships and customer services. A.I.-powered predictive logistics will let CEVA make supply chains easier for their customers and help them make better decisions, in near real-time. A.I. will also enable new services. For example, artificial intelligence can help determine what route to take from A to B to optimize speed and costs -- and also to make ‘greener’ decisions in terms of sustainability. Consider lean manufacturing, which requires low stock levels. However, if you have low stock levels, you need to work very fast. Working fast generally requires you to frequently change your shipping routes, and that is not a good option in terms of sustainability. A.I. can help here, because the better you are able to predict production and the better you are able to predict supply and demand, the more flexibility you have in your distribution supply chain. And that means more flexibility to make more sustainable decisions. A.I. helps make routing choices and it also helps predict supply and demand. 

Michael Rabaud | CEVA Logistics  

A.I. and predictive logistics will also allow us to be much more proactive and service-oriented with our customers. For example, instead of just reacting to delivery issues after they happen, we will be able to anticipate them, and proactively propose solutions and options to the customer. Artificial intelligence will make these sorts of services reliable enough to become a viable and realistic business strategy. 

 

What are the risks, and how can they be mitigated?

Guilhaume Leroy-Meline | IBM Consulting France 

There is the risk of bias: A.I. learns from data about the past to predict the future, and so the risk is thinking “We have always done it this way in the past so this way will always work in the future.” This is one of many situations where human intelligence and creativity are extremely important. We need humans to watch out for bias in the data that is used to train artificial intelligence. 

Another risk is about the robustness of the decisions made by A.I., and you mitigate this by understanding how the A.I. engine makes its decisions. We call this “trust-testing” the A.I. 

A third risk is about data privacy. We need ensure that the data we use to train an A.I. is data we have the right to use, that it is compliant with local or international regulations.

Michael Rabaud | CEVA Logistics  

There is also the risk of data security. It is said that “data is the new oil” or “data is the new gold” because it is so valuable. We mitigate this risk by protecting data with powerful, state-of-the-art cyber-security measures that cover both CEVA and the CMA CGM Group.

 

Does CEVA have any A.I.-powered services?

Michael Rabaud | CEVA Logistics  

CEVA is fully embracing predictive logistics as a company-wide strategic decision. We are looking at the full supply chain and determining which pieces of it could be disrupted or enhanced by A.I.-powered predictive logistics. We’re on a global journey to the logistics of tomorrow.

Predictive ETA is one very promising area, and in fact we have developed an A.I.-powered predictive ETA solution. This solution is very much like the app on your smartphone that you use to find out what route is the fastest to get where you’re going, but instead of being for cars and trucks, it’s for ocean-going cargo ships.

This solution was the winner of our 2022 company-wide innovation “ideathon” which tapped into the expertise and creativity of more than 70 CEVA Logistics employees from around the world. 

CEVA Logistics has a strong collaborative working relationship with IBM. They helped us run the ideathon, and then they helped the winning team build a prototype and then test and refine it. 

Philippe Goloubev | IBM Consulting France 

The relationship between IBM and the CMA CGM Group goes back 20 years. When CMA CGM acquired CEVA, IBM worked on how best to integrate the two companies. So, it was a real pleasure for our IBM Studio division to work with CEVA on the ideathon, and then to co-construct and co-execute the resulting solution. 

The service in question is based on three key elements: real-time data, a strong data engine and A.I. capabilities. CEVA wants to transition from being a transport operator to a service company, and this project is one element that will enable this transformation. 

Guilhaume Leroy-Meline | IBM Consulting France  

IBM and CEVA are a great strategic match. Using A.I. ethically is important to IBM, and CEVA has been clear that they want to use A.I. to help their employees make better and faster decisions and help their customers find more sustainable routes.

What we really appreciated in this whole experience was how there wasn’t a CEVA team and an IBM team; there was just one team composed of people from CEVA and people from IBM, all working together to deliver this project. We had the best logistics expertise and the best technology expertise. This was a complex project with new kinds of data that had to be managed in real time, and a whole new experience in terms of design. With those kinds of constraints, success can only happen if the best people are working together toward a single shared objective, and that’s what we had here.

 

When will this first fully A.I.-driven service go live?

Michael Rabaud | CEVA Logistics  

With services that can revolutionize our business like this one, we firmly believe in adopting a careful methodology and a phased rollout process. In the first half of 2023, we will deploy this new predictive logistics solution to our in-house customer service teams in a few countries. Then we will use the feedback from that soft launch to determine our roadmap for its full-scale deployment.

Philippe Goloubev | IBM Consulting France  

It’s clear that this solution is just a first step for CEVA Logistics. It builds a technical foundation, and it gives everyone involved ideas on the organizational model that will be needed to scale. This is a smart business decision. We believe that working in “incremental mode” like this is what leads to successful transformation. 

 

What does the future hold for the logistics sector?

Michael Rabaud | CEVA Logistics  
Supply chains are going to become meaningfully more agile. Today, if you make and sell T-shirts, people in the company must arrange to have T-shirts manufactured during Week W at Factory X and delivered to Warehouse Y so that it can be sold at Retail Store Z. Tomorrow, all the inventory will be managed virtually by digital twins. There will be A.I.-powered predictive logistics solutions which will know that Retail Store Z needs a specific quantity of T-shirts at a specific date, and those solutions will get the T-shirts manufactured and delivered automatically. You will still have people supervising the solutions, of course, but the whole process will be much smoother and more efficient. It’s going to change the way that supply chains are operated and completely change the way they are designed.

Guilhaume Leroy-Meline | IBM Consulting France

Technologically speaking, in the short-term future, we will be moving to broad A.I, also known as multimodal A.I. This is A.I. that can understand numbers, text and also images. It’s A.I. that can not only predict the best logistics route for a shipment, but can also explain its decision in natural language, and interact with the user to refine scenarios.

In the slightly longer term, quantum computing can support A.I for logistics. Quantum is a new way to approach intractable problems. Michael once told me there were more than 300 logistics ports in the world. That’s 10 to the power of 614 total possible shipping routes: a 615 digits number. That’s a huge number! A.I can only solve part of the challenges caused by the existence of that many routes. But, sometime in the next decade, we would be able to easily tackle this kind of complexity with a quantum computer.