Richard
Hello and welcome to this Lloyd's List podcast. My name is Richard Clayton, and I'm the Chief Correspondent at Lloyd's List. Now there's an increasing buzz around artificial intelligence and machine learning over the past two years. Uh, the shipping and logistics businesses have stopped writing them out in long hand and now talk openly about the application of AI and ML as tools in the digital toolbox. But I think there's a danger here. While some are happy to play at the foreground of advanced technologies, most people have enough on their plate already without having to learn a new language. So, what's the current position of AI in shipping and logistics? How can it be used to improve operational efficiency, and in what areas? What specific areas will it have the greatest impact? Joining me on this podcast to talk through their experience of AI and ML are Jaison Augustine, Head of the Shipping and Logistics business unit at WNS, and Thomas Heydorn, Global Head of Operations and Process Transformation at ECU worldwide. Jaison is based in New York and Thomas in Hamburg. Gentlemen, welcome to both of you. I'd like to begin by asking about your own business and the impact you have seen AI make to the way your business operates. Jaison, can you begin by sharing a specific example of how AI implementation has helped to optimize supply chain management in shipping and logistics?
Jaison
Absolutely. We look at the problem statements coming to us from our global customers, which are shipping and logistics companies like ECU Line, and one of the big problems that we first started implementing AI and ML algorithms to is the problem of documentation. It is a well-known fact that this industry, while moving cargo from point A to point B, generates an information/paper trail which is quite significant. As cargo movements happen globally, it has to deal with all kinds of regulations, business rules, restrictions, sanctions and what have you. So that's the first place that we implemented AI and ML, with a technology platform that we call Malkom that digitizes the shipping instructions and applies a myriad of business rules so that the human intervention for routine tasks can be minimized. So that's the first area that we have deployed. We've had six or seven implementations of Malkom on the trucking side, on the global logistics side, and it has had an effect of improving productivity up to 50 percent from the traditional ways of capturing the documentation information. So that's the first one. We are now expanding the role of that into more work streams, particularly different areas of customer interface.
Richard
What are the key challenges faced by businesses when they adopt AI technologies in these sectors, and how can these challenges be overcome?
Jaison
The fundamental challenge is that of having a normalized clean data set where you can train the AI and ML algorithms on, and then that has been a problem that the industry has been grappling with as they migrate from legacy systems to more modern platforms and move into the cloud. So, the first challenge that companies have to overcome is to serve up all that data on a data cloud so that the algorithms of AI and ML can then be trained to generate insights, patterns and information from that. And I'm sure Thomas will sympathize with this. It is an endless journey for companies as they continue down their journey and as they grow and expand and merge and acquire companies. You know, they restart that effort again and again as they expand.
Richard
Thomas, go on with that. You have been at your position for a year now. How have things developed in that year?
Thomas
Well, it's not that I am inventing something. I have joined the company in a new position, but based on experience I take along being in the operations for so many years and going for sometimes less radical, but nevertheless very steady development into the use of technology. And I think we are just at the beginning of a new era here, as Jaison said: access to clean data and making it available to the use of machine learning. That is probably one of the biggest challenges going ahead, in particular unleashed logistics. In the past, I could see legacy is that we entrust people and the complexity of our jobs to be done by very knowledgeable resources, but the complexity of our business does increase. The complexity or the willingness to absorb all this and manage that well - it is an uphill battle. We have in logistics, in particular the last couple of years, a certain lack of resources both by quantity and quality. So, in order to stay ahead of our workflow and to build a good experience for our customers, we have to settle in with new technology. And again, this is the beginning. I think we all have to take it ahead with a lot of investment. Not only that we pay for IT, but as well we need to rethink about our work processes a lot.
Richard
Sure. I asked Jaison earlier where he saw the greatest implementation potential. Answer the same question. Where do you see AI being used, most of all, in logistics?
Thomas
Not that we have real use cases already, but the thinking is and the complexity of our customer portal in terms of certain patterns of bookings, rating and pricing exercises we do where we need to match and compare different scenarios of building a routing that's in the complexity of our business model, moving LCL freight across the globe. So, the pricing thing is definitely one, but also the unnumbered rules and regulations we have to follow in terms of compliance, statutory rules, customs, etc., where there's not always a wealth of structural information in place where machine learning can help us to bring in the structure and build a use case available to our resources.
Richard
Jaison, where have you seen the most remarkable results already? I appreciate we are at the early stages of AI. What has caught your attention?
Jaison
I think Thomas is right when he says that it's still very early days. I think it would be a fallacy to believe that there's widespread deployment of AI across the industry. People are still creating use cases, but here is where I think a lot of investment dollars and attention are going to. I think that's a good indicator of where we will begin to see some impactful use cases. Every shipping and logistics company today interfaces with their customers, predominantly through e-mail. It is not uncommon to find a million e-mails a month from their customers on a plethora of topics. It could be a shipping instruction, a rate inquiry, a service request, an amendment request, an invoice dispute, a claim or a complaint. That is really the playground that sets the opportunity for something like AI to create some sort of a recognizable pattern, direct the traffic into the correct workstreams, digitize and automate as much as possible. And of course, the exciting thing about ChatGPT and Generative AI - it gives the possibility of using the large language models to interpret some of these messages, get the right information, and also serve it back to the customer. That’s where I believe it's a huge opportunity. I alluded to this when we started speaking. That is where we at WNS are also making very big investments to create use cases. We're starting with bookings and booking amendments, for example. The ability to turn around a booking request much faster than it was done manually. And also deal with booking amendments based on the information provided. This would have a huge impact to the customer. Then you start going down that list. There are 12 or 15 other areas that you could deploy it. Most companies are also implementing and investing in CRM solutions so they can track their customer interfaces. So, the bridge between the input from the customer, the workflow engine in the middle, and then the final CRM and tracking of where they are with the customer experience is where all of this is going to have an impact. And as you can imagine, this is not going to replace human beings. This is going to make them far more effective.
Thomas
It's a challenge which we see amongst all our offices across the world. We are working on the replacement of our ERP. The complexity of this world is not getting reduced by what customers want from us as a service. We are expanding the level of what services we can offer, but the complexity for an operator should not increase because of the learning and training exercises we have to invest in. And that is as well by the existing business model we are using or we are cooperating with WNS. That's a big challenge. New people come in, and to get them really up to speed and make them productive is a long-lasting process at our side, probably longer than at your side Jaison. I know that from visiting Nasik and all the good explanation I got over there. In our industry, there's a lot of talk these days about the beloved e-mails. It is an animal that’s very difficult to control. It’s a big cost, something people sometimes forget. All this data needs to be handled, understood, and handled. It needs to be stored and the flow of information should be categorized. ML can give us a big advancement on that. Malkom Pro is in discussion among WNS and ECU for worldwide deployment. So there is already a very tangible project in front of us.
Richard
Both of you have talked about the human element here; many people would see AI as getting rid of the human element. But it sounds like you need new skills and new talent to come. Where do you get those skills from?
Thomas
Walking away from a typical freight forwarding or shipping line qualification in some locations means that we need to have people who can go the extra mile. That means have the same fundamental training in our industry or are really from the IT side, and are into work process definitions and the relevant translation into an IT scripting or hard coding. That’s an experience we took when we deployed RPA a couple of years ago. Some people were actually asked to get a deployment to learn that and to think about it differently in order to build that extra process, which gives you a return if everything goes right. RPA is probably not the best case for a good long-lasting automation because as soon as you change something on your main European system, the scripting might not always be stable, but it is already a use case for getting in to a different level of technology. It requires a different education level of people we employ.
Jaison
Obviously, WNS is a very big employer. We have 62,000 employees worldwide. ECU is also a global company. The best practice is to find a partner to work with and not try to solve everything in-house because they have core competencies about onboarding their customer and moving their freight. They have their core solutions. To your question. We have a big management trainee program, and no matter what we hire them for, we know we're not hiring them for any expertise. We want to groom them. There is inherently a much higher technological orientation among the new breed of employees that we are bringing into the company. It is amazing to see how many of these young kids already can do some form of coding, which is very different from when I began 30 plus years ago. We don't have to over-engineer it. I think the orientation of the new workforce is far more technologically savvy. As some of the traditional legacy employees fade out and the younger ones come up, I think they will be far more friendly with technological solutions. The other dimension to the industry's challenge is that every company in this business has to fend for themselves because there are no common industry standards like the airlines or the banks have done. This industry stubbornly resists coming together to agree to a common set of standards, which adds to the additional complexity of making sense. No amount of artificial intelligence is going to be successful when you have so much variation and disparity. It makes it that much more challenging.
Richard
A lot of people say maritime is far behind aviation in so many ways. Jaison, I do want to ask about the maritime side - ships and shipping. How will AI make itself felt in the maritime sector in the near future?
Jaison
Interesting question! I think there are two aspects to that answer, Richard. One is the management of the physical assets. They are talking about autonomous ships at some point. It's less crowded compared to driving an autonomous car on busy streets, so there might be some value to that. There is also the whole gamut of predictive analytics, and for maintenance and repair, I think that it’s going to make it a lot more efficient to manage. When it comes to information management, I think there are some good opportunities in the whole load optimization/vessel planning area. This has traditionally been a high-end skill, employing master mariners who understand that well. I think that’s extremely suited for an AI engine where you feed in all the constraints, the details about the Bay Plan and it generates the load factor for these ships and also the terminals. I think optimizing some of that would be great areas of opportunity. And for operators, it is the entire document management and information flow, which we already spoke about.
Richard
Thomas, what are your expectations for the evolution of AI in shipping and logistics?
Thomas
It will take over information management and make data much more transparent. We have had setbacks. If we look at the development of blockchain in our industry, with TradeLens being discontinued. That's a little bit of a setback. But setting standards in terms of bringing forward information reference numbers of particular shipments that’s all coming in patches. Every operator putting his hand on cargo has to start all over again. The collaboration seen in the airline industry needs to come [for us too]. It might come from some bigger players in the market with the resources to set certain standards that others have to follow. Or we do that through a kind of association. But standard setting, building and managing data nevertheless on the last leverage of doing decision-making processes. I believe humans will play a big role and it will take a while until that is getting challenged. So, if young people ask me if it is worth joining our industry, it’s a big ‘yes’. The business must come to a certain pace of development in terms of using technology. Beyond this, there’s enough space to improve and look at the cost of logistics, with more intelligent management of load factors of a combination of transport modes. There is still a lot of good things we can do for the economical results of the company and for our environment.
Richard
Jaison, look two years – five years – ahead. How do you expect AI to become more adopted?
Jaison
I'm actually optimistic, Richard, that we will continue to advance this initiative. I think the massive growth in computing power and the availability of big data are two game-changers. But it's not going to be a science-fiction kind of scenario that I anticipate. There would be much better access to information and quicker responsiveness to visibility - track and trace documentation requests from the customers. I think those would see step changes, but at the same time, the industry is going to continue to expand and just like we had Brexit a couple of years ago that added a whole bunch of complexity, those new complexities will continue to evolve, [such as] geopolitical situations, which will create levels of complexity that we haven't anticipated. It will be a little progress, a little step back, but overall, it will be positive.
Richard
Thank you, Jaison and Thomas, for a fascinating conversation. As we've heard, artificial intelligence is no longer just for advanced businesses, its benefits are starting to trickle down to mid-level players and while it will be a little while until the shipping and logistics sectors fully embrace such innovative technologies, the push towards decarbonization is stimulating transparency, operational efficiency, and as we've heard, the search for new skills and talent required to get the best out of AI.