AI creates technological ‘inflection point’

Greg Kefer discusses how AI has created a global inflection point, akin to those created by the introduction of the personal computer and the internet, in this episode of “On Air with Air Cargo Next.”


Yael Katzwer, Deputy Editor at Air Cargo Next

Hello and welcome to a new episode of On Air with Air Cargo Next. I am Yael Katzwer, the deputy editor with Air Cargo Next. Today, we will be speaking with Greg Kefer, Chief Marketing Officer at Raft. AI [Artificial Intelligence] is the big buzzword in the air cargo industry today with stakeholders across the supply chain, looking at how this technology can improve efficiency and assume responsibility for repetitive tasks, thereby freeing up time for employees to do more meaningful work. But change does not come easily and the industry is currently undergoing some AI growing pains as stakeholders evaluate the best uses for AI and determine where AI does not work. Raft's Greg Kefer discusses how AI is changing the landscape of the air cargo industry.

Greg Kefer, CMO at Raft

You know AI is a lot of things and when you think of air cargo, right, it's a fairly fast process, right? And I think one of the things that AI is able to do is pick up a lot of the, you know what I'll call mundane redundant tasks that are generally performed by people that actually people don't really want to do. I'm talking about rekeying data, right? Where you get a document from a partner or a carrier that needs to be loaded into another system and somebody has to type that in.There's not a magical way to do that despite three decades of trying and I think AI and its ability to read and interpret information and different types of documents and classified things. It's gonna make a big difference and add a lot to the speed of which I think is really important when you're talking about a move that takes 24 hours or so.

Yael Katzwer

What are the current obstacles that you're seeing to implementing AI?

Greg Kefer

Well, any kind of new technology, there's always a fair amount of uncertainty and newness to it that people don't fully understand. So I think part of it is just market awareness and understanding of what AI can and can't do. It's not a black box, it may someday become one. There's still a fair amount of orchestration and process to get it right. So I think like with any new technology, I mean, we saw the same thing back in the early days of cloud and I imagine even the days of the personal computer, it takes a while for people to embrace different technologies and the software itself, the tech itself to catch up with the demands of a workflow. It's something like supply chain and logistics, right. Which is fairly complicated and different. It'll take time, like anything, but I think it's here to stay. I don't think this is  another Blockchain moment where it's all hyped and hot for a couple of years and then nobody can figure it out and it fades away even though Blockchain actually hasn't faded away. It's become something much different than it was in 2015 when everyone was all excited about it.

Yael Katzwer

You said we need market awareness on what AI can and can't do, can you talk more about that? What can AI do? What can't it do for us?

Greg Kefer

First off, it's getting better at the order of magnitude level. The machines are teaching themselves now. It doesn't require an engineer to teach it to write new code. It's writing its own code. So, it's becoming smarter by itself. But again, you'll see use cases that are that kind of lead the way like writing an essay on the history of air freight. And it'll do it and you say give it, make it 200 words long and it'll do it. And that used to take someone like me a day to do that. And you know, a writer might look at it and go, that's not that great. But it checks the box. When you get down a few levels into orchestrating commerce and trade and logistics and all of the complexities that we all in the industry understand. It's gonna take a little bit of time for it to be this magical black box that does it all by itself and it may never get there. 

I think that there is a very important role that well trained humans that know this stuff have to take versus just entrusting. There will be people that get very good at writing prompts and asking it the right way to do the right things. But I think that people just have to get used to living with it and using it and, and, and the funny irony is a lot of us have been using it for years. How many people used to get a map out to find directions, when now you just talk to your phone and say, tell me the best way to the airport. It tells you. 

And I think that is probably a future state but for the time being it's going to be really finding ways where you can apply AI or Large Language Models [LLMs] to pieces of the process to really begin to create that efficiency, agility, and velocity.

Yael Katzwer

You mentioned cloud technology, I understand you were with GT Nexus when the industry really began using cloud technology, what similarities and differences do you see between how the industry is embracing and implementing these two different technologies?

Greg Kefer

Yeah, I was at GT Nexus when the internet was this new thing and nobody really knew what to do with it. The internet was going on Google and looking up information about the history of air freight or something and it was just beginning to make its way to the world of business and there was a lot of fear, uncertainty, and doubt about that. Like is it safe? Do we have our own software servers? I can't use it on the internet?  Or what happens if my web browser breaks?  And there are all sorts of security concerns. It took a good decade. The handle of cloud didn't come in really until like 2008. There were all these different terms, it was hosted on demand, software as a service platforms, and really what happened was it was largely driven by Marc Benioff and Salesforce, Is this idea of cloud, which was based on those architecture diagrams that engineers make and they had the little cloud, which meant the internet that somebody started calling it that. And that was what kind of helped people get this idea that, wow, this actually is a thing and there's some very big successful companies that are out there delivering software, as a service of a browser, on a rental basis versus having to buy a disk and install and maintain it. So the paradigm shifted but it took 10 years for it to really, to a point where GT Nexus started getting RPs coming into our inbox and we were trying to forge a category. 

We always felt that supply chain is in such desperate need of innovation because it's still being done with email and PDFs. We might have eliminated faxing, but there's still a lot of instant docs and processes. I see a very similar maturation cycle, but I think AI might go quicker. I kind of look at technology inflection points over the history of time. There's only been a handful of big ones. The personal computer was one. The internet was one. I would say maybe smartphones were one. Social networks, that information model of Facebook and LinkedIn with one. And this is the next one. It's not going away and I think that practitioners and buyers need to really get in and understand it and start testing with it. 

It doesn't mean trusting your entire operation to the AI black box and it's gonna be magical, but there are some definite proven areas where this stuff is working. At Raft that’s what we’re doing. We are definitely applying it to the consumption of unstructured data and documents and then allowing it to drive into the systems that they already use without having to rekey everything. It's just having a profound impact on data quality and velocity.

Yael Katzwer

What lessons can the air cargo industry learn from cloud implementation? As we now look to AI?

Greg Kefer

Well, when you look at any new industry and innovation like this, there's always gonna be, leaders and laggers. Leaders are companies that are more aggressive and are willing to take risks. They are the guinea pigs, if you will. They will go in early and experiment and try to take on this new technology. And we saw that with cloud. Some are huge companies, some are small companies. But what they're getting is, they're learning before everybody else. But they're also the ones that are gonna be doing the testimonials and the case studies and talking about the proof points. So that the masses, and the majority are laggards, the people that aren't sure about it and want to wait and kind of hedge their bets can come in. So we are in that phase now with AI. Where you are seeing companies that are using this today doing a lot of really cool, impactful things. These stories get out, and it's gonna lower the fear that AI is going to take over your computers and crash all your planes. There's always gonna be fear about anything different but people felt the same way about putting the credit cards on the cloud. They were worried that someone would steal their credit cards and by the way that happens. We're all using Ecommerce now, and that was a big hurdle, but with any new innovation there's gonna be bumps. In the end it's going to be a superior outcome to not doing anything. The laggards need to decide when they want to jump on because again, I don't think this is going anywhere.

Yael Katzwer

And how do you decide which new technologies to embrace? You mentioned Blockchain and how that didn't go as far as everyone thought it was going to, how do you determine “Yes, this is something we need to invest in, this is here to stay.”?

Greg Kefer

Blockchain is a great one. I couldn't understand it and I'm  in the industry. I understand the idea that you can have this Lego brick and once you made a change, it was forever etched on there. I understood that it took just enormous amounts of computing power and energy to make it happen, but I couldn't understand it and I couldn't connect it to a real workflow business process in supply chain. I think that is not happening here with AI. There are very simple things like running the report of the last 16 shipments on Lufthansa between Berlin and New York. Just ask it and, and the executive gets the report on his, on their phone.

The other one that I mentioned earlier, this idea that you relieve expensive unhappy low-level humans of the hassle of having to swivel chair key data between different systems and from reading things in multiple languages. Those are real things and they're really happening today. That’s how it’s going to happen, to really go very specific into something that people can understand. From Raft’s perspective, you know, we've got over 50 large freight forwarders doing this today. We're not the only ones, there's a lot of companies that are doing it. So, going back to what I said a minute ago as a practitioner or a buyer, somebody's trying to understand what this is. You really need to read those case studies and you really need to understand it. And when you engage a vendor about doing it, it's got to be practical. It can't be this, “Oh, it's magical”, like it goes in and it comes out, it goes in this side, it comes out perfectly on the other end. That, if that's all you're getting then don't buy it. It's probably more the realm of Blockchain, but we're not seeing that here with AI. We are definitely seeing some very pragmatic real world things that people can understand. 

And there's a lot of companies like us that are out talking about it and bringing their stories to the market like I'm doing right here with you. It's all gonna add up to a point where there's gonna be enough on there where people can really begin to see what this is and what it isn't. And when they really understand it, they're gonna be impressed about what it is.

Yael Katzwer

Speaking about specific use cases. How is Raft using AI, what are some of these use cases?

Greg Kefer

As I mentioned before, Raft is really trying to tackle this issue of information; integrity and quality and speed and accuracy. If you think of any part of the supply chain, whether it's air freight or ocean freight or trucking or whatever, one of the things that the technology has struggled with for the past 30 years is once you get past the planning phase and you go into your network to orchestrate movement of freight, for lack of a better and simpler example, you suddenly are dealing with a constellation of partners and other companies. And in some cases, this could number in the thousands of companies that are big and small that all have their own systems and their own ways of doing business. The industry has really struggled to get all of that data, all of the updates and information related, that they need to make decisions, and choose what to expedite, what they can put on air, what they can put on a ship, etc. So no matter how great the software is, without good data [there’s no point]. It's like having a Lamborghini in your driveway and gasoline hasn't been invented yet.

This is a big problem and with an email is pervasive to this day. You're seeing emails that have very important information like bills of lading, packing lists, and invoices. It's changing. A single shipment could have five updates over the course of the week and somebody has to scan an inbox and key that into some other system. So that's a really big challenge. 

At Raft, that's what we're tackling. We're focused primarily right now on AP [Accounts payable], the settlement of the bills that the freight forwarders get from their carriers. As you can imagine it for a big company like Expeditors, which is one of our customers, is a massive amount of data that's coming in paper. It's a PDF or a JPEG or a Word doc, but it's all over the place. So we're doing it for AP, we're doing it for Customs. Which is a similar experience. If you think of all the information that a customs agent/clearance person has to collect before they could file it for clearance. It's a lot. And then we're getting into logistics; the act of shipping. Thinking about booking and tracking the goods, which I think is the holy grail, is getting to that hyper granular level of visibility. But if you begin to think about visibility, all of the things that we use for finance and customs and warehousing and execution all go into that picture. We think that's a very juicy area.

And by the way, it’s not just AI that's magically doing this. The applications that people use to operate have to have this embedded into the workflows. It doesn't stand on by itself and it doesn't replace a TMS [Transportation Management System] or a system of record. The idea is that you would use the data. And then you update the system of record like a TMS or an AP system or something like that, to improve the value of those investments because those were probably bought 5 to 25 years ago. If you bought it 25 years ago, you've got 25 year old technology. But you've embedded it in your operations. So, how do you improve that without a full ripper place restart, which is gonna take 10 years and by the time you get it done, you're 10 years behind again.

$1M annual savings & 2,000 extra hours a month await.

Explore how, on average, automating workflows for 3,000 shipments a month can lead to impressive annual savings. 
It all starts with a demo.

By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.