It’s so hard to talk about AI without sounding pretentious or annoying. The more I engage with the topic in public industry forums, the more it feels like hype.
Recently, I was on a panel discussing AI in the supply chain industry, and much of the conversation revolved around its applications in forecasting. Immediately afterward, a group of experts privately argued that forecasting is precisely where AI models tend to hallucinate the most.
In other words, the conversation about AI is completely divorced from the reality of AI.
What struck me the most was when an actual AI scientist at the conference told me that, when he looked around, all he saw were hundreds of companies poised to waste a lot of money applying AI solutions to problems they didn’t actually have. People are implementing AI for its own sake. I couldn’t help but feel a sense of déjà vu, recalling the blockchain frenzy a couple of years ago.
Let me be clear on this – from a practitioner’s perspective, AI is far from the solution-looking-for-a-problem that was blockchain
AI Solutions Are Real. But Only if Used Right.
It’s already clear that, once the dust settles, many amazing AI solutions will emerge. The key for us at Freight Right, a freight forwarder that has always gone tech-first, is identifying the problems that lend themselves to these solutions. Fortunately, we have a lot of smart people working across the company to pinpoint key areas. A pragmatic strategy about where to concentrate efforts is also crucial in an industry as technologically slow-moving as logistics.
While I’m airing frustrations about the industry, I also have to mention the lack of useful content surrounding AI. The real change-makers are quietly working on solutions. They aren’t producing content because they’re busy doing the actual work. They also don’t typically share their work publicly because they don’t want to give competitors an edge.
Meanwhile, everyone else seems to be generating content for the sake of content, flooding the internet with AI-related noise. I will try to be as open as possible but keep things general—I don’t want to divulge proprietary secrets. Besides, we have smarter people who understand the intricacies far better than I can articulate.
What Freight Right Doesn’t Do With AI:
We aren’t using chatbots. Yes, I know that is bucking the trend.
We’ve always thought user-first. And if the user has to enter the same data into a chatbot that they would enter into a pricing tool to generate a quote, then they’ll just use the pricing tool. It’s a more natural format, and they can use it with greater confidence. Similarly, if your chatbot asks the user to enter a tracking number to give them the same results they’d get on the tracking page—with the same number of keystrokes—then I don’t see any value in that. In fact, as a business, I’d feel like I just spent my customer’s money on something they don’t need. And that is something we categorically don’t do.
We’ve also stayed away from AP automations.
Our operational processes are designed in such a way that this isn’t an issue. This is also not really an AI application. There is some model training involved, where the system learns to identify the fields needed to match invoices to jobs, but it’s more of a traditional algorithmic solution or, at best, a minor AI use case that represents an evolution, not revolution.
The existing products in this area are priced competitively in relation to labor costs in developed countries. However, if you have a shared resource office or back office in a developing country, this becomes another expensive solution in search of a problem.
We chose not to purchase a solution that suggests sell rates.
The model would be trained on our own customer data. This felt like cheating. It also felt impersonal and somewhat insensitive to our customers, as well as to our account managers, who work hard to build relationships and partner with clients. Our Account Managers are trained and coached to find ways to reduce costs for their customers. AI solutions in this space tend to focus on optimizing conversion and/or margin. I can’t yet envision a model that would learn to genuinely care about a customer’s logistics spend.
So what do we do with AI?
We focus on areas where we can tangibly improve three things:
Customer satisfaction
Increased efficiency
Reduced costs.
One of the battles we constantly fight is rate management.
If you’re a traditional forwarder with a primary focus on transpacific routes, you don’t need much. Rates are emailed to you by your agents, and most quotes you prepare are for transpac. There isn’t much to train on, and sophisticated solutions aren’t necessary. On the far side of the spectrum, if you handle unique, complex shipments that require custom pricing and planning—like project cargo—your business doesn’t lend itself well to automation.
However, if you handle shipments across various trade lanes and at a high volume, you’re likely fighting the ongoing battle of maintaining rates. I’m talking about origin trucking, destination trucking, main freight, terminal, and warehouse fees at both origin and destination, and more. Getting this information regularly from all your partners and carriers, with enough coverage to automate quoting, is a challenge that only a few in the industry are successfully tackling. Fortunately, this is where AI solutions are making a real impact.
For the same reasons, our staff spends countless tedious hours auditing rates, contracts, and benchmarks. Here, too, we are testing AI solutions to minimize the drudgery.
Most importantly, we have always fostered a culture of experimentation. We’re frequently approached by startups, and we listen to most of them. If the idea has merit and is non-invasive, we’ll test it. If it aligns with our principles of improving customer experience, efficiency, and reducing costs, we’ll even help them develop it. This philosophy isn’t just external. Our staff, too, is inquisitive and empowered to experiment, especially when it comes to AI.
