Traditionally, big companies had the edge: They had more market information, a full-fledged IT department and a bigger reach. Because building your own IT system was complex and expensive, this created an even bigger advantage, forming an ever-increasing gap between different sizes of companies.
Nowhere is this more true than in the logistics industry. With trucking clocking in as a $720 billion market in the U.S., the land-freight industry was the breeding ground for the internet of things (IoT), where companies basically put a beacon on every truck.
Back in 1995, SalesForce led the “No Software” campaign, introducing the software-as-a-service business model. Fast forward to 2022, everything runs on SaaS: Gmail, Facebook, LinkedIn and many other corporate platforms and IT systems. When required, these platforms also integrate with APIs in order to constantly pull data from the web, process it and view it as “one screen.” Advanced organizations can even opt to move forward from traditional raw data APIs into “insight as a service,” effectively creating a power multiplier for the data being pulled into their platforms and setting their organizations up for further growth.
Software is indeed eating the world, and many SaaS applications have become household names, including Miro (whiteboards), Salesforce (CRM) and many others. Today, SaaS solutions are the norm, not the exception. But where does AI come into all of this? Can SaaS solutions and AI live in the same boat? Don’t they contradict each other?
AI Becoming Mainstream
Deep learning and AI are relatively new fields in the computational world, and they originated with Alan Turing.
Andrew Ng recently advocated in an interview with Fortune that today almost every company can get access to the same level of AI models that Google and Nasa might have by leveraging open-source technologies and code. According to Ng, AI needs a combination of AI models that are open source and high-quality data that has been “smart sized” I’ve advocated previously that best-in-class AI models and data are great when solving generalized problems, but when it comes down to domain-specific problems, the use of vertical AI is the right approach. Simply put, if you can’t understand a problem and the data around it, you won’t be able to build a model that solves it.
In recent years, we have seen a whole ecosystem evolve around the concept of what it means to bring AI into day-to-day SaaS products in the cloud. In short, as per Andy Scherpenberg, an AI architect in KPMG Belgium, companies tend to think the value chain is much shorter than it really is. In actuality, AI requires a complex set of good data, domain expertise and an engineering setup usually referred to as ML Ops.
Or in other words, there’s whole lot of engineering required to bring multiple AI models into production, and it really used to be a thing more fitting for bigger companies.
Could AI Give Shippers A Competitive Edge?
The world of logistics is built from approximately 10 million shippers across multiple verticals. The key metric is how many containers each one of them ships around the world. The biggest of which are the likes of Nike and Walmart. It is logical to assume that these companies make full use of AI.
These shippers ship between tens of containers a year to several hundreds of thousands of containers a year—obviously a big spread. These folks might be shipping anything from almonds to tires or drinks. It is these organizations that don’t have the capability to deploy AI at scale.
This problem is most apparent in ocean freight, which has a reliability score of 40%, per Freightos and SeaIntelligence. For organizations in this industry and others, the ability to include AI-predicted ETAs for day-to-day operations and more would be incredibly valuable.
How do you get started with AI implementation of this sort? Here are a few first steps.
1. Be Clear With Your Intent: What do you want to optimize(e.g., visibility on shipments)?
2. Understand: Where do you simply need data (visibility), and where does AI come in (predictions)?
3. Measure: How do you measure AI’s impact? Ask your vendor for stats and clear comparisons to industry benchmarks.
4. Start Small, Then Grow: Using AI can be a daunting task. It’s better to start with a clear, smaller projected outcome, then wait for a huge one. For example, start using an AI solution for some of your shipments before getting into the thick of TMS integration.
5. Assign Ownership: Make sure there is an internal champion for the adoption and usage of your solution.
AI can level the playing field in logistics, allowing small shippers and freight forwarders the ability to get to the same level of predictability and actionability as their much bigger counterparts. It’s time for more organizations to start leveraging that ability.
Scroll To Top