You’ve likely already heard some big claims about how machine learning can streamline your supply chain and benefit your business. But while the bold promises are prevalent, the proof is harder to come by.
In a recent webinar, Eric Fullerton, Senior Director of Product Marketing at project44, chatted with Cole Newell, Principal Product Manager at project44, and Ben Zifkin, project44’s Manager of Engineering, to demystify machine learning and talk through five specific and beneficial use cases.
1. Smarter Geofences
You’re likely already somewhat familiar with a geofence — a virtual boundary of a physical location that triggers a ping whenever a person or vehicle crosses. But you’re probably also already familiar with their shortcomings.
While they enable real-time visibility and more accurate milestones for truckload shipments, they’re also challenging to scale to thousands of locations. And doing so often generates inaccurate results that can prematurely end tracking, create false arrivals, and lead to missed appointments and rescheduling.
Machine learning in project44 improves geofences to give more accurate and actionable data. It automatically generates smaller geofences based on actual pings and uses historical data to consider how a customer’s truck typically moves. As Eric explains, “We’re taking all of that information and enhancing that geofence to make it more accurate.”
2. Predictive Truckload ETAs
ETAs have “estimate” right in the name, which means they’re hardly exact. Most times, they’re notoriously unreliable and present a massive challenge for over the road shippers. “Shippers can’t take action on an ETA if they can’t trust it,” Coleman says.
Carriers also typically don’t provide dynamically-updated ETAs and missed appointment windows are incredibly common. “In fact, the median truckload appointment window is off by over six hours,” he adds.
While a shipment’s drive time is fairly predictable, it’s much harder to calculate things like facility dwell and unpredictable driver behavior — which are some of the biggest causes of late arrivals and inaccurate ETAs.
“project44 has developed a machine learning-based, best-in-class predictive truckload ETA. That ETA is calculated by a machine learning model trained on billions of data points across tens of millions of truckloads,” explains Coleman.
The machine learning models take all kinds of factors into account, including driver behavior, seasonality, and truck and load characteristics to provide better insights for shippers. Plus, ETAs are dynamically updated at every stage of a shipment.
“It accurately detects late loads and reduces errors by over 60% compared to static appointment windows and schedules,” Coleman adds.
3. Better Ocean Carrier Data
One of the biggest challenges in ocean shipping is receiving reliable milestones. You need these insights to be as close to real-time as possible to inform strategic decision-making and performance monitoring.
“project44 generates berthing geofences around all major ocean ports and terminals that are really precise and accurate,” Coleman says of another way machine learning can provide more accurate and helpful data.
That’s combined with satellite tracking data to get thorough visibility into ocean milestones. Because these milestones use GPS tracking data and geofences, they’re discovered in real-time.
“As soon as that vessel enters the geofence, we can detect the milestone,” shares Coleman. “Or if it departs the geofence, we can detect the departure milestone. And on average, that’s about four times faster than waiting for a carrier-reported event to show up.”
4. Predictive Ocean ETAs
Unreliability is one of the biggest challenges with ETAs — and that holds true on the ocean as well, particularly with seemingly relentless disruptions.
“We all know that the container shipping landscape has rapidly changed over the past three years in multiple different directions,” shares Coleman. “There have been a number of dramatic disruptions and abatements. Keeping up with that is really difficult and makes ETAs more challenging.”
Ultimately, it prevents shippers from doing what they need to do with an ETA: take action to improve their supply chain. Fortunately, project44 has machine learning ocean ETAs that are far more detailed and specific. For example, you can receive a milestone for not only when a vessel arrives at a stop, but also when the container is discharged.
“project44’s ETAs for ocean shipments take into account sailing schedules, carrier ETAs, rotation performance and lane performance, port congestion, and attributes of the vessel and the container itself,” Coleman adds.
The algorithm is also constantly tuned and optimized to do something slightly different depending on where the container is in the journey, so you get the most meaningful insights at the right time.
5. Movement GPT
Movement is project44’s all-in-one supply chain visibility platform. While it’s intuitive to use, the inherent complexity of supply chains means some filter queries can still get pretty complex.
That’s why we incorporated Chat GPT into the platform to create Movement GPT, which means you can use natural language to search for shipments. Simply ask which carrier lanes to use or where there’s more capacity and the platform will provide a helpful answer.
“Given that we have visibility across regions and modes, this allows us to give the most complete and in-depth answer to user searches rather than what other platforms can provide,” Ben explains.
Machine Learning: Less Mystery, More Meaningful Impacts
AI and machine learning have garnered a lot of buzz in recent years. But while it’s easy to find lofty claims and bold promises, it’s a lot harder to figure out just how machine learning can improve your supply chain and operations.
As the above five use cases show, machine learning isn’t just a buzzword — it can have some real, meaningful impacts on your ability to do what’s most important: make strategic decisions to streamline your supply chain.
Want to learn even more about advanced technologies and how they can drive efficiency and effectiveness? Watch the replays of the webinars in this series and register for the next one.
