In just a few years, artificial intelligence has transformed from a fun novelty to an everyday business tool. Many supply chain professionals are already experimenting with large language models (LLMs) like ChatGPT to analyze data, answer questions, and brainstorm ideas.
But shipping teams are still figuring out where AI fits. Generic LLMs can be helpful, but they often lack the context to guide real decisions.
That’s the key idea in a recent webinar hosted by Supply Chain Now, featuring Lori Boyer (host of Unboxing Logistics), Tom Butt (director of platform analytics at EasyPost), and Tyler Diestel (senior product manager at EasyPost).
Together, they explored how shipping teams use LLMs today, where the technology still has limitations, and how purpose-built AI tools are beginning to change the way organizations manage shipping.
Here’s what they discussed.
(Note: Quotations have been lightly edited for clarity and concision.)
Where shipping teams are using LLMs today
Across the industry, teams are already experimenting with LLMs in small but useful ways.
Tom explained that supply chains generate enormous amounts of data, and LLMs can help sift through that information quickly.
“Supply chains are a natively quantified space,” he said. “You can use [AI] to take all of the data, all of the statistics, and create summaries and sort through a lot of the noise to get the signal at what might be going on.”
Instead of manually digging through dashboards and spreadsheets, teams can ask direct questions about performance or trends. Many of the early use cases look something like this:
- Summarizing shipping data and dashboards so teams can quickly understand trends
- Identifying patterns or anomalies in shipping performance
- Generating ideas for cost reductions or operational improvements
- Brainstorming logistics strategies before making operational changes
Much of this experimentation is happening at the individual level; employees are trying AI tools on their own to make everyday work easier.
Lori pointed out that many companies don’t realize how widely these tools are already being used.
“Over 90% of employees are using LLMs,” she said. “Although only about 23% of shipping companies actually have an official LLM.”
The gap suggests that many teams are still early in their adoption journey.
Why generic AI tools have limits in logistics
Despite their usefulness, general-purpose LLMs weren’t designed specifically for logistics. Because they’re trained on broad information across many industries, they’re helpful for general questions but less reliable for operational decisions.
Data access
Tyler explained that the biggest challenge with generic LLMs is data access.
“ChatGPT is only as powerful as the data that it has access to,” he said. “What shippers need to do is find an LLM that actually has access to their pertinent information.”
Without access to real shipping data, an AI model simply can’t understand how your logistics operation actually works, and that missing context limits the value of its recommendations.
Security
Security also plays an important role.
Many public AI tools store or learn from the information users provide, creating risks if teams upload sensitive operational or contractual data. That means the safest and most useful AI tools are those designed specifically for shipping environments, where data won’t be leaked to individuals or organizations outside the company.
Human judgment still matters
While AI can help analyze data and surface recommendations, it’s not ready to replace human decision-making in shipping operations.
LLMs still struggle with precision tasks like complex calculations or detailed operational rules. They can also sound very confident even when the answer is incorrect.
Tyler explained why they sometimes get things wrong (and what you should do about it).
“LLMs are really just trying to predict what the most likely next digit in a sequence should be,” he said. “Sometimes it gets the answer right, but you should always be double-checking those numbers.”
Tom summed it up with a simple but effective approach: use AI to assist with decision-making, but verify the results.
The role of specialized AI in shipping
One way to overcome the limits of generic tools is to build AI systems trained specifically on logistics data.
That’s the idea behind Luma AI Advisor from EasyPost.
What is Luma AI Advisor?
Simply put, Luma AI Advisor is the expert shipping advisor that never sleeps. Like ChatGPT, it responds to natural-language questions with natural-language answers.
But unlike ChatGPT, it has deep knowledge of your specific shipping operations.
Drawing on your own shipping information, as well as EasyPost’s vast store of data, it provides tailored recommendations, makes predictions, and leads the way to smarter shipping.
Tyler described how it works: “Users have this nice LLM that they can talk to directly to ask personal logistics questions. How can I reduce my shipping cost? What carriers could I add to my logistics strategy while maintaining my same three-day SLA?”
Use cases for AI-driven shipping insights
Luma AI Advisor makes shipping less risky by allowing teams to model the potential impact of changes before actually making those changes.
For example, Tyler explained, “Logistics managers are testing what it would actually look like if they were to add a carrier like Amazon and what impacts that would have on their overall bottom line.”
Other scenarios might include:
- Identifying services that deliver the same speed at a lower price
- Evaluating how shipping strategies affect delivery timelines
- Finding opportunities to improve existing carrier usage
Tom emphasized that even small adjustments can add up quickly.
“We can use that data to literally impact where the rubber’s hitting the road and save people a bunch of money to their bottom line.”
Getting started with AI in shipping operations
For teams that are new to AI tools, the first step is often identifying where the technology could save time. Tom recommended starting with everyday tasks.
“Take a catalog of the types of things that you’re doing in your day-to-day and assess where an LLM can help save time or eliminate confusion,” he said.
Tyler suggested defining a clear operational goal first. “Set a goal for your logistics strategy,” he advised. “Whether that’s reducing parcel spend or providing a two-day SLA at the cheapest price.”
Once a goal is established, AI tools can help evaluate potential strategies and highlight opportunities for improvement.
Shipping decisions still belong to humans
At this point, you might be thinking, “If I rely so heavily on artificial intelligence, where does that leave my human team members?”
The answer?
They’re just as essential as ever.
AI is becoming a powerful tool for logistics teams, but it works best when paired with real operational expertise.
Shipping managers understand their customers, service expectations, and operational constraints in ways that no model can fully replicate. While AI can accelerate analysis and surface useful insights, experienced professionals still guide the final decisions.
As Tom put it, the goal is not to replace people but to support them.
He said it simply: “You’re still the pilot.”
Hear the full conversation
Want to explore the world of shipping LLMs even further? For the full conversation, watch the webinar recording here.
Ship smarter with Luma AI Advisor
Looking for a shipping AI that knows your business inside and out? Meet Luma AI Advisor.
Luma saves time and cuts costs by analyzing your shipping data, recommending new strategies, and giving you visibility into the results. Get in touch today to get started!
