Artificial intelligence is a hot topic that has landed on the agendas of conferences across the trucking industry. AI, being an umbrella term, has leaders across the supply chain asking how best to use it in their operations.
That’s a question that 1Logtech CEO JP Wiggins set out to answer during Jarrett Logistics’ recent supply chain summit held this month. It’s a question that experts at trucking events have offered solutions to. But one question that isn’t answered as often – if at all – is what not to use AI for.
Wiggins provided some dos and don’ts during his AI-generated presentation: “AI in Logistics: What’s Real, What’s Not, And Why Connectivity Matters.” What AI can’t do, he said, is make cross-partner decisions or adapt to unstructured, real-world logistics chaos. That’s where his team comes in; it provides self‑service connectivity for EDI and API.
[RELATED: Experts weigh in on logistics jobs that could be replaced by AI]
“AI can't react to anything it can't see,” Wiggins said.
He used an analogy to explain: if he uses AI to pay his bills every month, it’s going to miss the kid down the street that he hired to mow his yard.
“That's the same thing that's going to happen with any system,” he said. “If it can't see it, it can't make decisions related to it, and that's the challenge in logistics because we're all business to business.”
Each partner a logistics company deals with controls its own data and systems. Even if two companies run the same version of a system, they use it differently. Siloed and fragmented data and broken workflows are the greatest challenge to a large language model or AI agent, he said.
A real-world example
1Logtech had a 3PL customer that needed to integrate with a global manufacturer’s ERP system. It required the 3PL to update the purchase order, but the finance department made the decision that no PO updates could happen after an order was released, meaning AI could not update the PO.
A human had to change the workflow. Once that flow was created, the AI took care of the technical details.
“If you're trying to integrate a carrier to a shipper, there are 700,000 carriers in North America alone at all levels of tech stack. You have carriers that are very sophisticated that put a billion dollars into their IT structure every year … but then you have down to the lowest level of a mom-and-pop carrier where the only technology is the phone that they're holding,” Wiggins said. “You have to integrate all those different tech levels, so that's why, if you're trying to make AI decisions there across that, that's where you need that human to augment. The AI can help you make the technical aspects, but your human has to manage the chaos.”
He said the connectivity between trading partners is AI’s weak point in logistics because AI is built on data, and if it can’t see the data, it can’t process it.
“For us in logistics, when you think about trying to use AI to solve your problems, look back towards the integration because clean and clear data across trading partners is what's going to allow the AI to work,” Wiggins said. “If you have an isolated AI, it can only answer questions in that isolation.”
Where AI can work in logistics
Wiggins said he had attorneys quote him $15,000 to $20,000 to file for a patent. He opted instead to use AI, which gave him the links to fill out the forms and suggested arguments to use in filling out the patents. Then it edited it and filed it.
He also talked about how logistics companies can use AI for something as simple as putting together a PowerPoint to using it for marketing.
Two areas where experts widely agree that AI is useful for logistics is dynamic pricing and customer service with chat and voice bots that can free up customer service agents to handle more complicated matters.
Document handling, Wiggins said, is another great area because AI can extract and validate data from things like bills of lading, invoices and proofs of delivery, for example, using natural language processes and optical character recognition. It takes a document and turns it into something a computer can read and react to.
[RELATED: Getting started with AI means starting small]
AI can also interrogate software to determine how the programming should work. That’s what his company does for Jarrett. Jarrett provided 1Logtech with a list of 30 LTL carriers to connect APIs, and just one of 1Logtech’s non-technical people was able to set up one API connection with a carrier per day using AI. At his previous company, he said it would take $10,000 and a couple of months to write a program.
“A person who was not a developer before is now able to go do a task that a developer only could do with their full research and development team,” he said. “You've eliminated the technical aspect, and you just have to focus on the human aspect. Basically, the AI handles the heavy technical lifting and that integration coding itself, but you're still doing it in such a way that the human, using their intellectual property and logistics experience, is the one that injects in where the AI falls short.”
