Artificial Intelligence Comes to Brokerage

HubTran CEO, Matt Bernstein, had this article published in the April 2016 issue of The Logistics Journal.

Artificial intelligence is about to transform perhaps the grittiest part of the brokerage business — the back office.

 Today, most brokers do not realize how costly their back office really is. As a broker, chances are that you are rightfully proud of your team and the shortcuts implemented. These workarounds do in fact create efficiencies… within an incredibly inefficient environment. In contrast, some of the largest brokers in the industry are now working to automate the back office, with an understanding that unnecessary cost has now become absolutely fatal.

 Payables is a typical back office process with special procedures unique to each brokerage. A list of the individual steps for managing payables can fill an entire page, single spaced. Teams keep track of all this work only through long experience and individual heroics. And the whole process becomes exponentially more error prone and costly with any kind exception — a missing POD (Proof of Delivery), a misplaced invoice, an invoice amount that does not match the carrier charge in the TMS (Transportation Management System), a payment address that does not get updated… You get the idea.

This is a tough problem to solve because of the sheer number of players in trucking: millions of shippers, hundreds of thousands of trucking companies, over 10,000 brokers, and hundreds of factoring companies. All with their own requirements and custom documents. Not to mention stuff like scale tickets, lumper receipts, or gate passes which originate from a completely different set of companies.

Some 3PLs deal with this by having their carriers send invoices by EDI (Electronic Data Interchange). But EDI is expensive and does nothing for brokers who require PODs and assessorial documentation. Other brokers purchase document management systems, which help store and retrieve documents. But they do not address the most significant problems: manual validation, manual key-stroking, and manual exception resolution.

Now artificial intelligence does. 

Just a bit of background: Companies like Facebook have developed photo-reading systems based on what is called deep learning, a technique that identifies faces and objects in photos on the social network. Using interconnected machines that approximate the web of neurons in the human brain, Facebook learns to identify photos by analyzing enormous numbers of similar images. To identify your face, for instance, it feeds all known pictures of you into the neural network, and over time, the system develops a good idea of what you look like. This is how Facebook seems to recognize you and your friends when you upload a photo.

That type of artificial intelligence can also be applied to the millions of documents circulated every day in the freight industry. Using Optical Character Recognition (OCR), the machine learns what kind of document it has just scanned, who it belongs to, and what information is in it. This allows computers to do a lot of the manual work that people currently do — an incredible leap beyond what was possible before.

As you can see this type of automation involves a lot more than just imaging and document management. And even OCR and artificial intelligence are not enough. Intelligent back-office software needs three more things to work:

  • Ease of use: Complicated software can work if you are willing to spend weeks and months re-engineering your business and training your people. But that simply is not an attractive option for most companies.
  • Integration with your TMS: Without integration, you are right back to where we started — manually entering data and attaching documents into the TMS.
  • Network architecture: Installed software cannot do the job. Artificial intelligence cannot learn all it needs to know within the four walls of a single company. There are far too many interactions and market participants. Like Facebook, it must be cloud-based software.

So the solution is super intelligent, cloud-based software that adapts to your business and people. This kind of hosted software serves many users from the same platform; it is always learning and adapting. And as a hosted, cloud-based service, the cost is low.

If you are a broker, you owe it to yourself to check out the new breed of transportation back-office software.

Matt Bernstein is CEO of HubTran Inc. He can be reached at 630–544–0459 or

Ready for your team to work 4X faster?