Transitions can take a long time for any company, but Jonathan Salama, CEO and co-founder of Transfix, said that moving from being a digital brokerage to a supplier of SaaS solutions is a path made easier by the fact that Transfix was always building and improving the broker-focused technology it was going to need for its own operations.
“It was a lot faster for us because we already had the right building blocks,” Salama said in an interview with FreightWaves.
The Transfix transformation has its first tangible product, a suite of AI-based solutions that the one-time digital brokerage was developing for its own internal use. The company announced its availability last week.
The full extent of just how much Transfix has changed could be seen when Salama was asked if Transfix had eyed the possibility of remaining a digital broker while concurrently offering software solutions to the brokerage community.
Brokers won’t buy from brokers
“I don’t believe a broker would buy software from another broker,” Salama said. “I don’t think we were ever going to attempt that.”
Just weeks after Transfix announced it was selling its digital brokerage operations to truckload carrier NFI and instead focusing on become a provider of SaaS products targeted to brokers and carriers, the company rolled out a suite of five solutions that are mostly using machine learning to automate more decision-making and drive greater productivity out of brokerage operations (though its announcement also noted it can be utilized by carriers). There is no one name for the entire package as its five components all perform different functions.
There is one common theme in the individual components’ names: they all begin with “auto.” The five are AutoBid, AutoAccept, AutoSched, AutoBook and AutoValidate. The latter offering is the only one that is utilizing the capabilities of generative AI.
The Transfix announcement of the product disclosed its first customer: NFI, which bought the brokerage business.
In discussing the transition to a software provider, Salama said Transfix goes into a new line of business bolstered by what it learned as a digital brokerage. “There are a lot of pain points that we needed tech to optimize to give us a better ability to get insights from our data,” Salama said. The development of solutions in-house for those issues meant that “we were building those for ourselves, but now we’re talking to customers or brokers.”
Some of those companies Transfix is targeting as a customer have not thought about these issues for a long time, “but we’ve been thinking about them for ten years so it’s really exciting,” Salama said.
The ‘tree of AI’
Salama said there is a “tree of AI.” “It’s sort of like AI is the big umbrella that has everything underneath it,” he said. “You’ve got GenAI on one side and you’ve got machine learning on the other side.” Salama added that AI also includes “neural networking,” another AI process that he said was not being utilized by any of the Transfix offerings.
In the prepared statement, David Broering, president of Integrated Logistics Solution at NFI, said that the AI evolution — ”revolution” was not used — “will accelerate the unbundling of the (transportation management system) in favor of powerful and dynamic components.”
Asked to elaborate on that theme, Salama said TMS applications “have existed forever, but they haven’t really evolved in a very, very long time.” He described them more as a CRM — customer relationship management tool — ”or more of just a database of shipments that isn’t very connected, not very logical, not very powerful or set up in a way that adapts to how you work.”
The basic “thesis” of the AI offerings from Transfix is that “there needs to be providers that bolt onto the TMS to help them be more efficient for their customers.” The goal with that efficiency, Salama said, will be to drive down operational costs, which he said is “what the broker’s life is all about.”
Asked to give an example of how the product might work, Salama turned to AutoAccept.
Most brokers now either accept 100% of the freight tendered to it, “or they need a human to make a decision on a shipment by shipment basis,” Salama said.
But a machine learning tool can be programmed to reject or accept freight based on guidelines that can be established in part by the contractual relationship between the broker/carrier and the shipper, Salama said.
When Transfix was a brokerage, Salama said calculations on how much freight should be accepted from a certain shipper often got done “in the broker’s head. We did not understand, why did you accept this load but not that load?”
The AutoAccept tool creates a “decision tree for a specific shipper” that, for example, would be trained to follow an agreed-upon tender acceptance ratio between broker/carrier and the shipper, Salam said.
And there’s a human efficiency aspect to that, as well, Salama added. “Some of our account managers had to log in on Sundays for hours on end, just to accept shipments, because customers send you shipments on Sunday and you have to accept them within the hour,” he said. “They would log in just to do this. It’s not worth the time.”
More articles by John Kingston
Bid to block Biden independent contractor rule moves to federal appeals court
EPA taking comments on California waiver for Advanced Clean Fleets rule
Werner battling unions on 2 fronts; 1 fight may have broader significance