The thousands of small trucking companies that help move goods around the United States have a fairly old-school way of doing things, according to Paul Singer. He would know — he left his product manager job at Uber Freight to start a company called FleetWorks, which he believes will modernize things.
Created during Y Combinator’s Summer 2023 batch, FleetWorks has been developing a marketplace that leverages artificial intelligence to make faster matches between carrier companies and goods that need to be moved, freeing up more time for workers on both sides of the transaction.
Singer and co-founder Quang Tran, who worked on “moonshot projects” at Airbnb, think it’s a massive opportunity, and they’ve found some serious buy-in: FleetWorks claims to have brought more than 10,000 carriers and dozens of brokers (including Singer’s old employer Uber Freight) onto its marketplace in the first six months.
To push things further, FleetWorks has raised $17 million to put toward hiring, commercial expansion, and product development, including a newly launched “always-on” AI dispatcher. The funding includes a $15 million Series A round led by First Round Capital’s Bill Trenchard, who also led Uber’s seed round in 2010. Y Combinator, Saga Ventures, and LFX Venture Partners also participated in the FleetWorks Series A.
“We ultimately went with First Round Capital to lead the round because, I mean, I think they are the preeminent investor in early-stage investing,” Singer said in an interview. “They were one of the few who really understood that we were building a marketplace company.”
Trenchard, who was also an early investor in Flexport, told TechCrunch in an interview that he thinks AI is the best way to manage all of these transactions, especially for small businesses.
“Traditional software is just not good at this. You’re structuring data before you even know exactly all of the elements that you need to structure, and you’re pushing people through your cheese grater,” he said. “This is obviously much more open-ended in the way that you can have these conversations with people and discover what their real interests are.”
There are a number of startups and established companies trying to apply AI to shipping and logistics. Oway, another YC-backed startup, is working on a sort of Uber for freight, packing trucks full that would otherwise be half-empty. Uber Freight is also getting its Fortune 500 customers to use a customized LLM to sort through all their data. Flexport, which deals more with global shipping, rolled out a suite of AI tools for its customers in February.
With FleetWorks, Singer and Tran have focused on the communication that drives the trucking world. Singer said he tries to understand how each carrier wants to communicate and will give them a mix of off-the-shelf and custom-developed voice and text models depending on what they need.
“Do they want a phone call? Do they want a text message? Do they want to come to our portal and talk to an agent there?” he said. From there, FleetWorks’ AI agents help match goods with truckers, finding out where the truckers will be, when they will be there, and what price they need to move the cargo, among other details. FleetWorks’ AI can also handle more nuanced but important details, like: Is the driver heading to a facility that requires steel-toed boots? Or does the trucker need to be home by Friday to be with his family?
These are all things that carriers and truckers already do in the current system. But it often takes dozens (or more) of calls, texts, and emails to get this part of the job done, Singer said. For small carrier fleets, which make up the overwhelming majority of the industry, that time is money.
“We sometimes imagine freight as this very rigid thing, but it’s actually this very fluid system on both sides. So for brokers, the pickup times for the appointment might be changing, or the price might be changing as the load becomes more critical when we get closer to the pickup time,” Singer said. “So the data that these two sides are collecting kind of feed this AI-powered brain, and when we detect a match, we’re able to just insert one of the carriers that uses FleetWorks directly into the broker’s system.”
That’s a lot of detail and nuance for AI software to handle, so to cut down on hallucinations, Singer said FleetWorks relies on a number of models in the background that it has designed to be specialized around particular tasks. The always-on dispatcher then pulls data from those discrete agents.
But it’s not just about the tech — a lesson Singer says he learned from his time at Uber Freight. “[It’s] about helping your customers through change management, teaching the team how to use it, and showing them that there are opportunities for this to impact your business,” he said. “We’ve gotten really good at AI implementation in addition to just building the core agents.”
That was a big attractor for Trenchard as well. “One of the things we’ve been excited about with AI just as a general thing is it’s fitting into the behaviors that people already have,” he said. “It’s not requiring you to change how you do business.
While there’s a lot of Uber DNA in what FleetWorks is building, Singer isn’t myopic. There is a lot of young software talent who want to work on solving problems in the physical world, he said. In fact, Singer joked he’s already hired some engineers who haven’t seen “Shrek” (2001).
“I never thought I’d feel like an old head,” he said.