Throughout the recent years of rapid technological innovation, one of the world’s largest industries has lagged behind: construction.
To tackle a small piece of this, BigRentz—a California-based company that since 2012 has matched contractors with rental yards for heavy equipment like forklifts, backhoes, and excavators across the U.S.—reinvented its business from one still operating via phone calls to one running completely on AI that it built internally from the ground up. The models are old-school machine learning, showing there’s still value in earlier AI techniques other than large language models. Now the company is launching a stand-alone software platform for large contractors, which is powered by the same AI system but allows customers to run smarter procurement on their existing lists of suppliers.
“I mentioned spreadsheets, but it’s also been on email chains, text messages, telephone calls, and scribbles on paper,” said BigRentz CEO Scott Cannon, referring to how contractors have historically handled their vendor relationships. “It’s a very inefficient industry—based on productivity gains on an annual basis—and with thin margins. So giving contractors the ability to make better decisions gives them a competitive advantage.”
The plan from day one had always been to leverage the massive amount of data the company would be working with, but when BigRentz launched it wasn’t clear how to go about it, Cannon said. The company tracked every customer interaction and associated data point as it conducted its day-to-day business. When a contractor submitted a request for a rental, for example, a BigRentz sales employee would take down the type of equipment, jobsite location, dates the rental would be needed for, and any special requirements like delivery constraints or required accessories. The employee would then call local vendors to see if they could fulfill the order and connect the contractor to one that could. BigRentz stored all that data for future use—creating a rich trove of information ranging from a supplier’s decision about whether it could fulfill the order, to price increases, service charges, and customer feedback.
In 2018 the company decided to start digging into the data. The team created a grid of the entire U.S. down to the square kilometer to represent where specific suppliers will deliver, delivery time, and costs accounting for bridges, tolls, and other contingencies in order to determine what price to charge in different locations. This was all done manually, often on whiteboards, and the tediousness spurred the decision to find a better way.
“The challenges of trying to mine that information and wield it forced us into the decision to use AI,” says Cannon.
Over the years, BigRentz started building up its technology team—including hiring data scientists, a full-stack engineering team, and a QA team—and creating machine learning models around different datasets. In 2022 it brought those models together to create its new AI system, SiteStack, relying solely on technology it built in-house. The company officially rolled out the system internally in January to autonomously handle vendor selection. Now, when a customer submits a rental request, rather than a team member calling a dozen or so vendors to fulfill the order, the system analyzes millions of historic pricing and fulfillment records, ranks suppliers in real time based on cost, proximity, and reliability, and selects the optimal vendor automatically.
Cannon said the system got much better as they obtained more information to train it on; the AI system was ultimately built on $500 million in sales data and more than $1 billion in interactions (the latter being sales the company didn’t win but which nonetheless provided valuable data). The data includes more than 13 million supplier decisions about order requests, a dozen pricing datasets, customer feedback, and millions of other data points that can predict what an all-in cost will be or what a supplier will do, according to Cannon.
Having a machine learning system determine the best vendor match for a contractor’s specific need is a huge shift from the company’s previous process in which salespeople spent all day on the phone calling rental yards. The company that’s come out on the other side of this AI project looks completely different than the one that launched years ago.
“The company had some tension between two different cultures for a bit. The tech culture [on the teams building the platform] was different than the sales and marketing on the marketplace side. That was always a bit of a challenge. But we reduced the headcount by so much [gradually over time] due to automation that we’re basically just a tech company at this point,” Cannon said, adding that working in an industry that’s averse to change has been the biggest hurdle.
Since it began using the new system in January, Cannon said BigRentz has saved over 3,000 hours every week in terms of time spent on procurement for rental services (the equivalent of over 80 roles) and has reduced errors by 40%. Today, the company is launching a customer-facing version of the system, also called SiteStack, which it hopes will make it possible to further pass on the types of efficiencies and cost savings it has realized to its customers. The launch is transforming the company yet again—from one that connects contractors and vendors to one that sells construction firms software so they can do it themselves with more information and control than ever before.
The new platform uses the same underlying AI but offers customers the ability to input information on the suppliers they already have relationships with. When they search for a rental and get the stack-ranked results, they can see how all their vendors compare for that specific rental, as well as additional vendors not in their current system.
Cannon said the idea is to streamline and bring more transparency to pricing in the industry, which he said is fragmented and “intentionally opaque” with some vendors offering day rates, others offering week rates, and other factors that make it difficult to compare apples to apples.
“What we’re trying to solve for evolved,” Cannon said. “So not just access to equipment, which is a problem, just not a big problem—no pun intended. It’s the decision-making that leads into which vendor you use, which is really the bigger problem. We didn’t set out to build our company around AI. It just turned out to be the best tool for the job.”
