Through rapid technological innovation in recent years, construction, one of the world's largest industries, has been lagging behind.
McKinsey reports that despite moving $10 trillion each year, the sector has averaged 1% productivity growth over the past 20 years and 2.8% of the global economy. Construction was also last ranked for perceived innovation in a survey of 600 US workers who deemed the sector “not technically competent” among 10 industries. This rug comes with serious costs. A study from the Said Business School at Oxford University found that over 90% of global infrastructure projects are behind or have budgets gone through. And in the US alone, $177 billion is wasted every year due to inefficiency, according to a survey of 600 construction leaders.
To tackle a small part of this, Bigrentz, a California-based company that has been matching contractors with contractors for heavy machinery such as forklifts, backhoes and excavators across the United States since 2012, has thwarted the business to one that operates on one phone, running entirely on AI built from the hierarchy. The model is old-fashioned machine learning, indicating that previous AI technologies other than large-scale language models still have value. Currently, the company is launching a standalone software platform for large contractors. It has the same AI system but allows for smarter sourcing on existing supplier lists.
“We mentioned spreadsheets, but they're also on paper on email chains, text messages, calls and graffiti,” said Scott Cannon, CEO of Bigrentz. “It's a very inefficient industry, and uses a yearly basis and thin margins based on productivity improvements. So giving contractors the ability to make better decisions gives them a competitive advantage.”
It all starts with a data strategy
The plan from day one has always been to leverage the vast amount of data the company is working with, but when Bigrentz launched it wasn't clear how it would go about it, Cannon said. The company tracked all its customer interactions and associated data points as it ran its day-to-day business. If a contractor submits a rental request, for example, Bigrentz sales employees will remove special requirements such as the type of equipment, site location, dates to which rental is required, delivery restrictions and accessories required. Employees then call the local vendor to see if they can meet their orders and connect with what they may be able to do. Bigrentz has saved all of its data for future use. This created a wealth of information from supplier decisions on whether orders can be met, price increases, service charges, and whether customer feedback would lead to.
In 2018, the company decided to start digging into data. The team created a US-wide grid up to square kilometres to represent where a particular supplier will deliver, delivery times, and where it will determine the price to charge at various locations, taking into account bridges, tolls and other contingencies. All this was done manually, often on whiteboards, and boredom spurred the decision to find a better way.
“The challenges of mining that information and trying to exercise it forced us to decide to use AI,” Cannon says.
New systems…and new companies
Over the years, Bigrentz began building technology teams, including employment data scientists, full stack engineering teams, and QA teams, creating machine learning models around various data sets. In 2022, we put together these models and created a new AI system, Sitestack, relying solely on technology built in-house. The company deployed its systems internally in January, autonomously handling vendor choices. Now, when a customer submits a rental request, instead of calling up around 12 vendors to meet the order, team members analyze millions of historic pricing and fulfillment records, rank suppliers in real time based on cost, proximity and reliability, and automatically select the best vendor.
Cannon said the system has become much better for training more information. The AI system was ultimately built on $500 million in sales data and over $1 billion in exchanges (the latter the company didn't win, but still provided valuable data). The data includes over 13 million supplier decisions on order requests, dozens of pricing data sets, customer feedback, and millions of other data points that allow you to predict what all-in-cost is or what a supplier will do.
Having a machine learning system means determining the best match for the vendor to suit the specific needs of the contractor. This is a major shift from the company's previous process where salespeople spent the whole day on the phone in rental yards. The companies that appear on the other side of this AI project look quite different to those that were launched a few years ago.
“The company was a bit nervous between the two different cultures. Technological culture. [on the teams building the platform] Market sales and marketing were different. It was always a small challenge. But we've reduced many people [gradually over time] We are basically just a tech company at this point for automation,” Canon said, adding that working in an industry that hates change is the biggest hurdle.
AI as the perfect tool for your work
Since launching the use of the new system in January, Canon said Bigrentz has saved more than 3,000 hours per week in terms of the time spent procuring rental services (equivalent to over 80 roles) and reduced errors by 40%. Today, the company is launching a customer-friendly version of the system, also known as SiteStack. This hopes to give customers more of the kinds of efficiency and cost savings they have achieved. From those connecting contractors and vendors to those selling construction company software, to companies that can do that with more information and management than ever before, to companies that are able to do that.
The new platform uses the same underlying AI, but provides the ability to enter information to suppliers that are already relevant to customers. Searching rentals to get stack rank results will help you see how all vendors compare that particular rental with additional vendors that are not in your current system.
Canon said the idea is to streamline industry pricing and bring more transparency. He says it is fragmented and “intentionally opaque”, some vendors offer daily rates and others provide weekly rates, making it difficult to compare apples to apples.
“What we're trying to solve because we evolved,” Cannon said. “So it's a problem, not a big problem, not just access to equipment, not just the intention. It's a decision that leads to the vendor you use. This is a really big problem. We didn't set out to build a company around AI.
