New AI-powered agents aim to streamline everyday construction workflows, helping contractors improve productivity while keeping people firmly in control of project decisions.
An improved interface results in an enhanced user experience. Image courtesy of CMiC
Although artificial intelligence has become one of construction’s hottest technology topics, many contractors struggle to arrive at meaningful adoption. While a number of organizations have experimented with chatbots and large language models, few have successfully integrated AI into everyday operational workflows where measurable business value can be realized.
In fact, according to Steve Cangiano, Chief Product Officer at CMiC, that gap between experimentation and implementation is exactly what the company’s NEXUS AI-powered ERP platform was designed to address.
“I think it’s easy to deploy AI without a specific problem in mind,” he says. “People often expect that putting a chatbot in front of a huge database will solve everything, but that’s not how people actually work. We wanted to build AI that performs practical tasks users are already doing every day.”
Rather than focusing on generalized AI capabilities, NEXUS includes approximately 25 purpose-built agents designed to automate common construction accounting, project controls and administrative functions, while also allowing customers to build their own customized agents as they become more comfortable with the technology.
Building confidence through everyday use
For many Canadian general contractors, productivity gains often come from eliminating small inefficiencies rather than replacing entire workflows. And Cangiano believes that’s where AI can deliver its fastest return.
Among the platform’s capabilities are agents that clean and import messy spreadsheets, simulate accounting batch postings before they’re committed, and answer complex operational questions using natural language.
“The immediate return on investment is the speed at which people can get answers,” Cangiano asserts. “The information has always been inside the ERP, but historically someone had to build a report or dashboard. Now an executive can simply ask a question like, ‘What’s the margin on tilt-up concrete projects in Texas?’ and receive an answer almost immediately.”
Beyond faster reporting, Cangiano expects contractors to see measurable improvements through task automation, particularly in project accounting and cost control.
“We’re not necessarily making every process dramatically different,” he says. “We’re making them just a little faster and a little easier. Those incremental improvements add up across an organization.”
Keeping humans in control
Despite growing AI capabilities, however, Cangiano rejects the notion that artificial intelligence will replace experienced construction professionals any time soon.
“I still see a very important role for the human in the loop,” he says. “AI can handle the volume of repetitive work. But people still need to apply judgment, creativity and experience.”
He goes on to explain that it’s a philosophy that also shaped NEXUS’ user interface. Rather than forcing users into a separate AI application, agents appear alongside existing ERP screens, allowing employees to review and validate the AI’s work before committee changes.
“Trust has to be earned,” Cangiano says. “Users can let the agent complete 80 per cent of a task, then immediately click into the record it created and verify everything before they finalize it.”
He believes that gradual approach will encourage adoption among construction professionals who remain cautious about relying too heavily on AI-generated output.
Surfacing problems sooner
In light of this, perhaps the greatest opportunity for AI lies not in making project decisions, but in identifying issues before they become expensive problems.
“If AI can tell you that concrete productivity has started to decline over the past two weeks, that’s incredibly valuable,” he says. “The AI isn’t deciding what to do about it – it’s simply surfacing the issue much earlier than you might have found it otherwise.”
Considering this, project teams should remain responsible for determining the appropriate response, but earlier visibility into issues can significantly improve human decision-making and reduce risk.
“I think we’re very much in a ‘trust but verify’ stage,” he suggests. “If AI alerts you to a potential problem, you still validate the information, but you’ve gained valuable time to act before the issue grows.”
And as labour shortages continue to pressure Canada’s construction industry, Cangiano sees AI less as a workforce replacement and more as a productivity multiplier.
“Construction is only becoming more complex. If we can help people accomplish a little more with the same resources they already have, that’s a meaningful step forward. Our goal is to make these tools practical enough that contractors stop seeing AI as something experimental and start viewing it as another tool that helps them build projects more efficiently.”