Construction firms are using AI to reshape roles and workflows more often than to reduce headcount. BRG’s survey of industry executives points to changes in how companies organize teams, hire and manage project information, even as broad deployment remains limited. That distinction matters: buying software is straightforward compared with changing how project teams make decisions and check their work.
Workforce changes outweigh reductions
Among North American respondents, 46% said AI had changed their team structure and 38% had retrained employees. Eighteen percent added staff in some areas, while 11% reduced headcount because of AI. The pattern suggests firms are adjusting workforce capacity around new tasks rather than replacing workers at scale. For now, the survey describes adaptation more clearly than displacement.
Hiring is changing, too. Twenty-nine percent of respondents created new role types, and 20% altered the skills they seek from candidates. Nearly a third redeployed employees to different work, while 51% changed their reliance on external vendors, consultants or contractors. Taken together, those shifts suggest companies are reconsidering where expertise sits: on staff, with outside advisers, or in a mix of both.
BRG surveyed 109 executives responsible for AI implementation in July 2026: 58 builders and 51 owners. Most findings came from 65 North American respondents. Builders were mainly small and midsized general contractors and construction managers; owners were largely larger developers and real estate investment trusts. That sample offers a view from people overseeing implementation, but it doesn't establish how widely these practices have spread across the full construction workforce.
Information-heavy work is the early focus
Richard Fultineer, BRG’s managing director and head of global construction, sees AI’s immediate value in work that depends on processing project information. He pointed to cost estimating, procurement, design review, schedule oversight, and contract and compliance review as areas where teams are applying the technology. These tasks involve large volumes of documents and project data, making them natural candidates for tools that can search, compare and summarize information.
Eight in 10 North American builders surveyed use AI to assess and plan proposed projects. The tools can also scan construction drawings for errors or missing details that might otherwise trigger expensive corrections. Estimating models, Fultineer said, can compare a proposed job with hundreds of completed similar projects to calculate material, labor and equipment costs. Those comparisons may help teams spot gaps earlier, though the usefulness of an output still depends on the quality and relevance of the information behind it.
“In recruiting, it is critical to focus on team members who will be able to manage and progress AI-type applications in the future,” Fultineer said. That shift puts a premium on employees who can work with these tools and apply their output to project decisions. The role is less about accepting a machine’s answer than knowing when to question it.
Spending is ahead of full integration
Investment has been substantial: more than two-thirds of North American respondents spent at least $250,000 on AI in the previous two years, and 37% spent $1 million or more. Yet just 15% said they had fully integrated AI across projects and operations. The gap between spending and organization-wide use points to implementation hurdles beyond software procurement. This is more significant than it looks: a tool can be funded and tested without becoming part of the routines that guide a project.
Forty percent cited a shortage of skilled talent or expertise as a barrier. Another 43% raised concerns about liability for AI-generated errors. Both issues are practical. Teams need people who can assess outputs, and companies need clarity about how those outputs are checked before they influence estimates, schedules or other project work.
Fultineer said training needs to fit each stage of construction and become part of regular practice, with senior leaders encouraging continued use. A one-time introduction is unlikely to settle questions that arise as a project moves from planning into execution. Here's the thing: adoption depends on whether staff can use the tools in the work already underway, not simply whether they have access to them.
Oversight remains a weak point
AI was active in most core functions for 42% of construction respondents, close to BRG’s figures for retail at 48% and financial institutions at 54%. Construction had the lowest policy coverage among the four sectors surveyed for human oversight and accountability, at 42%, and for accuracy and reliability, at 34%. Fultineer linked the gap in part to stronger regulatory pressure on banking and energy companies to formalize oversight.
That comparison points to a gap in documented controls, though it doesn't by itself show how carefully any individual firm checks its tools. Construction work often involves information moving among owners, designers, contractors and vendors, so a mistake can travel through several decisions before it becomes visible. (And this is the part most people overlook.) Clear responsibility for reviewing AI output matters as much as the tool's ability to produce it.
What adoption will depend on
For contractors, the practical test is whether AI improves project outcomes. Fultineer recommends tracking estimating accuracy, schedule performance, rework and change orders, while applying quality checks as teams use the tools on active projects. Those measures connect adoption to familiar project concerns and can help firms distinguish useful applications from activity that merely looks like progress.
If you're working in this space, the survey suggests the next challenge is organizational: matching training, oversight and responsibility to the tasks where AI is already being used. And yet the reported investment and limited full integration show that spending alone hasn't settled that challenge. The next phase of adoption will depend on building that discipline into everyday work. Still a work in progress.
