This article was last updated on July 14, 2026.
AI agents for construction are autonomous agentic AI systems that project teams configure to connect to project data across the systems where that data already lives: Procore and Autodesk Construction Cloud for project management, Primavera P6 for scheduling, and Navisworks, Revit, Revizto, or SYNCHRO for BIM and model coordination, alongside the underlying project files.
AI agents can then execute multi-step workflows such as RFI validation, submittal review, compliance checks, drawing comparison, site-safety review, and project-file search. They can also flag cost, schedule, or quality implications in RFIs.
Contractor optimism about AI runs high, but the adoption data tells a more layered story, separating investment intent from workflows that actually changed, deployments that scaled, data that's governed, and ROI that's been measured.
The statistics in this article point to where AI-agent workflows are worth testing first: RFI volume, rework rates, project-data waste, and platform-scale metrics. Two caveats travel with every number below. ROI needs its own validation on your projects, and safety-critical outputs still need human review, especially in the early weeks when validation rules produce false positives that require tuning.
AI agents for construction adoption rates by survey
Investment intent, active AI use, and scaled deployment are different stages of construction AI maturity. AGC's 2026 outlook reported 61% use AI or plan to increase AI investment. The outlook also reported current AI deployment by function:
Office or administrative tasks: 45%
Estimating: 23%
Design and preconstruction: 20%
HR: 16%
These percentages put office workflows ahead of estimating, design and preconstruction, and HR in current AI deployment.
Investment intent and current AI use
AGC's 2025 outlook reported that 44% planned AI investment increases in 2025. AI was the top-ranked technology investment category.
RICS's Q2 2025 construction report found 15% current adoption of AI in construction. Additional findings included:
Construction organizations exploring AI: 45%
Construction organizations with no AI capability or plans: 29%
These figures separate current AI use from exploration and no-plan status.
Scaled adoption and readiness
KPMG's 2025/2026 global construction survey reported 24% scaled AI adoption globally, based on 375 engineering and construction executives. The survey also reported:
Australian construction firms reporting scaled AI adoption: 43.8%
Respondents citing generative AI as the second technology most likely to change the industry: 43%
Dodge Construction Network's AI for Contractors brief reported that 19% adapted workflows to use AI.
Deloitte's 2026 construction digital adoption survey found 46% using AI or machine learning across construction and engineering businesses in Australia, Hong Kong, India, Japan, Singapore, and Vietnam.
BST Global / ENR reported that 82% expect AI to change AEC from AI in AEC. The report also found:
AEC firms expecting AI-driven change within five years: 78%
AEC firms claiming mature or advanced AI readiness: ~20%
AEC firms citing business-value use-case identification as the top barrier: 91%
These readiness figures show why scaled adoption should be read separately from expectations for change.
Procore platform network and AI agent adoption statistics
Platform scale and integrations
Procore's 2025 annual report listed 17,850 organic accounts as of Dec. 31, 2025, up from 17,088 in 2024. It also listed:
App marketplace integrations: Over 500
App marketplace categories: 25
App marketplace categories include accounting, compliance, drone technology, and workforce management
Those integration counts show the scale of connected systems around Procore project data.
Revenue, products, and workflow AI
Procore's FY 2025 results reported 115 accounts with more than $1M ARR, a 34% year-over-year increase. Other metrics included:
Gross revenue retention: 95%
Net revenue retention: 106%
ARR from customer accounts using four or more products: 78%
ARR from customer accounts using six or more products: 52%
Procore's Q1 2026 results reported $359.283 million in quarterly revenue, up from $310.632 million in Q1 2025. Calculated from those figures, that was 15.6% year-over-year quarterly revenue growth for the three months ended March 31, 2026.
ENR reported that Procore announced 3 workflow areas after the Datagrid from Procore integration: RFI analysis, submittal cross-checking, and compliance auditing.
Procore's Q1 2026 earnings transcript described Datagrid from Procore integration progress as rapidly proceeded, a qualitative milestone from the earnings transcript.
Procore's Q4 2025 earnings transcript reported nearly 450 accounts for Procore Pay at the end of 2025. The transcript also reported 70% year-over-year Procore Pay account growth.
Built world role statistics where AI agents meet the work
Role-focused baselines map AI-agent business cases to the people who feel the workflow burden. Those groups include operations leaders, PMs, estimators, preconstruction teams, field supervisors, BD/proposal teams, and account teams. Some statistics are role-specific; others are role-relevant baselines instead of individual-role hour breakdowns.
Workflow burden by role
PlanGrid/FMI's Construction Disconnected report found 35% working hours spent on non-optimal activities in 2018. The report also found:
Across surveyed construction roles: More than 14 hours lost per person each week to non-optimal activities, equivalent to almost two working days
Across surveyed construction roles: 5.5 hours per week spent looking for project data or information
Across surveyed construction roles: 4.7 hours per week spent on conflict resolution
Across surveyed construction roles: 3.9 hours per week spent dealing with mistakes and rework
Construction firms overall: Over $177 billion in U.S. labor cost of non-optimal activities
Poor communication and poor project data: $31.3 billion combined annual cost
The burden data points to search, coordination, and rework as workflow targets for AI-agent testing.
Coordination and governed project data
Dodge Construction Network reported 33% coordination issues as the root cause of quality challenges among contractors in 2024.
FMI reported that 55% have data plans, a baseline for account and client teams whose service consistency depends on project data being governed and reusable.
RFI, rework, project-data, and overrun baselines
RFIs and rework
Navigant / CMAA's RFI impact study found 796 RFIs per average project, based on 1,362 projects and 1.1 million RFIs. The study also found:
RFIs per $1 million of construction cost: 9.9
RFIs per $1 million on $5M–$50M projects: 17.2
Average time per RFI review and response: 8 hours
Median administrative cost per RFI hour: $82/hour
Median technical review cost per RFI hour: $188/hour
Average total cost per RFI review and response: $1,080
Estimated total RFI review cost per average project: $859,680
Navigant / CMAA's rework impact study reported 5.04% direct rework cost on an original contract cost basis. The study also reported:
Median total rework cost including indirect costs: 9.07%
Project delay attributed to rework: 52%
Rework-attributable schedule growth: 9.82%
ASCE / Love reported 0.38% precompletion rework cost in January 2026, based on actual contractor cost data and contract value. Additional figures included:
Rework cost including postcompletion corrections: 0.76%
Underreported actual rework costs in contractor records: 300%
These rework figures give a conservative counterpoint to older RFI and rework impact baselines.
Bad data, disputes, and overruns
Autodesk/FMI reported 14% bad-data rework in 2020. The report also found:
Avoidable rework cost from bad data: $88.69 billion
Respondents saying 20%–50% of project data is bad: 33%
Respondents saying more than half of project data is bad: 30%
FMI modeled a $7.1 million avoidable rework cost from bad data for a contractor with $1B revenue.
Flyvbjerg / Oxford reported 47.9% on budget from a 2023 database of 16,000+ projects across 136 countries. The dataset also reported:
Projects delivered on budget and on time: 8.5%
HKA CRUX's 2025 report found 33.4% disputed costs as a share of contract budgets globally, based on 2,200+ projects with US$2.433 trillion in value. It also found:
Claimed extensions of time as a share of works schedules globally: 66%
HKA's North America analysis found 58.6% time extensions as a share of planned schedules across 405 U.S. and 91 Canadian projects. The analysis also found:
Disputed costs as a share of CapEx: 33.4%
MDPI Buildings reported 85% overrun budgets globally in a 2025 peer-reviewed literature review. The review also reported:
Average global budget overrun: 28%
McKinsey reported that large projects finish 20% longer than scheduled and exceed budget by up to 80%.
Construction jobsite safety statistics and AI evidence
Government safety baselines
BLS reported 1,034 fatal injuries in private construction in 2024, down 3.8% from 1,075 in 2023.
The BLS CFOI summary reported 1,032 fatalities among construction and extraction workers in 2024. The summary also reported:
Fatal falls, slips, and trips among construction and extraction workers: 370
Decrease in fatal falls, slips, and trips: 7.5%
OSHA reported 389 fatal falls to a lower level in construction, out of 1,034 construction fatalities in 2024.
OSHA reported 20% federal reduction in fatal falls after the fall emphasis program, from 234 to 189. The same program period reported:
State OSHA fall fatality reduction: 15%
BLS reported 30.0 cases per 10,000 FTE workers for nonfatal construction falls, slips, and trips requiring at least one day away from work, compared with 22.6 across all private industry.
BLS industry-rate data listed 1.8 per 100 FTE workers as the 2024 TRC rate for heavy and civil engineering construction. The table also listed:
Highway, street, and bridge construction TRC rate: 2.5 per 100 FTE workers
Specialty trade contractors TRC rate: 2.3 per 100 FTE workers
The rate differences keep fall risk and civil-work incident rates in the safety baseline.
AI safety evidence and limits
Deloitte's 2026 engineering and construction outlook disclosed no specific percentage for safety-focused computer vision incident reduction, while noting adoption.
MDPI's 2025 systematic review covered 148 peer-reviewed articles on AI in construction health and safety from 2013–2025 and disclosed no meta-analysis percentage for incident reduction, while supporting monitoring and warning uses.
ENR reported 97%–100% compliance on Arrowsight projects from contractor-supplied field data. The report also included:
Reported workers' compensation claims decrease: 72%
Limitation was contractor-supplied without independent audit
CPWR's 2025 small-study funding cycle showed no AI-specific priority for construction safety research, and independently audited AI safety percentages remain scarce.
AI in construction and agentic AI market forecasts
Definitions of "AI in construction" vary, so base-year values vary widely.
AI in construction forecasts
Research and Markets states a $2.28 billion 2025 base market size for AI in construction in its 2026 report. They also forecast:
2030 forecast market size: $9.48 billion
Forecast CAGR: 33.1%
Mordor Intelligence estimated an $11.1 billion 2025 base market size for AI in construction. Its estimate also included:
2031 forecast market size: $27.92 billion
Forecast CAGR: 16.62%
Fortune Business Insights estimated a $4.86 billion 2025 base market size for AI in construction. The forecast included:
2034 forecast market size: $35.53 billion
Forecast CAGR: 24.80%
Grand View Research estimated a $2.93 billion 2023 base market size for AI in construction. The forecast also included:
2030 forecast market size: $16.96 billion
SNS Insider estimated a $6.48 billion 2026 base market size for AI in construction. Its forecast included:
2035 forecast market size: $53.32 billion
Forecast CAGR: 26.38%
Market Research Future estimated a $0.67 billion 2024 base market size for AI in construction. The projection also included:
Forecast CAGR: 32.66%
Forecast period to 2035
Precedence Research estimated a $404.63 million 2025 base market size for generative AI in construction. Its forecast included:
2035 forecast market size: ~$7.65 billion
Forecast CAGR: 34.18%
The spread across these forecasts reflects differences in base year, scope, and market definition.
Agentic AI and AI-agent forecasts
MarketsandMarkets forecast a $6.76 billion 2025 base market size for enterprise agentic AI. The forecast also included:
2030 forecast market size: $46.04 billion
Forecast CAGR: 47%
MarketsandMarkets forecast a $7.06 billion 2025 base market size for agentic AI. The forecast included:
2032 forecast market size: $93.20 billion
Forecast CAGR: 44.6%
MarketsandMarkets forecast a $7.84 billion 2025 base market size for AI agents. Its forecast included:
2030 forecast market size: $52.62 billion
Forecast CAGR: 46.3%
Grand View Research forecast a $24.50 billion 2030 market size for enterprise agentic AI. The forecast also included:
Forecast CAGR: 46.2%
Mordor Intelligence estimated a $6.96 billion 2025 base market size for agentic AI. Its forecast included:
2031 forecast market size: $57.42 billion
Forecast CAGR: 42.14%
These agentic AI and AI-agent forecasts show high growth expectations outside construction-specific market definitions.
Enterprise agent adoption forecasts
Gartner's August 2025 prediction forecast 40% of enterprise apps featuring task-specific AI agents by 2026, up from less than 5% in 2025.
Gartner's June 2025 agentic AI forecast projected 33% of enterprise software applications including agentic AI by 2028, up from less than 1% in 2024. The same forecast included:
Agentic-AI-driven enterprise software revenue by 2028: Over $450 billion
Work decisions made autonomously through agentic AI by 2028: 15%
Agentic AI projects expected to be canceled by the end of 2027: Over 40%
A separate Gartner multi-agent forecast projected 70% of AI apps using multi-agent systems by 2028.
Gartner's May–June 2025 survey of IT leaders found 15% of IT application leaders considering, piloting, or deploying fully autonomous AI agents.
Construction labor and productivity baselines
Labor and productivity baselines explain why project teams are considering AI agents for more operating capacity while labor, productivity, and hiring constraints persist.
Workforce constraints
ABC estimated 349,000 workers were needed in 2026 to meet construction demand.
ABC's 2025 workforce-shortage coverage projected 439,000 workers needed in 2025.
ABC projected 456,000 new workers needed in 2027 as spending growth resumes.
AGC's 2026 outlook release reported 82% craft difficulty filling hourly craft positions. The release also reported:
Firms reporting difficulty filling salaried openings: 80%
Firms expecting to add headcount in 2026: 63%
Firms expecting to decrease headcount in 2026: 15%
AGC / NCCER's 2025 workforce survey reported 92% filling difficulty for open positions.
AGC reported workforce shortages as the leading delay cause, with no percentage disclosed.
AGC's January 2026 employment digest reported 8,303,000 employed in construction in December 2025, down 11,000 from November and up 14,000 year over year. The digest also reported:
Year-over-year construction employment growth: 0.2%
ABC reported the slowest hiring rate on record for construction in February 2026, since BLS began the survey in December 2000.
Productivity gap
McKinsey reported 1% annual growth in construction labor productivity over two decades. The report also stated:
Total world economy productivity growth over the same period: 2.8% annually
Manufacturing productivity growth over the same period: 3.6% annually
Construction firms matching overall-economy productivity growth: Less than 25%
Opportunity from closing construction's productivity gap: $1.6 trillion
McKinsey's full productivity report stated that productivity growth in manufacturing, retail, and agriculture since 1945 reached up to 1,500%, a cross-sector comparison for construction productivity.
Performance claims and limitations to keep in view
Outcome percentages from vendor studies, pilots, and broad cross-industry research need separate scrutiny when independently audited construction datasets are unavailable.
Enterprise AI value signals
McKinsey estimated a $90–$150 billion potential generative AI productivity impact in construction. The same estimate was 1.4%–2.3% of construction revenue.
Deloitte's 2026 enterprise AI survey reported 66% productivity gains from enterprise AI across industries. The survey also reported:
Organizations reporting cost reduction from enterprise AI: 40%
Organizations reporting improved insights and decision-making from enterprise AI: 53%
These cross-industry gains provide context, while construction-specific ROI still needs separate measurement.
McKinsey's September 2025 value-paradox report found 80% using AI, while the same 80% reported no significant topline or bottom-line gains from the latest generation of AI. The figure is cross-industry and shows the adoption-to-value gap.
McKinsey's 2026 organizations survey found 47% expect reduction in administrative work from AI. The survey also reported:
Leaders expecting AI to bring exponential productivity: 55%
C-suite respondents describing generative AI rollouts as mature: 1%
Respondents reporting AI-accelerated revenue increases above 5%: 19%
The survey results reinforce the gap between AI rollout maturity and measured revenue acceleration.
Construction-specific caution
MIT Sloan's 2025 study found a 26% task increase from GitHub Copilot access among software developers. The study also found:
Junior developer task-completion gains from GitHub Copilot access: 27%–39%
Senior developer task-completion gains from GitHub Copilot access: 8%–13%
Limitation was a software-developer proxy for construction workflows
Treat the Copilot study as productivity context and a software-developer proxy for construction workflows.
AI agents can automate complex procedures such as RFI validation, submittal cross-checking, compliance auditing, drawing comparison, site-safety review, and project-file search. The strongest public construction data still measures adoption, workflow burden, project-data waste, and platform scale more reliably than achieved ROI percentages.
For operators, the practical sequence is narrow. Connect Procore, Autodesk Construction Cloud, Primavera P6, drawings, specs, RFIs, and submittals. Run an AI agent against one governed workflow. Review false positives. Expand the playbook after PMs trust the exceptions it flags.
