.png)
AI integration at Lehigh
Lehigh Construction Group is a Western New York general contractor that self-performs much of its own field labor. That hands-on culture is part of what makes the company different, and it shaped how the team approached AI.
GCM worked with Lehigh to put AI inside the work the team was already doing, from estimates and invoices to manpower logs, safety records, quote templates, project communication, and cost coding.
The goal was not to change how Lehigh works. It was to reduce the friction around the work so the people closest to the projects could spend more time on what matters.
The results
In one engagement, Lehigh put around 20 distinct AI workflows into real use across estimating, accounting, safety, and project management.
What started as eight planned pilots grew into twelve additional workflows the team built organically as staff began seeing where AI could support their day-to-day work.
Key outcomes included:
- ~20 AI workflows in use
- 11–20% time saved per most respondents
- 45–60 minutes saved per day on safety routing
- 8 workflows delivered in the Starter Prompt Library
One respondent reported saving 21–40% of their time, and several team members also reported completing 10–20% more tasks each week.
Where AI made the biggest difference
Estimating and proposals
Estimators built quote letter workflows that pull estimate and subcontractor data into a formatted letter in about ten minutes. The team also built workflows to support first passes at takeoffs, reducing time in preconstruction.
Safety
The safety manager uses a Claude project to score jobsite hazard risk across five categories and build a prioritized daily route in about five minutes. That workflow saves roughly 45 minutes to an hour each day and creates more room for site visits.
The safety team is also using AI to identify OSHA citations from site photos, draft toolbox-talk topics, and generate site-specific safety plans with OSHA language.
Accounting and cost coding
The accounting team built an invoice coder that stamps each invoice with vendor, job, project manager, cost code, and tax status.
They also built an assistant that converts a 46-page gas-card statement into clean Excel, splitting gas and diesel accounts.
Project management and communication
A project manager uses a single Claude project as a living assistant for a major residential job, including monthly budget reconciliation. In one instance, Claude caught a billing discrepancy.
Another team member rebuilt trade reports for a 63-unit project into a clean Excel sheet, work that would have taken weeks by hand.
The team also built a voice-based chief-of-staff workflow that drafts replies and updates a task list, and uses Claude to compare drawing sets.
Adoption beyond the first users
By the final support call, staff across estimating, accounting, safety, field engineering, and project management were using AI in their core work, with several using it daily.
The team also created a voluntary weekly show-and-tell where staff share what is working, what is not, and what they are building next.
That is the deeper signal of adoption: AI was no longer limited to a few power users. The team had started building its own tools.
The bottom line
Lehigh did not add headcount. It gave its team more room to work.
The company competes by self-performing the job and keeping relationships close. AI did not change that. It helped remove the repetitive, time-consuming friction around the work so Lehigh’s team could spend more time applying judgment, solving problems, and moving projects forward.
This is what AI adoption looks like when the training is built around real work.
