AI safety risk scores for a general contractor: Hensel Phelps predicts jobsite incidents before they happen
Hensel Phelps
What they did with AI
Hensel Phelps, a general contracting and construction management company, uses an AI application built around pre-incident indicators (PII) to give each jobsite a forward-looking safety risk profile, according to a case study in Associated Builders and Contractors' 6th Annual Construction Technology Report: Health and Safety, as reported by ABC's Under the Hard Hat on November 9, 2026. Rather than counting incidents after the fact, the tool reviews project data the company already has, with no extra input from project teams, and weighs risk factors such as weather, workforce changes, observation frequency and production pressure. Leaders use the risk score to adjust staffing, add supervision, change work plans or tighten safety controls. The company piloted it in 2025, refined it with field feedback, and has now fully integrated it into its safety management. The report cites earlier risk identification, more targeted actions and better decisions, and says field feedback has been "overwhelmingly positive." Because it scores project conditions rather than individual workers, it keeps blame off the crew; people still decide what to change. The report gives no incident-rate figures and does not name the software; it reads as an in-house tool.
Related use cases
AI field apps for an excavation and concrete contractor: Zemba replaces paper hazard forms and builds a crew scheduler
The Zemba Companies
The Zemba Companies, a construction group working in concrete products, materials, logistics, excavation and environmental services (including residential, industrial and septic work), used AI as a coding partner so its small IT team could design, write and test web apps in weeks instead of months, Executive Director of IT Dr. Joshua J. Tilton wrote in Construction Tech Review. First came safety: paper Task Hazard Analysis forms became a phone form crews fill out before a job, and supervisors can review, export and spot patterns across sites. Next was a crew scheduler with admin, crew and manager views, where a supervisor builds the day's plan and texts or emails the whole crew in one step, replacing phone calls and whiteboard photos. The team is building a phone version of the state pumping report septic drivers file for every load, so the state report and the company's own disposal-cost tracking come from one entry. Zemba also runs "Ask Zemba," an internal AI chat assistant with focused helpers for common questions, has linked payroll, field time, trucking and accounting data that used to be re-keyed by hand, and is building a single onboarding hub. Tilton's rules: start with the person doing the work, keep people reviewing safety forms and approvals, and ship small tools often. No figures were published.
AI construction safety reporting by text: EllisDon's AI Safety Assistant and auto-written toolbox talks
EllisDon
EllisDon, a large general contractor, gave crews on four sites a named AI Safety Assistant from Crewscope with its own phone number and email. Workers text a photo or voice note, the AI logs and classifies it by hazard type, routes it to the right trade by text and tracks it to close-out, with the site team staying in control of what goes on the project record. Crewscope's case study says crews logged about 556 field observations from January to July 2026, 99.5% of them submitted directly by workers, with roughly three positive observations for every problem flagged and most issues closed in under a day. The AI also surfaced a blind spot (electrical hazards were barely reported) and turned positive observations from the site into bespoke toolbox talks, sent by text with a follow-up for a signed attendance list. EllisDon's Doug Van Moorsel, Regional Director of Health, Safety and Environment, is quoted in the case study. This is a vendor-published case study and the figures are Crewscope's; the PDF itself shows both 556 and 566 for the observation count.
AI construction safety software: voice pre-task plans and photo hazard detection for every foreman
Lee Construction Group
Lee Construction Group, a fifth-generation family-owned contractor in Charlottesville, Virginia with 175+ employees across concrete, metal framing and drywall, waterproofing and facilities divisions, had safety paperwork split between paper and a project management system used mostly for storage. It moved its safety program to Field1st: supervisors complete daily pre-task safety forms by voice (Voice1st), AI analyzes jobsite photos to flag hazards and suggest controls, and an in-app AI safety agent answers questions with guidance aligned to OSHA and company policy, with dashboards for leadership. Field1st reports that within the first several months 100% of supervisors were using it for daily pre-task planning, with faster documentation and better hazard closeout. Shelomith Gonzalez, VP of Strategic Operations, is quoted saying supervisors use it daily "because talking through a pre-task form is faster than filling one out." Results are vendor-published and not independently audited.