Joerg Hiller
Mar 05, 2026 13:38
Oracle’s new Advisor for Safety uses predictive AI trained on 10,000+ project-years of data to forecast construction incidents before they happen.
Oracle rolled out its Construction and Engineering Advisor for Safety on March 5, 2026, bringing predictive AI to an industry that still largely reacts to accidents after they happen. The company claims early customers have seen incident rates drop by up to 50% and workers’ compensation costs fall by as much as 75% in year one.
The pitch is straightforward: instead of waiting for someone to get hurt, the system forecasts which projects are most likely to have safety incidents in the coming week. Oracle says its model can identify the top 20% of projects that typically account for 80% of safety incidents, letting firms concentrate resources where they matter most.
What Makes This Different
Most predictive safety tools require companies to feed them years of proprietary data before they become useful. Oracle built its model on the equivalent of 10,000+ project-years of safety data spanning different project types and locations. That means construction firms can plug in and get predictions immediately, regardless of how mature their existing safety programs are.
The system pulls from multiple sources—safety observations, incident reports, payroll data, project schedules—aggregating information from Oracle Aconex, Primavera Unifier Accelerator, and Oracle Fusion Cloud ERP. Third-party systems can feed into it too. Over time, the model fine-tunes itself using each customer’s specific data.
New Observation Tools for Field Teams
Alongside the AI launch, Oracle introduced an Observation capability in Aconex and Primavera Unifier Accelerator. Field workers from project engineers to senior executives can now capture structured safety data—including severity and frequency scores—through mobile devices or web browsers. Standardizing this data collection improves the predictive model’s accuracy while reinforcing safer behaviors on site.
“Advisor for Safety marks a significant step forward in safety management, giving construction companies and owners the tools to predict and prevent incidents,” said Mark Webster, senior vice president and general manager of Oracle Infrastructure Industries.
The Business Case
Construction remains one of the most dangerous industries. Beyond the human cost, workplace injuries carry substantial financial weight through compensation claims, project delays, and regulatory penalties. Oracle’s positioning here targets both concerns simultaneously.
The company’s AI push in construction safety builds on its 2020 acquisition of Newmetrix, which developed the ‘Vinnie’ safety prediction platform. This latest release represents a more integrated approach, tying predictive capabilities directly into Oracle’s broader construction management ecosystem.
Oracle will showcase the technology at its Customer Edge Summit in Austin, Texas, scheduled for April 12-14, 2026. For enterprise software investors watching Oracle’s AI strategy, construction safety represents a vertical where predictive analytics can demonstrate clear, measurable ROI—something harder to prove in more abstract AI applications.
Image source: Shutterstock