Case Study: Suffolk achieves predictive safety alerts and $1.4M-$3.6M in avoided incident costs with Smartvid.io

How Suffolk & Smartvid.io learned to predict and prevent construction incidents

Suffolk, a national building construction firm, wanted to move beyond observing safety issues to predicting incidents before they occurred. To do this it partnered with Smartvid.io and its AI engine “Vinnie” as part of Suffolk’s “Safer Together” program, applying Smartvid.io’s machine‑learning tools to project photos and construction management data to surface leading safety indicators like PPE compliance.

Smartvid.io ingested ten years of photos and project data from Procore, Autodesk BIM 360 and OxBlue, used Vinnie to detect risks and build an early‑warning model, and validated it on three years of unseen data. The model predicted 20% of incidents at 80% accuracy (about 4 alerts/project/year with one false alarm), or up to 40% at 66% accuracy (≈12 alerts/project/year); with conservative assumptions Suffolk estimated $1.4M–$3.6M in annual safety savings for a 50‑project portfolio (rising if more alerts prevent incidents).

Suffolk

Alex Hall

EVP, Environmental Health & Safety