Taeef Najib

AgeProof AI turns everyday photos of a home into a home-safety screening an older person can take to their occupational therapist. It's not an image analyzer that asks an AI what looks risky. The language model never decides where a hazard is. SAM 3.1 outlines the objects in the room at pixel level, 13 unit-tested spatial rules measure how they relate (a rug in the walking path, a cord on the floor, no grab rail visible within reach of the toilet), and Muse Spark explains what may matter in plain language. A validator then drops any finding that points at a region that doesn't exist, so every highlight on the photo traces back to a real mask and, for rule-based findings, a real measurement. The result is something a family can act on: numbered hazards marked on their own photos, questions to ask their OT, a checklist of what a photo can't show, and a PDF report for the visit. It knows its limits too. It says a rail is "not visible in this photo" rather than "missing", turns away photos that are too dark or blurry, blurs people automatically, and stores nothing on the server. It never calls a home safe. Built by a data analyst and AI engineer in Australian aged care and tested on 29 labelled room photos (78% precision, 77% recall).