Scan a product. Log how your skin reacted. Find out which ingredients are the problem.
Most ingredient scanners stop at "here's what's in it." SkinSync goes one step further — it learns your skin from your history, and tells you which specific ingredients are likely behind your bad reactions. Deterministically, with the receipts.
-
Scan and see. Point your phone at a barcode and get the full ingredient list, with sourced safety flags — every flag cites an official list (e.g. EU Cosmetics Regulation allergens), never our opinion.
-
Keep a diary. Add products to your shelf and record how your skin got on — loved it, broke me out, went red. Reactions, severity, notes.
-
It figures out your skin. Once you've logged a few things, it ranks the ingredients most likely causing trouble — and always shows why: "every product that broke you out had this exact thing in it; none of the ones you liked did."
The personal side is the point. Scanners that just list ingredients already exist — the diary plus "it learns your skin specifically" is what doesn't.
- Sourced, never editorial. We don't call things "toxic." We cite what official regulatory lists actually say, and link the source.
- Deterministic and explainable. The tolerance engine is careful comparison and maths — not a black box. It can always justify a verdict.
- Privacy by design. Skin-reaction data is health data. We keep it minimal, ask permission clearly, and let anyone export or delete everything, anytime. (Privacy policy + DPIA live alongside the code.)
- Cheap to run. AI is reserved for the wordy jobs (label OCR, plain-English explanations), gated behind a toggle and off by default. The core loop costs next to nothing.
A PWA-first Rails 8 app — add it to your home screen and it feels native; proper iOS/Android come later, once the web version proves the loop.
Rails 8 · Hotwire (Turbo + Stimulus) · Tailwind CSS v4 ·
PostgreSQL + pgvector · Solid Queue / Cache / Cable (no Redis) ·
Docker · Prometheus + Grafana · Cloudflare Tunnel
Early and active — the core scan → log → insight loop is the current focus. Built and tested on a private homelab first, then opened up.