MedMatchAI
A travel health PWA that helps people find safe medication equivalents when they're abroad.
the problem
When you travel, your medications often don't. A British traveller in Spain looking for Nurofen, an American in France searching for Tylenol, a Turk in Germany trying to find Parol — they're all looking for the same active ingredient under a different brand name. Get the matching wrong and the consequences range from "I bought the wrong thing" to "I just took something I'm allergic to."
Most existing tools either don't address this at all, or treat it as a simple lookup problem. It isn't. It's a brand-name-to-generic-name reconciliation problem across an unbounded global catalog, where AI can plausibly map most cases and be confidently wrong on the rest.
What I'm building
MedMatchAI is a mobile-first PWA designed for iOS Safari, backed by 47,000+ medication records sourced from regulatory authorities across the US, UK, EU, Turkey, and Canada (FDA, MHRA, ANSM, AEMPS, BfArM, AIFA, and others — every record attributed to its source).
The interesting design work isn't the search box. It's everything around it: how the app communicates uncertainty, how it prevents AI suggestions from quietly overriding a user's allergy data, how it routes high-risk matches to a pharmacist instead of resolving them silently.
I designed and built the whole thing myself — design, front-end, data pipelines, deployment — using AI tooling throughout the process.
The case study (in progress)
I'm writing the long-form case study now. The throughline: what it actually takes to integrate AI into a high-stakes domain when the cost of being wrong is real, and AI being confidently wrong is the failure mode you're designing against — not just AI being inaccurate.
Coming to this page when ready. If you want to follow along or talk about the work in the meantime, I'm at https://www.linkedin.com/in/cara-ceren-sarper/
Trey the Live app
It's in active development — expect things to change week to week. Feedback welcome.