RxAll: the pill scanner that learned to work without a data center
A Nigerian startup builds a handheld scanner that spots fake pills in seconds. When the internet failed during a live demo, the fix reshaped the whole product.
I came across this one in an IEEE Spectrum piece about "small AI," and the origin story stuck with me more than the technology itself. In 2019, Adebayo Alonge was in a Cape Town hotel room, about to demo his startup's device for spotting fake medicine. It didn't work. His server was in the US, and getting the result of a single pill scan took over five minutes.
RxAll's product, the RxScanner, is a handheld spectrometer: point it at a pill, it shines infrared light on it, reads the molecular profile, and matches that against a pharmaceutical database to tell you in seconds whether the drug is genuine. Pharmacies across a dozen countries, including Ghana, Kenya, Myanmar and Nigeria, already use it.
The clever part isn't the spectrometer. It's what happened after the failed demo: Alonge's engineers shrank the AI model down to run entirely on an Android phone, no server, no connection, two hours of work. That constraint became the actual product: a scanner that authenticates medicine in places without broadband, computers, or even reliable electricity.
The stakes make sense of the effort. The counterfeit drug trade in Africa is estimated at $30 billion, and roughly one in ten drugs sold in developing markets is fake. Alonge isn't solving this in the abstract: in 2006 he nearly died after taking pills contaminated with lethal levels of diazepam.
RxAll fits a pattern IEEE Spectrum calls "small AI": models pruned or distilled down to a few megabytes, running on cheap chips, sometimes on solar power, built for the roughly 99 percent of the world's poorest internet users who have never touched a chatbot. It's a quieter story than most AI news this year, and a more useful one: a model small enough to run on a five year old phone, deciding whether the pill in your hand will help you or kill you.