Azumio Inc.

The trick behind Instant Heart Rate was almost too simple. Press a fingertip over the camera lens, let the flash illuminate the skin, and the software watches for the faint color shifts each pulse of blood produces. No strap, no dongle, no clinic. The app landed near the top of the App Store charts and, by the company's own account, passed ten million users - a scale most medical-device makers never reach.

That success became a company. Azumio was founded in 2011 by Peter Kuhar, Bojan Bernard Bostjancic - a former postdoctoral researcher at CERN - and Tom Xu, who had built one of the earliest diabetes-logging apps, Glucose Buddy. The founding bet was that the smartphone was already a capable health instrument, and that most people did not need new hardware so much as better software.

From that premise came a portfolio. Fitness Buddy for workouts. Sleep Time for sleep cycles. Glucose Buddy for blood-glucose logging. In 2013 the company launched Argus, an attempt to gather heart, food, activity and sleep data into a single timeline. Collectively, Azumio says its apps have been downloaded more than 100 million times.

Then, in 2016, the company made a quieter but more consequential move. Calorie Mama applied deep learning to a stubborn problem: looking at a photo of a meal and naming what is on the plate. It is the kind of task humans do without thinking and machines find genuinely hard - a green curry from Thailand does not look like one from Japan, and calories hinge on the difference.

The model is a loop. On one side, Azumio publishes consumer apps monetized through subscriptions and premium features. On the other, it licenses the AI inside those apps - especially food recognition and health-data tooling - to businesses through paid APIs and SDKs. The consumer apps are more than a revenue line: every logged meal and measured pulse feeds the data that sharpens the models sold to enterprise partners.

Where it differs. Plenty of apps count calories. Fewer own the recognition engine underneath, and fewer still sell it. Azumio's distinction is that it built the hard computer-vision core itself and packages it as infrastructure - so a company that would compete with Calorie Mama can instead run on it. That is the same move that put its technology inside Samsung's Bixby.

The real moat is the dataset. Food recognition is not won with a clever model alone; it is won with millions of labeled meals across cuisines, preparations and regional variations. Years of consumer usage gave Azumio that corpus - the unglamorous asset that is hardest for a newcomer to replicate.

Where it sits in the market. On the consumer side, Azumio shares shelf space with MyFitnessPal, Lose It!, Noom and Samsung Health. On the API side, it competes with food-recognition providers such as Passio AI, Clarifai and Nutritionix. Its unusual position is standing in both camps at once - a consumer brand and a quiet supplier to other builders.