Sensor information including acceleration, audio, image, light level, location, magnetic field, orientation, proximity, temperature, and video information are all available in modern mobile devices. Nunchi transparently samples these sensors and transmits certain information to one or more remote servers for processing. Nunchi servers compile sensor information through time and construct basic template models of user’s needs through the actions and reactions to the activities and interactions represented by the current and historic sensor data.

The first obvious question is how much can really be determined about a user’s activities and requirements -- the answer is plenty. It’s difficult to explain how the data is actually processed but let’s use a simple descriptive example for illustration. Two brothers are riding their bikes home from school. Nunchi identifies the rhythmic swaying of the bikes as they pedal as well as their average speed. Nunchi also sees that both boys are on their normal path home. Nunchi can see the boys are bumping into each other, and at one point in the ride the youngest falls off of his bike. When he quickly gets back on his bike and starts pedaling rapidly, Nunchi sees that the fall is directly related to the boys’ roughhousing, but since the boy is not injured Nunchi classifies the event as a non-emergency.