In this paper, we explain one approach for a real-time solution in a mobile device, which would automatically sense and respond to the context of the consumer. This approach uses bag of words to record the state and context of the device. Using the Experience Matrix device will predict the next states, without a use of a schema, processing of bags of words using the Random Index (RI) algorithm. The index forms an associative storage of data that conveys proper data instruction pairs for the next steps or states. It turned out that this was also a very effective approach to provide real time response and almost static storage requirements. Moreover there was no need for training sequences for learning and recording the context. In this article we explain the overall solution architecture, where a dictionary is the model of the world. The context is specialized by inserting the information from mobile sensors to a bag of words, which is used to record and predict events on the mobile device. The Random Index algorithm is used to recall similar events from past. After recall the consumer is selecting an individual way to continue processing from predicted choices and a new state is entered. We would say that this new approach is a base for activity based computing that could potentially replace current application based paradigm in mobile communication devices. As an example we have implemented a predictive browser that uses the event matrix to predict next pages to be navigated into.