In this paper, we propose a C-based approach to develop a Virtual Assistant for the Linux operating system. The goal is to perform tasks with ease so that the user effort is minimum. We are giving input sentences in the form of Natural Language which is processed by our Pattern classifier after which virtual assistant is performing the tasks based on the results of the pattern classifier. Pattern Classifier’s result is the name of a category such as a restaurant, weather, media, etc based on the highest score among all the categories. We are using pattern classifiers, which is much like Multinomial Naive Bayes, the difference is that instead of calculating probabilities this approach is comparing scores of categories which makes this algorithm much simplified. This algorithm is naive because it consists of different independent “features”, in this case: words. Each word is considered as a different entity and does not show any relation with other words in the sentence that is being classified.