In order to build an intelligent chair capable of posture detection and correction we developed a prototype that gathers the pressure map of the chair's seat pad and backrest and classifies the user posture and changes its conformation. We gathered the pressure maps for eleven standardized postures in order to perform the automatic posture classification, using neural networks. First we tried to find the best parameters for the neural network classification of our data, obtaining an overall classification of around 80% for eleven postures. Those neural networks were exported to a mobile application in order to do real-time classification of those postures. Results showed a real-time classification of around 70% for eleven standardized postures, but we improved the overall classification score to 93.4% when we reduced the posture identification to eight postures, even when this classification was done with unfamiliar users to the posture identification system.