A personalized measure for thermal comfort has been applied for use in combination with smart controls for building automation. Using data from a field study, we first show the superiority of personalized measures for thermal comfort compared to standard non-adaptive methods. Based on this knowledge we describe a methodology, using logistic regression techniques, to convert user votes to a probability of comfort. We also describe the interface used to collect the votes. We show that, for a given subject, our thermal profile converges against the probabilities found in the field study. As a case study we implemented the measure in a control algorithm to control the shading devices. The results clarify the mode of action and also show the effectiveness of the method.