Automated agents should be able to persuade people in the same way people persuade each other, namely via dialog. Today, automated persuasion modeling and investigation use unnatural assumptions in persuasive interactions which create doubt regarding their applicability in real world deployment with people. In this work we present a novel methodology for persuading people through argumentative dialog. Our methodology combines theoretical argumentation modeling, machine learning and Markovian optimization techniques, which together form an innovative agent named SPA. Preliminary field experiments indicate that SPA provides higher levels of attitude change among subjects compared to two baseline agents.