There is a big difference between teaching offline and online. In an offline, in-room situation, the teacher gets immediate feedback about students when he/she sees the face of the student and responds accordingly and that’s what makes teaching effective and interesting. Whereas when students are remote and learning through an app, the estimate of students’ involvement and nudge based on the student’s state of mind is a key missing element. When a student uses an app, a lot of events are generated. Data comes from accelerometers and clicks. Each click and accelerometer data is in turn associated with what is there on the page and the current mental state of the student. All these events also have a time sequence. All these factors prompted us to choose a Kalman filter approach to solve the problem. With page content as input, events as output, and the mental state of the student as an internal state, a Klaman filter has been designed to estimate the students’ current state. The estimate of the student’s state is used to decide the next item in the course of study. In turn, achieving a student-centric teaching approach that is scalable and economically viable. Think of it as we managed to achieve a 1:1 teacher-to-student ratio. AI wonders!