Swirl-based separators are widely used in the separation of two-phase flows with phases of different densities, e.g. in the oil and gas industry. The separation is based on centrifugal forces associated with the swirl motion, that splits the original mixture into a light-phase core and a heavy-phase annulus, captured by two separate outlets equipped with control valves. Due to the complex (unsteady) dynamics of the flow and variations in the mixture that reaches the separator, real-time controllers are required to keep the separation optimal. Traditionally, the control is based on pressure measurements at the boundaries of the separator. However, pressure is only indirectly related to the separation. A more direct (and potentially more effective) control variable is the distribution of phases inside the separator, which can be monitored using soft-field tomography. This idea is explored in this work using an air-water inline-swirl separator facility equipped with an Electrical Resistance Tomography (ERT) sensor, that determines the size of the air core in real-time using a newly developed fast ERT reconstruction algorithm. The dynamics of the system was studied for different valve actions and summarized into a Transfer Function of the process, which is used to design a PI controller. The approach was successful and the control strategy implemented kept the air core near the setpoint in the presence of disturbances in the air flow rate reaching the separator.