We present a human-in-the-loop remote robot supervision platform that enables workers in warehouses and distribution centers to increase their throughput. In our platform humans take a central role in two ways: (i) through their contribution in graceful exception handling and (ii) in improving the machine learning model performance. A preliminary case study shows the feasibility of our approach using data from a deployed pick-and-place system in a production setting. We argue for a more human-centric view of automation that elevates human labor instead of replacing it.