Android is the most widely used smartphone OS with 82.8% market share in\n2015. It is therefore the most widely targeted system by malware authors.\nResearchers rely on dynamic analysis to extract malware behaviors and often use\nemulators to do so. However, using emulators lead to new issues. Malware may\ndetect emulation and as a result it does not execute the payload to prevent the\nanalysis. Dealing with virtual device evasion is a never-ending war and comes\nwith a non-negligible computation cost. To overcome this state of affairs, we\npropose a system that does not use virtual devices for analysing malware\nbehavior. Glassbox is a functional prototype for the dynamic analysis of\nmalware applications. It executes applications on real devices in a monitored\nand controlled environment. It is a fully automated system that installs, tests\nand extracts features from the application for further analysis. We present the\narchitecture of the platform and we compare it with existing Android dynamic\nanalysis platforms. Lastly, we evaluate the capacity of Glassbox to trigger\napplication behaviors by measuring the average coverage of basic blocks on the\nAndroCoverage dataset. We show that it executes on average 13.52% more basic\nblocks than the Monkey program.\n