During the late 80's, while most of Aperture employees were focused in the GLaDOS project, a small group of scientists intended to develop a sentient AI, but without the restrictions that were put into GLaDOS; in order to achieve this, they worked in an old abandoned Enrichment Shaft. The result was the Testing Protocol 'Martin', a sentient AI protocol with the primary function of designing, building and testing advanced test chambers, without any bit of 'morality'. Several test subjects died through its tests, which were too hard for anyone to complete. Nobody has been able to reach the end of the testing course, but not only because of the difficulty, but because 'Martin' is constantly building new chambers.