State-of-the-art: Current Machine Learning and AI solutions deployed in industry are monolithic, highly customised and are designed to operate a single process with minimum process reconfiguration, raw material variations, low material handling complexity, and in majority of cases, fixed position condition monitoring (e.g. surface substrate treatment, cutting). The current solutions normally consist of a highly efficient imaging technology for detecting known normalities and anomalies where a decision tree type cause-effect analytics and linking the evaluation (normally quality parameters) with known process failures reside on top of the quality monitoring apparatus. Such solutions are effective for highly repetitive and low variety-complexity scenarios – confining the systems to moderate levels of state-space definitions. They are prone to regular functional line breakdown/misconception incidents caused by unknown events (un-programmed events).