In flexographic printing technology, quality is measured mainly through correct depiction of color. Color quality is checked mainly manually, organoleptically by a person, and/or semi-automatically by measuring optical density with a spectrophotometer. Unfortunately, the effects are not immediately visible, only after the product has left the machine. Therefore it is a time-consuming and costly process. The printer is expected to identify perfectly the cause of the error, but that requires thorough analysis of the situation. A quick and reckless decision not only may not solve the problem, but also increase the costs related to further use of materials and production stoppage. Therefore the method is long-running and labor-intensive. In order to improve quality management it is necessary to automate the process. A cognitive agent can be used in this purpose. The aim of the paper is to develop an approach? for cognitive agent’s sensory memory and perceptual memory implementation. The agent will be used for improving quality management of the flexographic printing on packages. Such approach has not been developed before, therefore it is the main contribution of this paper. The main findings are as follows: Due to the use of Hierarchical Temporal Memory the Agent will analyze all parameters and their combinations with other parameters. Changes made by perceptual memory will gradually change the representations in a way that the ‘winning’ columns will be similar in similar signals, however this will occur after many modifications initiated by the learning process. Such approach can improve the quality of prints made and allow for increase in efficiency and decrease in cost of the entire process. It is necessary in the face of growing competition from digital printing. The approach presented in this paper allows for implementation of the remaining modules of LIDA cognitive agent – it is a subject of? future works.?