The aim of the present study is to present the results of the assessment of clinical application of the robotic assistant for patients suffering from mild cognitive impairments (MCI) and Alzheimer Disease (AD). The human-robot interaction (HRI) evaluation approach taken within the study is a novelty in the field of social robotics. The proposed assessment of the robotic functionalities are based on end-user perception of attractiveness, usability and potential societal impact of the device. The methods of evaluation applied consist of User Experience Questionnaire (UEQ), AttrakDiff and the societal impact inventory tailored for the project purposes. The prototype version of the Robotic Assistant for MCI patients at Home (RAMCIP) was tested in a semi-controlled environment at the Department of Neurology (Lublin, Poland). Eighteen elderly participants, 10 healthy and 8 MCI, performed everyday tasks and functions facilitated by RAMCIP. The tasks consisted of semi-structuralized scenarios like: medication intake, hazardous events prevention, and social interaction. No differences between the groups of subjects were observed in terms of perceived attractiveness, usability nor-societal impact of the device. The robotic assistant societal impact and attractiveness were highly assessed. The usability of the device was reported as neutral due to the short time of interaction.