The paper is a retrospective look on about a decade long development of the concept map based intelligent knowledge assessment system IKAS. The research and development process is divided into several periods. During the period before the IKAS, research was focused on intelligent tutoring systems (ITS) based on agent paradigm. The beginning of the IKAS is related to studies of concept maps (CM) as teaching, learning, and knowledge assessment tools and definition of basic characteristics of the IKAS: the ability to enable systems thinking, adaptability to learners’ current knowledge level, and the capacity of flexible knowledge assessment. Evolution of five versions of the IKAS, characterized by continuous extension of its functionality, went on in the period of enthusiastic work. Success and failures as well as lessons learnt during this period are described. The current situation with the IKAS and its further development directed at elaboration of improved scoring methods of CMs and formal evaluation of CM complexity is presented.