Machine learning (ML) and artificial intelligence (AI) have become hot topics in many information processing areas, from chatbots to scientific data analysis. At the same time, there is uncertainty about the possibility of extending predominant ML technologies to become general solutions with continuous learning capabilities. Here, a simple, yet comprehensive, theoretical framework for intelligent systems is presented. A combination of Mirror Compositional Representations (MCR) and a Solution-Critic Loop (SCL) is proposed as a generic approach for different types of problems. A prototype implementation is presented for document comparison using English Wikipedia corpus.