摘要: Agent-based computing represents an exciting new synthesis for both Artificial Intelligence and, more generally, Computer Science. It has the potential to improve theory and practice of modelling, designing implementing complex computer systems. Yet, date, there been little systematic analysis what makes agent-based approach such appealing powerful computational model. To rectify this situation, paper aims tackle exactly issue. The standpoint is role software in solving complex, realworld problems. In particular, it will be argued that development robust scalable systems requires autonomous agents can complete their objectives while situated a dynamic uncertain environment, engage rich, high-level interactions, operate within flexible organisational structures.