Artificial Intelligence aims to create computer programs that display human intelligence features. This can be done in two ways: to create computers that model the human brain, or to treat the computer as a system for manipulating mental symbols. The first way uses neural network and learning. The second way, where we are focused in this paper, uses a formal representation of the world, human-readable representations of problems, logic and search strategies, to provide a human-like intelligence in a machine. So in the levels’ hierarchy of computer systems there must be a new level; level of knowledge. The new challenge of computer scientists is to shift from traditional data processing to knowledge processing. An application of this level is found in expert systems. In this paper we propose that this level might be an agent with its goals, actions and knowledge. The agent processes the knowledge and determines the actions to be performed. In choosing the right action, the agent takes into account the goal. As knowledge is added or modified, the behavior of the agent changes, transforming it into an autonomous and intelligent element.