AbstractThis paper focuses on human-machine communication with intelligent agents, it proposes a generic architecture with an algorithm for natural language (NL) command interpretation which makes it easy to dene dierent applications using the description and domains of the dierent agents, since all that is required is their respective codes and domain ontologies. There are two classical approaches for NL command interpretation: the top-down approach, which relies on the syntactical constraints of the agent's model, and the bottom-up approach which relies on the set of the agent's possible actions. The present work combines the two in a new bottom-up based algorithm that makes use of agent's constraints. the three algorithms are then compared, and results show that the combined approach gives best results. Keywords. Human-Agent communication, introspection, domain ontology, top-down and bottom-up approach, comparative evaluation.