This paper presents what is, to the best of our knowledge, the very first successful attempt at running AgentSpeak(L) programs. AgentSpeak(L)is a programming language for BDI agents, created by Rao, with which he pointed for the first time towards bridging the gap between BDI logics and implemented BDI systems. Moreover, it has quite an elegant and neat notation for a BDI programming language, which could establish a turning point in the practice of implementing cognitive multi-agent systems, should it be turned into a working interpreter or compiler. Precisely because such (implemented) interpreter or compiler was unavailable, AgentSpeak(L)has been neglected, as have other agent-oriented programming languages with a strong theoretical support, by multi-agent system practitioners. This paper shows a way of turning AgentSpeak(L)agents into running programs within Sloman’s SIM_AGENT toolkit. We have called this prototype interpreter SIM_Speak, and we have tested it with a multi-agent traffic simulation. We also discuss the limitations and possible extensions to SIM_Speak.