Using hardware platforms combined with other paradigms stimulates learning and has become an apparent trend in education. The Multi-Agent Systems domain studies the use of autonomous entities to solve complex problems. When combined with the Belief-Desire-Intention (BDI), a cognitive model close to human thinking, it can contribute to facilitating the teaching of Artificial Intelligence (AI) and Computational Thinking (CT). This work presents a toolkit to support the teaching of AI supported by BDI intelligent agents using a prototyping approach. Four workshops were held at different levels, and a technological acceptance questionnaire was designed to analyze the proposed toolkit.