Self-adaptive software aims at anticipating changes which occur in a complex environment and to automatically deal with them at run-time. The increasing demand for complex networked software, which makes computing resources available to anyone, anywhere and at any time, is drawing attention to the engineering of self-adaptive software. The objective of our work is to define a process and a tool-supported design framework to develop self-adaptive systems, which consider Belief-Desire-Intention agent models as reference architectures. We adopt an agent-oriented approach, which allows to explicitly model system goals in requirements specifica-tion and in the system architecture design. Moreover, goal achieve-ment conditions are specified along with their relationships with the environment and with possible failures, and corresponding recov-ery actions. This paper aims at motivating and giving an overview of our approach with the help of an example.