At the time of the introduction of the Ambient Intelligence (AmI) concept many scenarios where considered to be visionary or even science fiction. Enabled by current technology, many aspects of these scenarios are slowly but inexorably becoming true. However, we are still facing important challenges that need further investments in research and industrialization. Current software engineering techniques and tools are not prepared to deal with the development of applications for what we could call AmI ecosystems, lacking a fixed architecture, controlled limits and even owners. The comfortable boundaries of static architectures and well-defined limits and owners are not existent in these AmI ecosystems. In its second year AmI.d again shows the heterogeneity of research challenges related to Ambient Intelligence. Many disciplines are involved and have to co-ordinate their efforts in resolving the strongly related research issues.