The Internet of Things (IoT) is a challenging combination of distribution and heterogeneity. A number of software engineering solutions address those challenges in isolation, but few solutions tackle them in combination, which poses a set of concrete challenges. The ThingML (Internet of Things Modeling Language) approach attempts to address those challenges. This model-driven, generative approach, which was inspired by UML, integrates concepts targeted at the IoT. Over the past six years, it has been continuously evolved and applied to cases in different domains, including a commercial e-health solution.