We describe the use of a suite of human-centric, domain-specific visual languages to manage knowledge for data-intensive systems. We use two exemplar system case studies – a smart home to support ageing people and a set of smart city technologies – to motivate the need for such an approach. We then describe aspects of these two example systems from abstract requirements to specific data analysis, implementation and deployment choices using our BiDaML representation. We discuss the strengths and limitations of the approach and key directions for further work in this area.