Urban crowdsourcing is a new form of open innovation that aims to develop smart cities. Due to the heterogeneity of partners in urban planning, knowing the type of knowledge gained from each stakeholder at each stage of the crowdsourcing process needs to be understood for effective collaborations and innovative knowledge flow activation. Using a qualitative approach with experts involved in the Casablanca Smart City project, this research investigates complex multi-stakeholder ecosystems. More specifically, this study identifies and examines the role of external stakeholders at each stage of the innovation process, and the resulting dynamic knowledge flows that help improve the process of urban crowdsourcing in the development of smart city solutions. Depending on the complexity of the innovation project, findings identify three primary stakeholders (citizens, public authorities, and private and public firms) and three secondary stakeholders (NGOs, universities, consulting companies). Results show at which of the four steps of urban crowdsourcing they intervene. Furthermore, the findings reveal a dynamic learning environment composed of two knowledge flows (“learning with” and “learning from” stakeholders).