We have conducted a large RCT at the pan-African level aiming to shed light on the impact of peer effects on innovation and entrepreneurship. It involved around 5000 entrepreneurs (some established, other just aspiring) from more than 40 African countries. All of them completed an online business course, while the treatment involved the possibility of interacting with peers (within groups of sixty), in three different setups: (a) face-to-face, (b) virtually “within” (i.e. all peers in the group being from the same country), (c) virtually across (peers spanning balanced country heterogeneity). Virtual interaction was mediated by a flexible and structured chatting platform that the agents could adapt to their needs. After two and a half months, all participants were asked to submit business proposals, which were evaluated by business professionals and senior investors for possible funding. This evaluation — conceived as a measure of entrepreneurship quality — defined one of our key outcome variables. In this respect, the main result is that virtual interaction within has positive effect on such quality, both in the extensive and intensive margin. Instead, if the interaction is face-to-face or virtual across countries, it has no statistically significant effect on quality but does have it on the rate of submission: positive in the former case and negative in the latter. To understand these results better, we turn to the rich panel-data available on the pattern of networking and the content of communication and combine the econometric, network and semantic analysis into an integrated study of the problem. In the end, our conclusion is that virtual peer interaction, suitably designed, can be a powerful and cost-effective manner of promoting entrepreneurship and hence economic development.