Good initiatives for sustainable innovations are everywhere. The biotope selects. How can a conducive biotope be created where such initiatives will flourish and develop into successful innovations? Where can initiators easily find partners and funds to make their dreams come true? What can innovation support agencies do concretely to make a difference? These are the central questions in the European Union Horizon 2020 project AgriSpin. Here, we share our first experiences from AgriSpin, in which 15 partner organisations in 12 European regions are learning from and with each other about successful approaches to innovation brokering. Firstly, we summarise some bottlenecks that are frequently mentioned in the literature. Then the design of the project is described. A key element is the series of ‘cross-visits’ hosted by the partners. At the time of writing, all cross-visits have been made, and the project has entered a digestion period in which we try to make sense of what has been observed. The next step is to design action plans for each partner organisation and the key actors in the regions where they operate. So, this paper reports work in progress. Nevertheless, some interesting ‘pearls’ and ‘puzzles’ can already be reported.

The turn of agrarian sciences and agricultural extension from reductionist and transfer of technology, respectively, towards systemic approaches has transformed agricultural/rural development thinking in the last decades. Nevertheless, the emergence of Agricultural Innovation Systems (AIS) has to confront a number of gaps among which the expert – lay knowledge gap is of major importance. This paper aims at exploring such a gap as well as obstacles to participatory development from a critical realist point of view. Critical realism (CR) with its realist, differentiated and stratified ontology aims at interpreting the world in order to ultimately bring about transformation. CR allows for new insights on the nature of knowledge as well as on development research and practice. It thus provides useful guidelines concerning the emerging ‘intermediation’ functions within AIS.