Historically, agroecology was constructed in specific locales through the articulation of professional, political and intellectual spaces: the ‘agricultural professionals’ space, spaces of scientific research (agronomy, biology, ecology, entomology, social sciences) and social movement spaces that are critical (to varying degrees) of the industrialisation of agriculture (Wezel et al., 2009, Abreu et al., 2009, Lamine and Abreu, 2009, Tomich et al., 2011, Francis et al., 2003). This paper, which is based on an ethnography carried out by the first author since 2013, participant observations by all authors in a series of international events, and lexical analysis of technical reports, traces how ‘agroecology’ is co-produced (Jasanoff, 2004) as a global socio-technical object. The site of co-production is the Global Dialogue, convened by the Food and Agriculture Organization of the United Nations (FAO) in different cities around the world between 2014 and 2018 (Rome 2014; Brasilia, Dakar, Bangkok 2015; La Paz, Kunming, Budapest 2016; Tunis, 2017; Rome 2018). We analyze these ‘expert’ symposia by exploring how knowledge about agroecology circulates and frames the terms of debate (Gherardi and Nicolini, 2000, Star, 1999, Callon, 1991). We focus on three key processes that contribute to the stabilization of a global agroecology: 1) the work carried out to define ‘agroecology’, 2) actors’ interests and strategies that are revealed through the politics of circulation, and 3) the emergence of the ‘evidence based’ logic within this dialogue and the ‘experts’ who are legitimized. We argue that the version of ‘agroecology’ that has stabilized through the Global Dialogue is one that has been highly influenced by civil society actors, even though they were not recognized as ‘experts’ in the process. We conclude with reflections upon the politics of ‘agroecological’ knowledge and what this means for the institutionalisations of agroecologies.