My thesis aims to describe and analyse the activities, strategies and experiences of trans and nonbinary people when dealing with the challenges that prompting visual generative AI tools entail. The challenges they encounter opens up to an exploration of the space of friction that occurs between the prompters’ imaginaries and the systems’ representations when the expectations aren’t met. The study is an investigation into the methods that the prompters employ to shape the systems’ output to correspond to their imaginaries. Analysing the activities surrounding the generation of visual synthetic media becomes a site for exposing the representational limitations of the AI models and ultimately what knowledge is perpetuated and made possible by these tools. This allows me to investigate the systems’ underlying infrastructure through the enforced limitations of the AI tools and the hegemonic knowledge production the tools enable.