Nodal Prompting: rethinking authorial practice in Generative AI

The growing integration of Generative AI tools in audiovisual production has turned prompting into the primary interface between human intent and machine output. However, existing theoretical frameworks treat the prompt as a discrete textual artifact, overlooking the increasingly complex interaction models enabled by node-based workflow environments such as ComfyUI, where generative instructions are distributed across interconnected parametric structures. This article proposes the concept of nodal prompting, a mode of human-AI interaction intrinsic to graph-based generative tools, in which authorial intent is wired across a network of nodes, each contributing partial specifications that jointly determine the generative output. The argument is developed through qualitative analysis of a production case study: Il partigiano Libero, an audiovisual series on a partisan during the Italian Resistance, developed by Motion Pixel and ANPI Chiomonte, and situated within a critical review of existing literature on prompt engineering, parametric authorship, and human-AI co-creation. The analysis identifies a constitutive paradox at the core of nodal prompting: the author delegates expressive execution to the machine while simultaneously exercising greater upstream control through system design. This redistribution of creative agency challenges linear models of AI-assisted authorship and suggests that efficiency gains in generative production are conditional on significant investment in workflow architecture. Ultimately, the figure of the author is not dissolved but fundamentally restructured.