Table of Contents
Fetching ...

Composable Prompting Workspaces for Creative Writing: Exploration and Iteration Using Dynamic Widgets

Rifat Mehreen Amin, Oliver Hans Kühle, Daniel Buschek, Andreas Butz

TL;DR

PromptCanvas addresses the bottleneck of static prompting interfaces by introducing a composable prompting workspace built from dynamic widgets on an infinite canvas. The system converts prompts into interactive, persistent widgets, enabling structured exploration, iterative refinement, and metacognitive support in creative writing. A within-subjects study with 18 participants shows significantly higher creativity support and lower mental demand and frustration compared with a baseline conversational UI, indicating meaningful improvements in user experience and efficiency. The work demonstrates the value of user-driven customization for prompting AI in writing and points to cross-domain extensions, such as visual content generation, where dynamic widgets can further empower creative workflows.

Abstract

Generative AI models offer many possibilities for text creation and transformation. Current graphical user interfaces (GUIs) for prompting them lack support for iterative exploration, as they do not represent prompts as actionable interface objects. We propose the concept of a composable prompting canvas for text exploration and iteration using dynamic widgets. Users generate widgets through system suggestions, prompting, or manually to capture task-relevant facets that affect the generated text. In a comparative study with a baseline (conversational UI), 18 participants worked on two writing tasks, creating diverse prompting environments with custom widgets and spatial layouts. They reported having more control over the generated text and preferred our system over the baseline. Our design significantly outperformed the baseline on the Creativity Support Index, and participants felt the results were worth the effort. This work highlights the need for GUIs that support user-driven customization and (re-)structuring to increase both the flexibility and efficiency of prompting.

Composable Prompting Workspaces for Creative Writing: Exploration and Iteration Using Dynamic Widgets

TL;DR

PromptCanvas addresses the bottleneck of static prompting interfaces by introducing a composable prompting workspace built from dynamic widgets on an infinite canvas. The system converts prompts into interactive, persistent widgets, enabling structured exploration, iterative refinement, and metacognitive support in creative writing. A within-subjects study with 18 participants shows significantly higher creativity support and lower mental demand and frustration compared with a baseline conversational UI, indicating meaningful improvements in user experience and efficiency. The work demonstrates the value of user-driven customization for prompting AI in writing and points to cross-domain extensions, such as visual content generation, where dynamic widgets can further empower creative workflows.

Abstract

Generative AI models offer many possibilities for text creation and transformation. Current graphical user interfaces (GUIs) for prompting them lack support for iterative exploration, as they do not represent prompts as actionable interface objects. We propose the concept of a composable prompting canvas for text exploration and iteration using dynamic widgets. Users generate widgets through system suggestions, prompting, or manually to capture task-relevant facets that affect the generated text. In a comparative study with a baseline (conversational UI), 18 participants worked on two writing tasks, creating diverse prompting environments with custom widgets and spatial layouts. They reported having more control over the generated text and preferred our system over the baseline. Our design significantly outperformed the baseline on the Creativity Support Index, and participants felt the results were worth the effort. This work highlights the need for GUIs that support user-driven customization and (re-)structuring to increase both the flexibility and efficiency of prompting.

Paper Structure

This paper contains 35 sections, 8 figures, 2 tables.

Figures (8)

  • Figure 1: Self-reported creativity support scores and preferences comparing PromptCanvas and the Conversational UI (N=18).
  • Figure 2: Self-reported NASA-TLX scores and ease-of-use ratings from participants in our lab study (N=18).
  • Figure 3: Self-reported cognitive load and preference scores comparing PromptCanvas and the conversational UI (N=18).
  • Figure 4: (1) Marina writes a prompt for the short story generation. (2) PromptCanvas generates widgets for her. (3) She chooses two widgets from the widget panel and drags them onto the canvas.
  • Figure 5: (1) Marina prompts in the widget panel to get more widgets. (2) She creates an empty widget on the canvas.
  • ...and 3 more figures