DashChat: Interactive Authoring of Performance Dashboard Design Prototypes through Conversation with LLM-Powered Agents

Performance dashboards are designed for and deployed within industrial settings to showcase and monitor their operational performance. In practice, the ideation and negotiation phases demand rapid prototyping and iteration to align with evolving client needs. However, existing tools compel designers to compromise either on iteration speed or on the meticulous handling of visual complexities. Addressing the gaps, we introduce DashChat, an interactive system for generating performance dashboard prototypes through conversation. We first derived common design patterns by analyzing 114 high-quality dashboards. We then developed a novel multi-agent pipeline that translates natural language intentions into high-quality prototypes. We evaluated DashChat through a user study with 28 participants, demonstrating its effectiveness in facilitating the prototyping process while ensuring design quality. Our work contributes not only an understanding of the challenges in real-world performance dashboard design workflows, but also a novel conversational system that embodies a new paradigm for AI-assisted creative work.