How do UI designers use Claude in their design workflows?

Why Read This from an HCI Perspective

This article shows how teams are embedding LLMs (large language models) into UI/UX design work—highlighting where real productivity is gained and where expectations may be inflated. From an HCI perspective, what matters less than prompt quality is the provision of context, along with the design system, pipeline, and governance. In other words, practical implications are that the work structure and collaboration model shape the user experience more than the tool itself.

CIT's Commentary

From a CIT perspective, Claude is closer to an assistive intelligence that accelerates design reasoning than to an AI that ‘replaces design.’ The recurring themes in the comments are structure, flow, edge cases, copy, and prototyping—areas that align better with the HCI stages of verification and refinement than with generation alone. In particular, the idea that output quality improves as the design system becomes more sophisticated suggests that AI performs best in environments with clear constraints, rather than in open-ended creative freedom. Conversely, ethical refusals and governance issues are not just matters of personal preference; they are research agendas that must address data provenance, accountability, and ownership within the organization.

Questions to Consider While Reading

• Q.Many people say Claude helps with generating UI ideas but is weaker at final visual design—could this be less about the model’s limitations and more about the input context and evaluation criteria?
• Q.There’s an opinion that AI adoption is more efficient in organizations with well-established design systems—what organizational structure and workflow are most effective from an HCI perspective?
• Q.Generative AI’s convenience and ethical concerns are both being raised—what governance standards should real-world teams have in place to use it safely?