I design with Claude more than Figma now

A designer at Jane Street now relies more on Claude AI than Figma for prototyping and development, marking a significant change in workflow driven by advances in AI capabilities.

The designer reports that recent improvements in AI models, combined with increased familiarity and scope, have made Claude capable of handling complex, large-scale prototypes that previously required manual work in Figma and engineering. Over the past two months, the designer’s use of Claude for creating functional prototypes—such as adding prompting to a JSQL input or implementing new app features—has surpassed reliance on Figma. This shift allows for faster iteration, reducing time spent on mockups and documentation, and directly producing working code that can be tested and refined. The process involves describing the idea, building with Claude, and then iterating quickly, with the final prototype often integrated into the codebase for testing. The approach is seen as empowering for designers, enabling rapid validation of ideas without extensive engineering overhead. However, it also introduces challenges, such as ensuring reviewers can treat prototypes as living proposals and maintaining creative fluidity. The team is experimenting with new review workflows, emphasizing iterative design and shared ownership of features that start as prototypes. The change is still evolving, and the full implications for collaboration and creativity are yet to be determined.

Why It Matters

This shift signifies a potential transformation in how design and development teams work, emphasizing rapid prototyping and iteration through AI. It could lead to faster product development cycles, more experimentation, and a democratization of prototyping, where non-engineers can produce functional ideas. However, it also raises questions about collaboration, review processes, and whether reliance on AI might constrain creative thinking or reduce opportunities for iterative refinement through traditional review and teamwork.

Background

Historically, designers relied heavily on tools like Figma for mockups and documentation, while engineers built prototypes based on these designs. The advent of large language models (LLMs) and AI-driven code generation has begun to blur these boundaries. At Jane Street, the recent adoption of Claude AI has enabled designers to produce working prototypes directly in code, bypassing some traditional steps. This change follows broader trends in AI-assisted development and design, which have seen mixed results in other contexts. The designer’s experience reflects a broader industry exploration of AI’s role in creative workflows, with a focus on efficiency and rapid iteration.

“Through some combination of improved models, my own facility with them, and carefully choosing the right scope, AI is now working for big stuff too.”

— Jane Street Designer

“Prototypes are living proposal docs, the code is disposable, and a reviewer’s job is to give feedback about the design and user experience.”

— Jane Street Designer

What Remains Unclear

It remains unclear how widespread this workflow will become within Jane Street or other organizations, and whether this approach will fully replace traditional design and review processes. The long-term impact on collaboration, creativity, and quality assurance is still being evaluated.

What’s Next

The team plans to continue refining the workflow, developing guidelines for review and iteration, and exploring how AI-generated prototypes can best support collaborative development. Further adoption and feedback will shape whether this becomes a standard practice.

Key Questions

How reliable is Claude for complex prototyping tasks?

While recent improvements have made Claude capable of handling larger, more complex prototypes, the reliability varies depending on scope and scope management. The team is still evaluating its consistency and robustness for production-quality work.

Will designers stop using Figma altogether?

Not necessarily. Figma remains useful for initial concepts, visual design, and collaborative review. The current shift favors rapid prototyping and iteration with AI, but Figma may still be used for certain tasks or final designs.

What are the risks of relying on AI for prototypes?

Potential risks include reduced creative exploration, over-reliance on AI outputs, and challenges in collaborative review. The team is experimenting with workflows to mitigate these issues by emphasizing iterative, shared ownership processes.

How does this change impact collaboration between designers and engineers?

The workflow aims to make prototypes more accessible for engineers to review and iterate on, fostering closer collaboration. However, it requires new practices to ensure feedback remains iterative and design-focused rather than solely code review.

Source: Hacker News