The Shift from Skeptic to Builder

Brett Williams, a veteran visual designer, long held a skeptical view of AI, fearing it would devalue the craft of design. However, after observing the industry's shift, he decided to experiment with building his own product, Gather, a Mac app for collecting design inspiration. His journey highlights a critical realization: AI tools, when wielded by someone with strong design judgment, do not replace the designer but rather act as an extension of their intent.

Rethinking the Role of Figma

Contrary to his initial assumption that he would need to design everything in Figma first, Brett found that he spent 80% of his time in the final 20% of the build process. He used Figma primarily for specific tasks—such as extracting CSS from Dev Mode for complex button styles or experimenting with high-fidelity aesthetics—but the majority of the app's structure and functionality was built directly through iterative prompting in Claude. He discovered that his ability to describe design language (strokes, opacity, spacing) was sufficient to guide the AI, often bypassing the need for a full Figma prototype.

Maintaining Control Through Iteration

Brett emphasizes that the "loss of control" often cited by critics is a misconception. He found that by breaking the build into granular phases—starting with the backend, then the frontend, then individual UI elements—he could maintain precise control over the outcome. He avoids "one-shot" prompts, preferring to drill into specific components one by one. This iterative approach allowed him to handle complex features like dark mode, multi-select, and free-flow canvases, which he notes are significantly more difficult to implement than they appear from the outside.

The Reality of Shipping

Brett spent roughly 100-150 hours building Gather, a stark contrast to the "build it in an hour" hype often seen on social media. He stresses that the true challenge of building a product isn't just the initial code generation, but navigating the edge cases, state management, and the technical debt that accumulates as features are added. His experience underscores that while AI lowers the barrier to entry, the "last mile" of polish and reliability still requires significant effort and a clear product vision.

Key Takeaways

  • Designers have a superpower: Your ability to describe visual intent (opacity, hierarchy, spacing) is a technical skill that AI models can interpret effectively.
  • Iterate in small chunks: Don't ask for an entire app at once. Build the backend, then the frontend, then drill into specific UI components one by one.
  • Figma is a tool, not a requirement: Use Figma for high-fidelity experimentation or extracting CSS, but don't feel obligated to design every pixel before writing code.
  • Respect the edge cases: The difference between a demo and a product is the time spent handling state, dark mode, and user interaction edge cases.
  • Focus on the 'why': Your value as a builder comes from your taste and your ability to solve a specific problem, not just the ability to write code.

Notable Quotes

  • "These tools in the hand of someone with good taste and good judgment is just crazy."
  • "I was shocked by... the control piece and being able to... there was no part of this app that I was not able to build exactly the way I wanted it to build."
  • "I think my life I feel like it's much easier to copy a tool one for one than like start with a blank slate... and then like have to think through all these edge cases."
  • "I don't really know how to talk about these things... but I'm doing it."