Design

Using LLMs to generate and refine UX design directions

Use LLMs as a day-to-day UX design assistant to generate initial UX/UI ideas, refine drafts, synthesize research, and iterate on directions while maintaining a clear design-thinking loop (generate → refine → redesign → repeat).

Why the human is still essential here

The designer must define the problem, set constraints, and apply judgment/taste to avoid iterating quickly on the wrong direction; AI accelerates iteration and drafting but doesn’t replace critical thinking or UX ownership.

How people use this

User flow & IA ideation

Use an LLM to propose multiple user flows, information architectures, and edge cases for a feature, then the designer selects and refines the most viable direction.

ChatGPT / Claude

Rapid wireframe variations from prompts

Generate and iterate low-fidelity wireframe/layout options from a short product brief to compare alternative screen structures before rebuilding the chosen version in design tools.

Uizard / Galileo AI

UI style and component direction boards

Create quick visual style explorations (color, typography, UI patterns) to inform a redesign direction and translate the selected elements into a Figma-based component library.

Midjourney / Figma

Research note synthesis

An LLM clusters interview or usability test notes into themes and opportunities for the designer to validate and refine.

Claude / ChatGPT

UX microcopy variants

AI proposes multiple button labels, error messages, and empty-state copy options aligned to a given tone and constraints.

Gemini / ChatGPT

Heuristic critique prompt

The designer describes a flow or shares key screen content and AI returns a heuristic-based critique checklist to review for issues.

Claude / ChatGPT

Community stories (2)

I couldn’t care less about AI

To be fair I use AI everyday in my design process, I pay Gemini and Claude, I have built apps with Claude Code and Figma MCP. AI is useful and impressive, but I miss the good old days when we were just designers focusing on the user experience. I feel that AI is turning companies into complete chaos. Making PMs feel that they can design the final experience just prompting mediocre UIs, making CEOs think designers are not needed, and wanting to turn designers into semi-developers and product managers to prove their value, because now “anybody can design”. Now we have a bunch of people in the organization jumping right into the solution, building mediocre and inconsistent user experience and forgetting completely about the process to understand the problem we’re trying to solve.

TheNuProgrammerUX designer

Mar 5, 2026

Daily routine of a “10x designer”👇

Daily routine of a “10x designer”👇

Generate. Refine. Redesign. Repeat.

If that’s the definition, then I’m a 10x designer too.

I’ve been using LLMs in my workflow for years. I even became an AI advocate at my full time job at Upwork

But here’s what I’ve learned:

AI speed often hides shallow thinking.

If the problem is wrong, you’re just iterating faster in the wrong direction.

Execution is cheap.
Thinking is expensive.

And most teams are cutting the wrong cost 🤔

Yanick JimenezSenior Product Designer

Feb 26, 2026