The future that is happening. Automation powered by GPT-4o generates Figma designs based on PRD. @figma

code is poetry.

just integrated this mcp into my demo, works really well

messing around with the local font access api recently. built a little tool to manage local fonts and added an llm for recommendations. it’s mostly just an excuse for me to learn more about typography. will open source it soon. here is a quick look at how it works.

a significant milestone in the evolution of design tools

my dream came true...

totally agree. though i’d argue that while simple prompts now yield results far beyond previous LLMs, achieving truly professional-grade quality still requires professional-level descriptions. for example: "Use GSAP Pinning to lock the screen. As the user scrolls, a 3D soda can rotates, translates, and tilts in the center, showcasing different parts (lid, body, texture) in three phases. dynamic reference lines: coordinated with CSS animations, lines extend from labels to specific parts of the 3D can." btw, i think it’ll be a total game-changer once this can be integrated with MCPs from component communities. i actually dropped "21st.dev/community/component…" into a prompt yesterday. even though gemini didn't explicitly browse or retrieve from the link, i suspect it was part of its training data—the final result featured several components that looked identical to ones from 21st.dev

elegant 🥲 got gemini 3 to code a webpage generator. gemini.google.com/share/25c6…

llm input boxes are blank canvases: max possibility, max friction. users don’t need smarter models—they need smarter starting points

make your figma shine with the Liquid Glass Effect💧