I’ve used quite a few AI design tools, but I was always thinking about what might happen when I stop generating a fresh screen and start working on a real design.
Generating a nice-looking landing page from a prompt has become easy. Importing an existing design, editing it without breaking its visual language, asking an AI agent to understand the canvas, creating another page that follows the same design system, and eventually getting that design into a real React workflow is a better test, I think.
That’s what I wanted to do with Brilliant.design.
For this hands-on review, I decided not to start with a blank canvas. I used a real developer portfolio I picked from Figma and brought it into Brilliant. From there, I worked through the editor manually, connected OpenAI Codex through MCP, asked Codex to inspect and modify the design, created a second canvas from the existing design system, tested Brilliant’s built-in AI with Codex, and tried its Playground mode.
I also looked at the parts of Brilliant that matter to developers beyond the visual editor, including its design system, Blueprint language, MCP workflow, and design-to-code options.
