Questions?
We’ve got answers.
What is Kombai?
Kombai is your AI design engineer. It helps you design and code standout websites and product UIs that don’t look like “AI slop”. Its taste agent makes high-quality design decisions for each task, instead of resorting to AI defaults. Its context tools find and reuse your existing design system, stack, tokens and components.
Explore designs on an infinite canvas, ship production-ready code in your real repo, and refine the result in an editable browser.
Where is Kombai available?
Kombai is available as a desktop app and an extension for popular IDEs, including VS Code, Cursor, Antigravity, Trae, and Kiro. You can install it directly from these IDEs’ extension marketplaces or download it from the Open VSX Registry to install on any compatible IDE.
Can I import designs from Figma?
Yes. You can import Figma designs into Kombai via its native integration. Kombai can parse complex design files, fix common issues like messy grouping and invisible layers, and extract icons and assets. No MCP needed.
Can I use it with my existing codebase?
Yes. Kombai works as an IDE extension inside your codebase. It can understand the codebase and generate production-ready code that fits natively. Additionally, Kombai can also reuse components, hooks, utilities, and design tokens from your npm packages and storybooks via Context Graphs.
Does it generate designs or working code?
Both. Kombai can generate standalone designs on canvases (in “Design mode”) as well as create production-ready code in your codebase, using your stacks and coding patterns (in “Code mode”).
Can I use Kombai alongside my existing coding agent?
Yes. Many people start with Kombai alongside Cursor, Claude Code, or Codex for design-heavy frontend tasks.
Kombai also has a complete frontend coding agent built in, so you can explore designs, implement them, and review and merge the code, without switching tools.
How does Kombai produce different results with the same AI models?
Kombai’s taste agent develops a design approach for your task, while its context tools identify the components, styles, and code patterns your product already uses. The canvas and editable browser let you compare and refine the result. Those choices help the same model produce a very different outcome.
