Show HN: Figma alternative where AI works with vector primitives, not code(brilliant.design)
brilliant.design
Show HN: Figma alternative where AI works with vector primitives, not code
https://brilliant.design/
5 comments
Hi HN, we're the team behind Brilliant.
Brilliant is a free Figma alternative where AI works with vector primitives, not code. Everything the AI produces is a real, fully editable vector element. Auto layout, components, vector editing, effects, masks, boolean ops, gradients, the whole surface area you'd expect (plus shaders and some other fun stuff figma doesn't offer).
What makes it different:
* The AI works with vector primitives, not code. It edits the same nodes, fills, frames, and auto-layout containers you do. Nothing compiles, nothing renders to a preview pane. Anything it produces you can grab and override at the node level.
* Many agents can work in parallel across different frames and across different design files. Each agent has it's own undo/redo stack so you continue working freely while they operate.
* Free unlimited MCP usage. Anything and everything you can do by hand can be done via MCP.
* Free forever. Export to PNG, JPEG, WebP, MOV, MP4 and even a cool replay for showing off your work. No feature whatsoever is gated.
* How do we make money? We offer an integrated AI chat. It's BYOK so you have zero markup on the underlying inference. You pay for the UX of the AI chat and the incredible context management. Teams also get centralized billing, seat management, dashboards etc.
* Files are a YAML-based .design format. Designs go in git, diffs are readable, branches and reviews work the way they should have all along.
A bit aobut us: we're Amichai & Lior, married with 2 children. Left Microsoft after ~5 years and are working on Brilliant full-time for about a year and a half. 100% bootstrapped, spending literally all our life savings with the irrational belief that the experience we're building will be superior to any UI/UX design tool ever in history.
We're stoked and nervous about the release. We'd love to get your feedback on: 1. Where the AI chat falls over for you. Prompts that didn't land, things you expected to work. 2. Performance issues on large and heavy canvases. 3. Anything whatsoever missing that would actually move the needle on switching.
We have both a built in Figma importer (simply paste a link to the figma file you want to import) and a Figma plugin that allows you to copy paste your Figma designs directly to Brilliant.
Download: https://brilliant.design
Manifesto, if you want the longer version of why we built this: https://brilliant.design/manifesto
We'll be here in the comments all day.
Brilliant is a free Figma alternative where AI works with vector primitives, not code. Everything the AI produces is a real, fully editable vector element. Auto layout, components, vector editing, effects, masks, boolean ops, gradients, the whole surface area you'd expect (plus shaders and some other fun stuff figma doesn't offer).
What makes it different:
* The AI works with vector primitives, not code. It edits the same nodes, fills, frames, and auto-layout containers you do. Nothing compiles, nothing renders to a preview pane. Anything it produces you can grab and override at the node level.
* Many agents can work in parallel across different frames and across different design files. Each agent has it's own undo/redo stack so you continue working freely while they operate.
* Free unlimited MCP usage. Anything and everything you can do by hand can be done via MCP.
* Free forever. Export to PNG, JPEG, WebP, MOV, MP4 and even a cool replay for showing off your work. No feature whatsoever is gated.
* How do we make money? We offer an integrated AI chat. It's BYOK so you have zero markup on the underlying inference. You pay for the UX of the AI chat and the incredible context management. Teams also get centralized billing, seat management, dashboards etc.
* Files are a YAML-based .design format. Designs go in git, diffs are readable, branches and reviews work the way they should have all along.
A bit aobut us: we're Amichai & Lior, married with 2 children. Left Microsoft after ~5 years and are working on Brilliant full-time for about a year and a half. 100% bootstrapped, spending literally all our life savings with the irrational belief that the experience we're building will be superior to any UI/UX design tool ever in history.
We're stoked and nervous about the release. We'd love to get your feedback on: 1. Where the AI chat falls over for you. Prompts that didn't land, things you expected to work. 2. Performance issues on large and heavy canvases. 3. Anything whatsoever missing that would actually move the needle on switching.
We have both a built in Figma importer (simply paste a link to the figma file you want to import) and a Figma plugin that allows you to copy paste your Figma designs directly to Brilliant.
Download: https://brilliant.design
Manifesto, if you want the longer version of why we built this: https://brilliant.design/manifesto
We'll be here in the comments all day.
> Old images get dropped > A screenshot attached four turns ago can be 50,000 tokens. It is not helping the agent decide what to do now. Once an image falls behind your last two user turns, Brilliant strips it and leaves a short [Image cleared] stub in its place. The recent ones stay, because those are the ones that matter for the next move. Zero user effort, zero config.
Ok the context gets smaller but doesn't it invalidate the LLM cache?