What we learned, what we're betting on next — and one decision we need this room to make.
Dear community, contributors, and investors,
This year the industry answered a question we'd circled for two years: can an LLM actually design, not merely describe design? Claude Design proved yes in April — then locked that capability behind a closed, cloud-only subscription. Our answer is the opposite bet, and it's the one we're asking this room to ratify again: keep the full design workflow open-source, local-first, and agent-agnostic, and let the community — not one vendor — own the roadmap.
That is not sentiment. It shows up in retention, contribution velocity, and the unit economics on the next few pages.
GitHub Stars
76K+
Apache-2.0. Forkable and self-hostable — zero vendor lock-in for any enterprise that adopts it.
Contributors
340+
Up from 210 a year ago — code, plugins, and design systems, not just issues.
The clearest evidence · GitHub Stars
76K+
340+ contributors · 260+ plugins · 150 design systems
Source: public GitHub data · github.com/nexu-io/open-design · Apache-2.0 open source.
The first commitment is the desktop. Design happens where files already live — local folders, Figma exports, and code repositories are readable in place, and the agent has the full power of the terminal. The second is agent-agnostic: Claude Code, Codex, and Cursor already on a contributor's machine are strong enough. We don't rebuild them — we wire them into one complete design workflow.
The third is the hardest, and the one that compounds: it learns you over time. Every choice settles into design systems, preferences, and memory, so the next generation lands closer to what a team meant. That compounding is why retention, not acquisition, is the number the board should watch.
Coding Agents
24
Auto-detects CLIs on your PATH. Zero config.
Design Systems
150
Imported from GitHub, Figma, or local files.
Monthly Cost
−80%
vs. Claude Design: $500 → ~$78.
The curve itself is not the surprise — its makeup is: 340+ contributors, 260+ plugins, 150 design systems. We have no growth team; the community is the growth team, and that's exactly the risk on the next page.
Midpoints are illustrative; the endpoint is the current public GitHub figure of 76K+ stars.
Our biggest risk was never losing users to a closed competitor. It's under-investing in the open pipe that lets 24 agents plug in — if that pipe breaks, the whole bet breaks with it.
— Maya Ortiz, Head of Platform Engineering · Open Design
GitHub Stars
76K+
Apache-2.0 open source
Contributors
340+
From a worldwide community
Plugins
260+
Prototypes, slides, posters, dashboards
Design Systems
150
Import from GitHub / Figma / local
Coding Agents
24
Auto-detects your PATH, zero config
Monthly Cost
−80%
vs. Claude Design, $500 → ~$78
The alternative we rejected was a closed, in-house model — higher margin on paper, but it would trade this entire ecosystem for one vendor's roadmap.
We are asking this room to approve three things: a $3.2M community-grants and infrastructure budget for the next fiscal year, keeping the local-inference path free for every contributor; renewing the Apache-2.0 license commitment unchanged, so no future round can quietly re-close the roadmap; and ratifying neutral terms with agent vendors — we integrate with Claude Code, Codex, and Cursor alike, and take no exclusivity payment from any one of them.
There are too many people to thank individually. Thank you to the 340+ contributors who left code, plugins, and design systems in the repository, and to every one of you who voted with a star, an issue, or a PR. In the year ahead we do exactly one thing: keep the open bet funded, and keep the roadmap in the community's hands.
Open · Local · Agent-native — accountable, named, and dated.
Founder/CEO — license & roadmap · Head of Platform — agent integrations · Head of Community — grants program
Vote requested by June 30 · github.com/nexu-io/open-design · discord.gg/mHAjSMV6gz