Skip to content

Repository files navigation

Launchable

MIT License TypeScript PRs Welcome

Need tighter visibility, control, and integration of your builds — at scale?

Launchable is an open-source build orchestration platform. Scaffold apps, design visually, get AI-powered feedback, pass quality gates, analyze market readiness, and export production code — all from a single pane of glass.

It's not a design tool. It's not a project manager. It's not an AI chatbot. It's the mission control for shipping apps.

Idea → Scaffold → Design → AI Review → Team Feedback → Quality Gates → Market Analysis → Export Code → Ship
         ↑                                                                                              |
         └──────────────────────── all visible in one dashboard ────────────────────────────────────────┘

The Problem

Building apps today means context-switching between 6+ tools:

Tool What it does What's missing
Figma Design mockups No code output, no feedback gates, no project tracking
Linear/Jira Track tasks Disconnected from the actual design
Slack/email Collect feedback Unstructured, no voting, no gates
Storybook Preview components No design system management, no AI
GitHub Actions CI/CD for code Nothing for design quality
Google Docs Document decisions Disconnected from everything

You have zero single-pane visibility into what's being built, who approved it, whether it's accessible, whether it's competitive, and whether it's ready to ship.

The Solution

Launchable puts everything in one place:

┌─────────────────────────────────────────────────────────────┐
│                    COMMAND CENTER                            │
│                                                             │
│  ┌─ Screens ──┐  ┌─ Tasks ────┐  ┌─ Design System ───────┐ │
│  │ Home    ✓  │  │ To Do   3  │  │ Colors  ████████████  │ │
│  │ Login  👍5 │  │ Active  2  │  │ Fonts   Inter + Lora  │ │
│  │ Dash   👎2 │  │ Review  1  │  │ Spacing ▪▪▪▪▪▪▪▪▪▪▪  │ │
│  │ Cart    ●  │  │ Done    8  │  │ Theme   ◑ Dark        │ │
│  └────────────┘  └────────────┘  └────────────────────────┘ │
│                                                             │
│  ┌─ Pipeline ─────────────────────────────────────────────┐ │
│  │ Draft ──→ Review ──→ Approved ──→ Export Ready ──→ 🚀  │ │
│  │            ✓ Feedback  ✓ A11y ≥80  ✓ No blockers       │ │
│  └────────────────────────────────────────────────────────┘ │
│                                                             │
│  ┌─ Market Intel ─────┐  ┌─ Activity ──────────────────┐   │
│  │ Score: 74/100      │  │ 2m ago  👍 Alex liked Login  │   │
│  │ vs. Stripe: -11    │  │ 5m ago  💬 "CTA too small"   │   │
│  │ "Add trust badges" │  │ 8m ago  ✓ Task completed     │   │
│  └────────────────────┘  └─────────────────────────────┘   │
└─────────────────────────────────────────────────────────────┘

What Launchable Does

1. Scaffold → full app structure in seconds

Pick a preset (SaaS, E-Commerce, Social, Dashboard, Health) or describe your app to AI. Launchable generates the complete app structure: screens, navigation flows, data models, and a starter design — ready to customize.

2. Design → visual canvas with AI co-designer

Drag-and-drop components onto a Konva.js canvas. Edit properties, tokens, variants. Or tell the AI: "Add a login form with Google sign-in" and watch it build it live via function calling.

3. AI Review → Quorum iterative refinement

6 AI design personas review your work one at a time, each making a single targeted improvement. Accept or reject each change. The design converges on quality through iteration — not parallel chaos.

4. Team Feedback → structured voting on everything

Like, dislike, or comment on any element — screens, colors, fonts, components, decisions. Feedback feeds directly into quality gates.

5. Quality Gates → CI/CD pipeline for design

Every screen progresses through: Draft → Review → Approved → Export Ready → Shipped. Gates enforce: accessibility score, positive feedback, team approvals, no blocking comments, no urgent tasks. You can't export code until the design passes.

6. Market Intelligence → competitive analysis

AI analyzes your design against the competitive landscape. Scores across 7 dimensions: category fit, intuitiveness, professionalism, visual quality, UX patterns, info architecture, conversion readiness. Benchmarked against top apps in your space.

7. Export → real code, not mockups

Scene graph → platform-neutral AST → React + Tailwind code. CSS variables. JSON tokens. Quorum attribution baked in.

8. Track → project management built in

Tasks, milestones, Kanban board — all linked to screens and components. No context-switching to Linear. Everything visible in the Command Center.


Quorum Design Refiner

Launchable integrates Quorum for iterative AI design refinement. Instead of dumping options, each AI persona reviews the current design and makes one targeted improvement.

User: "Professional SaaS pricing page, dark theme, trustworthy but modern"

  Pass 1: ♿ Aria (Accessibility)
  ─────────────────────────────────
  Focus: accessibility
  "Increased text contrast from 3.8:1 to 7.2:1 for AAA compliance.
   Darkened secondary text from #9ca3af to #4b5563."

   Before: aesthetics 5.2 | usability 5.8 | a11y 4.1 | consistency 5.5
   After:  aesthetics 5.2 | usability 6.0 | a11y 7.8 | consistency 5.5
   [✓ Accept]


  Pass 2: ◯ Mina (Minimalist)
  ─────────────────────────────────
  Focus: simplification
  "Removed heavy drop shadows in favor of subtle borders.
   Cleaner hierarchy — content breathes."

   Before: aesthetics 5.2 | usability 6.0 | a11y 7.8 | consistency 5.5
   After:  aesthetics 6.8 | usability 6.2 | a11y 7.8 | consistency 6.4
   [✓ Accept]


  Pass 3: ✦ Edith (Editorial)
  ─────────────────────────────────
  Focus: typography
  "Increased heading size contrast from 30px to 36px.
   Better typographic hierarchy tells a clearer story."

   Before: aesthetics 6.8 | usability 6.2 | a11y 7.8 | consistency 6.4
   After:  aesthetics 7.4 | usability 6.8 | a11y 7.8 | consistency 7.0
   [✓ Accept]


  Pass 4: ◆ Blaze (Bold)
  ─────────────────────────────────
  Focus: visual-hierarchy
  "Deepened CTA button color from #3b82f6 to #1d4ed8 for
   stronger visual impact."

   Before: aesthetics 7.4 | usability 6.8 | a11y 7.8 | consistency 7.0
   After:  aesthetics 7.9 | usability 6.8 | a11y 7.6 | consistency 7.0
   [✗ Skip — user preferred the lighter blue]


  ✓ Quorum Review Complete
  ─────────────────────────────────
  3 of 4 improvements accepted
  Overall improvement: +2.3 points

  "Design refined by Quorum AI — 3 iterative passes by
   ♿ Aria, ◯ Mina, ✦ Edith. Powered by Launchable."

The 6 AI Design Personas

Persona Avatar Philosophy Style
Mina Less is more Clean, whitespace, minimal
Blaze Make a statement Bold, contrast, dramatic
Aria Design for everyone Accessible, readable, clear
Edith Typography leads the narrative Editorial, elegant, serif
Pixel Delight is a feature Playful, friendly, colorful
Brix Function IS the aesthetic Brutalist, raw, structural

Design Pipeline (CI/CD for Design)

Draft → Design Review → Approved → Export Ready → Shipped
  │          │              │            │
  │     Feedback open    Approvals    A11y ≥ 80
  │     for voting       met (N/M)   No urgent tasks
  │                      No blocking  No blocking
  │                      comments     comments
  │
  Has components
Gate Required What It Checks
Has Components Yes Screen has at least 1 element
Positive Feedback Yes More likes than dislikes
No Blocking Comments Yes All blocking comments resolved
Team Approvals Yes Required number of approvals met
A11y Score ≥ 80 Yes Accessibility score passes threshold
No Urgent Tasks Yes All urgent tasks for this screen done
Market Readiness ≥ 70 Advisory Competitive analysis score
AAA Compliance Advisory A11y score ≥ 95

Market Intelligence

AI analyzes your design against the competitive landscape:

Dimension What It Measures
Category Fit Does it look like it belongs in this space?
User Intuitiveness Will users understand it without instructions?
Professionalism Does it look trustworthy and polished?
Visual Quality Is it visually competitive with top apps?
UX Patterns Does it follow proven patterns for this category?
Info Architecture Is the information structure clear?
Conversion Ready Would users sign up / buy / engage?

Benchmarked against top apps: Stripe, Linear, Headspace, Shopify, Instagram, Duolingo — depending on your category.

Covers 6 categories: Fintech, Health, SaaS, E-Commerce, Social, Education.


Use Cases

Indie hacker shipping a SaaS MVP Scaffold a dashboard app → customize the design → run Quorum review (Aria catches contrast issues, Mina simplifies the nav) → pass quality gates → export React + Tailwind → ship to Vercel.

Design team replacing Figma-to-dev handoff Design in the visual canvas → collect team feedback with structured voting → enforce accessibility gates before export → generate production code with design tokens baked in. No more "does the dev build match the mockup?"

Agency pitching a client Pick the client's industry category → scaffold the app → run Market Intelligence to benchmark against competitors → present the competitive analysis alongside the prototype. Data-backed design decisions, not opinions.

Open source maintainer building a contributor portal Scaffold the app structure → document every design decision with inline notes → set up quality gates so PRs can't merge without accessibility checks → export components that match the design system exactly.


Quick Start

git clone https://github.com/qinnovates/design-studio.git
cd design-studio
pnpm install

# Optional: start database for save/load
docker compose up -d
pnpm db:generate
pnpm db:push

# Start dev server
pnpm dev
# → http://localhost:3000

Configure AI (BYOAI)

  1. Open the editor → click AI panel → Configure AI
  2. Choose your provider: OpenAI, Anthropic, or Ollama (local, free)
  3. Enter your API key → Test → Save
  4. Start building: "Add a hero section with a signup form"

Your key stays in your browser. Never touches our servers.


Architecture

99 TypeScript source files. 11 packages. 8 Zustand stores. 0 type errors.

launchable/
├── apps/web/               # Next.js 15 + React 19
│   ├── components/editor/  # 25+ editor components
│   │   ├── canvas/         # Konva.js renderer
│   │   ├── CommandCenter   # Single pane of glass dashboard
│   │   ├── DesignArena     # AI swarm (Quorum + Arena modes)
│   │   ├── PipelineStatus  # CI/CD gate status bar
│   │   ├── MarketIntel     # Competitive analysis panel
│   │   └── ...             # Inspector, Toolbar, Fonts, Export, A11y, Notes, Versions, Plugins
│   └── stores/             # 8 Zustand stores (canvas, project, tokens, AI, feedback, PM, swarm, UI)
├── packages/
│   ├── ai/                 # BYOAI providers + Quorum + Pipeline + Market Intelligence
│   ├── canvas/             # Scene graph + node types
│   ├── components/         # 9 UI component definitions
│   ├── tokens/             # Design token resolver
│   ├── app/                # App manifest + scaffolder + fonts
│   ├── export/             # React/CSS code generators
│   ├── a11y/               # WCAG accessibility checks
│   ├── collab/             # Yjs CRDTs + version history
│   ├── plugins/            # Sandboxed iframe plugin runtime
│   ├── db/                 # Prisma schema (12 models)
│   └── ui/                 # Shared utilities
└── docker-compose.yml      # PostgreSQL + Redis

Tech Stack

Next.js 15, React 19, Konva.js, Zustand, Yjs, PostgreSQL, Prisma, Tailwind CSS 4, Zod, BYOAI (OpenAI/Anthropic/Ollama)


What's Built (v0.2.0)

Orchestration: Command Center dashboard, Design Pipeline (CI/CD gates), Market Intelligence (7-dimension competitive scoring), project management (tasks, milestones, Kanban), feedback system (votes + comments on any element), version history, activity feed

Design: Visual canvas (drag/drop, zoom/pan, selection, resize), 9 component definitions, 80+ design tokens with alias chains, light/dark themes, app map (screen relationships), font library (15 fonts, 8 pairings), responsive preview, inline annotations

AI: BYOAI (OpenAI, Anthropic, Ollama), 7 canvas tools via function calling, Quorum iterative refinement (6 personas), Arena parallel generation, market readiness analysis, Zod-validated tool calls

Export: React + Tailwind codegen, CSS variables, JSON tokens, Quorum attribution

Security: BaseUrl allowlist, Zod validation, CSP headers, plugin sandbox, XSS escaping

What's Next

  • Real-time multiplayer cursors
  • SwiftUI / Compose / Flutter export
  • Figma import
  • Plugin marketplace
  • Docker production image
  • Live demo at launchable.dev

Contributing

# Fork → clone → install → branch → code → typecheck → PR
pnpm typecheck  # must pass (23 tasks, 0 errors)

License

MIT — Kevin Qi / qinnovates


Build orchestration powered by Quorum — iterative multi-persona AI review.

About

Launchable — open-source build orchestration platform. Tighter visibility, control, and integration of your builds at scale. AI-powered design, feedback CI/CD, market intelligence, code export.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Used by

Contributors

Languages