Turn your creator niche into a fully structured content strategy — streamed live, visualized on Miro.

Built for LovHack Season 2 · Track: 🎨 Creators and Digital Communities

Solo creators — YouTubers, LinkedIn builders, newsletter writers, indie hackers — know what they want to make but have no structured system for strategy and execution. They have scattered ideas, no content pillars, no posting plan, and no brand consistency.

CreatorOS fixes that in under 2 minutes.

Fill in your niche, audience, tone, and goals. CreatorOS streams a full content strategy word by word using Llama 3.3 70B, parses it into structured pillars and post ideas, and automatically builds a color-coded Miro board you can act on immediately.

• Live Strategy Streaming — watch your content strategy generate word by word via Server-Sent Events
• 4 Content Pillars — color-coded, each with 3 post ideas (hook, format, summary)
• Auto Miro Board — programmatically creates a hero banner, pillar columns, and idea cards in your Miro workspace
• AI Content Coach — click any post idea to open a modal with a built-in chat to develop the idea further
• Notes + Links per Idea — save scripts, Notion links, YouTube references, or any URL per card (persisted to localStorage)
• Model Fallback Chain — if primary model is at capacity, automatically retries through 5 models

User fills intake form
        ↓
Next.js API Route (/api/generate)
        ↓                    ↓
Featherless AI          SSE stream to frontend
(Llama 3.3 70B)         (tokens → pillar cards appear)
        ↓
Zod validation + retry
        ↓
"complete" event → client creates Miro board
        ↓
Miro REST API (shapes + cards + frames)
        ↓
Board link returned → opens in new tab

Prerequisites: Node.js 18+, a Featherless AI account, a Miro developer account

git clone https://github.com/your-username/creatoros
cd creatoros
npm install

Create .env.local:

FEATHERLESS_API_KEY=your_featherless_api_key
MIRO_ACCESS_TOKEN=your_miro_oauth_token
NEXT_PUBLIC_MIRO_ACCESS_TOKEN=your_miro_oauth_token
NEXT_PUBLIC_BASE_URL=http://localhost:3000

npm run dev

Open http://localhost:3000

Featherless AI

1. Sign up at featherless.ai
2. Dashboard → API Keys → Create key
3. Connect your HuggingFace account to unlock Llama 3.3 70B (Settings → Integrations)

Miro

1. Sign up at miro.com (free)
2. Go to developers.miro.com → Create new app
3. Enable boards:read + boards:write scopes
4. Click "Install app and get OAuth token" → copy the token

src/
├── app/
│   ├── page.tsx                  # Intake form
│   ├── generate/page.tsx         # Strategy generation + streaming display
│   └── api/
│       ├── generate/route.ts     # SSE streaming + LLM orchestration
│       └── chat/route.ts         # Per-idea AI coach chat
├── components/
│   ├── PillarCard.tsx            # Color-coded pillar + post idea cards
│   ├── PostIdeaModal.tsx         # Expandable modal with notes, links, chat
│   ├── StepIndicator.tsx         # Progress steps UI
│   └── MiroLink.tsx              # Miro board link component
└── lib/
    ├── featherless.ts            # Featherless AI client + model chain
    ├── schema.ts                 # Zod schemas for strategy data
    ├── prompt.ts                 # System prompt builder
    ├── miro.ts                   # Server-side Miro utility (local dev)
    └── miro-client.ts            # Client-side Miro board creation

Pooja · LovHack Season 2 · March 2026

Sponsors: Featherless AI · Miro · DevSwarm · Vercel