Transform research papers into visual stories

1. Ingest: Paste any arXiv paper URL and watch as it's decomposed into digestible sections
2. Analyze: AI agents analyze each section to identify key concepts and visual opportunities
3. Generate: Multi-agent pipeline creates 3Blue1Brown-style Manim animations for complex ideas
4. Validate: Four-stage quality gates ensure syntactic correctness, spatial coherence, and runtime stability
5. Experience: Read through an interactive scrollytelling interface with embedded animated visualizations

• Node.js 18+
• Python 3.11+
• FFmpeg, Cairo, Pango (for Manim)
• API keys for Anthropic Claude, ElevenLabs, Dedalus Labs (free at signup)

cd frontend

npm install

npm run dev

cd backend

cp .env.example .env          # Add your API keys

uv sync

uv run uvicorn main:app --reload --host 0.0.0.0 --port 8000

Research papers arrive as monoliths — dense, opaque, intimidating. Within them lies a mosaic of brilliant ideas waiting to be seen.

arXivisual transforms fragments of academic text into animated visual explanations, making complex research accessible to everyone.