Popinion (Public Opinion) — Don't guess the future. Rehearse it.
Popinion is an AI prediction engine. It creates a "Digital Mirror World" — a simulation of your society where thousands of AI agents react to your ideas. Test a policy, a message, or a strategy here first. See how the world might react. Then act with confidence.
| Feature | Description |
|---|---|
| Smart Scraping | AI-driven search planning with specialized crawlers (Telegram, News, etc.) |
| Ghost Injection | Inject real-world data as a "World Agent" to seed simulations |
| Swarm Simulation | Thousands of autonomous agents with memory, personality, and social behavior |
| Prediction Reports | Deep analysis with agent interviews and trend forecasting |
Input: Upload seed materials (news, reports, social data) + describe your prediction scenario
Output: A detailed prediction report + an interactive digital world you can query
Popinion enables High-Stakes Social Wargaming:
- Policy Crash Testing: Test laws and announcements before they hit headlines
- Diplomatic Save Game: Simulate negotiation outcomes before entering the room
- Cognitive Vaccination: Pre-test counter-narratives against disinformation
From serious predictions to playful simulations, we let every "what if" see its outcome.
- Graph Building: Seed extraction → AI Search Planning → Smart Scraping (Telegram, News, etc.) → GraphRAG construction
- Environment Setup: Entity extraction → Persona generation → Agent configuration
- Simulation: Dual-platform (Twitter + Reddit) parallel evolution with dynamic memory
- Report Generation: ReportAgent with Deep Insight, Panorama Search, and Agent Interview tools
- Deep Interaction: Chat with any agent or query the ReportAgent for custom analysis
Note: Popinion was developed and tested on Mac and Linux. Windows compatibility is experimental.
| Tool | Version | Description | Check Installation |
|---|---|---|---|
| Bun | Latest | Fast JavaScript runtime & package manager | bun -v |
| Python | 3.11+ | Backend runtime | python --version |
| uv | Latest | Python package manager | uv --version |
| Docker | Latest | For Neo4j database | docker --version |
# Start Neo4j using Docker Compose
docker-compose up -d
# Neo4j Browser available at: http://localhost:7474
# Default credentials: neo4j / pubop123# Copy the example configuration file
cp .env.example .env
# Edit the .env file and fill in the required API keysRequired Environment Variables:
# LLM API Configuration (supports any LLM with OpenAI SDK format)
# Recommended: Use a capable model like GPT-4, Claude, or Qwen
LLM_API_KEY=your_api_key
LLM_BASE_URL=https://api.openai.com/v1
LLM_MODEL_NAME=gpt-4
# Neo4j Graph Database Configuration
# Use Docker Compose to start local Neo4j: docker-compose up -d
NEO4J_URI=bolt://localhost:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=pubop123
NEO4J_DATABASE=neo4j# One-click installation of all dependencies (root + frontend + backend)
bun run setup:allOr install step by step:
# Install dependencies (root + frontend)
bun install
# Install Python dependencies (auto-creates virtual environment)
bun run setup:backend# Start both frontend and backend (run from project root)
bun devService URLs:
- Frontend:
http://localhost:3000 - Backend API:
http://localhost:5001
Start Individually:
bun run backend # Start backend only
bun run frontend # Start frontend only- Frontend: Vue 3 + Vite + TypeScript
- Backend: Python + Flask
- Database: Neo4j (Graph Database)
- Simulation: OASIS (Open Agent Social Interaction Simulations)
- LLM Integration: OpenAI-compatible API
Popinion stands on the shoulders of giants:
| Project | Contribution |
|---|---|
| OASIS | Core multi-agent simulation engine (by CAMEL-AI) |
| MiroFish | Social simulation architecture & workflow design |
| BettaFish | Real-world data extraction & scraping framework |
We are grateful to these open-source communities for making Popinion possible.
This project is licensed under the MIT License - see the LICENSE file for details.