A sophisticated 3D visualization platform for multi-agent research systems with real-time AI-powered research capabilities.
MVA_Demo.-.Made.with.Clipchamp.mp4
Watch the multi-agent system in action: 3D visualization, real-time research, and intelligent agent interactions
- 3D Agent Architecture: Interactive visualization of specialized research agents
- Real-time Research: Integration with Perplexity API and Azure OpenAI
- Intelligent Interactions: Context-aware agent communication flows
- Embedding Space: Dynamic visualization of research embeddings
- Export System: Clean, minimal research result exports
- Dual Architecture: Express server + Astro framework support
-
Install dependencies:
npm install
-
Configure APIs:
cp .env.template env.local # Edit env.local with your API keys: # - AZURE_API_KEY: Your Azure OpenAI key # - AZURE_ENDPOINT: Your Azure endpoint # - PERPLEXITY_API_KEY: Your Perplexity API key
-
Start the server:
node server.js
-
Open your browser:
http://localhost:4321
- Start Research: Enter a topic in the search interface
- Watch Visualization: Real-time 3D agent interactions
- Monitor Embeddings: Top-right panel shows research space
- Export Results: Get clean, minimal research summaries with real API data
- Lead Orchestrator: Coordinates all research activities
- Memory System: Contextual data storage and retrieval
- Tool Agent: API management and resource coordination
- Search Cluster: Parallel information retrieval (Alpha, Beta, Gamma)
- Citations Agent: Source validation and reference management
- Synthesis Agent: Analysis and knowledge integration
- Frontend: Three.js + ES Modules for 3D visualization
- Backend: Node.js + Express + Socket.io for real-time communication
- AI Integration: Azure OpenAI GPT-4.1-nano + Perplexity API
- Framework: Astro (for language diversity)
- Database: SQLite with better-sqlite3
# Express server (main application)
node server.js
# Or Astro dev server (alternative)
npm run dev├── server.js # Main Express server
├── public/
│ ├── index.html # 3D visualization interface
│ └── css/styles.css # Styling
├── src/ # Astro components (for language stats)
├── data/agents.db # SQLite database
└── env.local # API configuration
Research results can be exported as structured data including:
- Search Results: Real-time Perplexity API search results
- Agent Analysis: Specialized AI analysis from multiple perspectives
- Academic Citations: Validated sources and references
- Visualization Statistics: 3D interaction and embedding metrics
- Security: Never commit your
env.localfile - it contains sensitive API keys - Languages: Astro framework included for GitHub language diversity
- Performance: Optimized for real-time 3D rendering and WebSocket communication