🚀 AgentFlow - AI Virtual Office Platform

Transform your startup idea into a complete business plan with AI agents working as your virtual team.

✨ Final MVP Features

🧠 8 Specialized AI Agents with Auto-Coordination

• 🎯 Cofounder Agent - Vision analysis and strategy
• 📋 Manager Agent - Auto task assignment and coordination
• 🛠️ Product Agent - MVP definition and user personas
• 💰 Finance Agent - Financial projections and pricing
• 📈 Marketing Agent - Marketing strategy and campaigns
• ⚖️ Legal Agent - Compliance and legal requirements
• 💼 Sales Agent - Sales strategy and forecasting
• 🔧 Operations Agent - Operations planning

🤝 Auto-Coordination System

• Manager automatically assigns tasks based on context
• Agents communicate and share insights with each other
• Coordinated execution with dependency management
• Real-time execution logs and status tracking

🔐 Authentication & Database

• Supabase authentication with demo mode fallback
• User projects and conversation persistence
• Agent outputs saved to database
• Row-level security for multi-tenant support

🧠 Advanced Memory Systems

• Neo4j Graph Memory - Agent relationships and context
• Qdrant Vector Memory - Semantic search and RAG
• Redis State Management - Real-time conversation state

🚀 Quick Start

1. Environment Setup

# Copy environment template
cp .env.example .env

# Edit with your API keys (only OPENROUTER_API_KEY required)
nano .env

2. Backend Setup

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

3. Frontend Setup

cd frontend
npm install
npm run dev

4. Access AgentFlow

• 🌐 Frontend: http://localhost:5173
• 🔧 API: http://localhost:8000
• 📚 API Docs: http://localhost:8000/docs

🎯 How to Use

Method 1: Auto-Execute (Recommended)

1. Visit http://localhost:5173
2. Sign up/Sign in (or use demo mode)
3. Start conversation with AI Cofounder
4. Approve vision → Auto-execution starts
5. Watch agents coordinate in real-time
6. Get comprehensive business analysis

Method 2: API Integration

# Start conversation
response = requests.post('http://localhost:8000/api/conversation/start', 
    json={'message': 'I want to create a productivity app'})

# Approve and auto-execute
requests.post(f'http://localhost:8000/api/conversation/{conversation_id}/approve')

# Watch execution logs
logs = requests.get('http://localhost:8000/api/agents/logs/live').json()

📊 Key API Endpoints

Authentication

• POST /api/auth/signup - User registration
• POST /api/auth/signin - User login
• GET /api/auth/user - Get current user

Auto-Coordination

• POST /api/auto-execute - Auto-execute project
• GET /api/auto-execute/status - Get execution status
• GET /api/agents/logs/live - Live execution logs

Projects & Data

• GET /api/projects - User's projects
• GET /api/outputs - Agent outputs
• GET /api/analytics/predictions - Predictive analytics

🏗️ Architecture

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   React Frontend │────│   FastAPI Backend │────│  Memory Systems │
│   • Auth UI      │    │   • Auto-Coord   │    │  • Neo4j Graph  │
│   • Live Logs    │    │   • Agent Comm    │    │  • Qdrant Vector│
│   • Real-time    │    │   • LLM Integration│    │  • Redis State  │
└─────────────────┘    └──────────────────┘    └─────────────────┘
                                │
                       ┌────────┴────────┐
                       │   8 AI Agents   │
                       │  • Auto-Coord   │
                       │  • Inter-Comm   │
                       │  • LangGraph    │
                       └─────────────────┘

🔧 Configuration

Required

• OPENROUTER_API_KEY - Get free key at OpenRouter.ai

Optional (Demo Mode Available)

• SUPABASE_URL & SUPABASE_ANON_KEY - Database
• NEO4J_URI & credentials - Graph memory
• QDRANT_URL & QDRANT_API_KEY - Vector memory
• REDIS_URL - State management

🚀 Production Deployment

Backend

gunicorn main:app --workers 4 --worker-class uvicorn.workers.UvicornWorker

Frontend

npm run build
serve -s dist

Database Setup

1. Create Supabase project
2. Run database/schema.sql in SQL editor
3. Add environment variables

📈 Current Status

✅ 100% Complete - Production Ready MVP

• 8 specialized agents with auto-coordination
• Real-time agent communication and execution
• Authentication and database integration
• Advanced memory systems with RAG
• Professional UI with live execution logs
• Comprehensive business analysis outputs
• Demo mode for easy testing

🎯 Success Metrics

• Technical Excellence: Full-stack AI platform with auto-coordination
• Business Value: Complete startup analysis with agent collaboration
• User Experience: Real-time visibility into AI agent workflows
• Scalability: Multi-tenant architecture with Supabase
• Innovation: First AI platform with true agent auto-coordination

🤝 Contributing

AgentFlow demonstrates advanced AI engineering patterns:

• LangGraph-based agent orchestration
• Auto-coordination and inter-agent communication
• Multi-modal memory systems with RAG
• Production-ready authentication and database
• Real-time execution visibility

📄 License

MIT License - Built for portfolio demonstration and commercial use.

---

Ready to see AI agents coordinate and build your business plan automatically? Start AgentFlow and watch the magic happen! 🚀