🤖 Agent BABA: The First RAG-Enhanced Trading Agent on Solana

Autonomous Trading Agent with Vector-Enhanced Decision Making

🏆 Solana AI Agent Hackathon Submission

Agent BABA represents a breakthrough in autonomous trading by combining cutting-edge AI technologies:

• 🧠 RAG-Enhanced Decision Making: Learns from historical trades through vector embeddings
• 🤝 Autonomous Asset Management: Self-adjusting strategies based on market conditions
• 📈 Real-time Market Analysis: Intelligent price impact and liquidity monitoring
• 🛡️ Vector-Based Risk Management: Pattern recognition for trade optimization

🎥 Demo & Presentation

View Demo Video | View Pitch Deck

🌟 Key Features

1. Advanced Trading AI

• Vector embeddings of historical trades using pgAI
• Semantic search with Ollama (all-minilm model)
• Self-adjusting trading parameters
• Continuous learning from outcomes

2. Institutional-Grade UI

• Real-time performance monitoring
• Advanced charting with TradingView integration
• Trade execution interface
• Risk management dashboard

3. RAG-Powered Analytics

• Vector-based trade pattern analysis
• Semantic similarity search for strategy optimization
• AI-driven market insights
• Real-time performance metrics

💻 Technical Stack

• Frontend: Next.js 13, Shadcn UI, React Query
• Backend: TypeScript, Bun, HONO
• Blockchain: Solana, Meteora DEX
• AI & ML: Claude AI (Anthropic), pgAI Vectorizer
• Data: TimescaleDB, pgVector
• Monitoring: BirdEye API

🏗️ Architecture

flowchart TB
    subgraph On-Chain
        MP[Meteora Pool] <--> SA[Solana Agent Kit]
        SA <--> JE[Jupiter Exchange]
        BE[BirdEye API] --> PS[Price Service]
        PS --> AB
    end

    subgraph Agent BABA Core
        AB[Agent BABA] --> SA
        AB --> DB[(PGAI Vector DB)]
        AB --> CA[Claude AI]
        DB --> RAG[RAG Engine]
        RAG --> CA
        VW[Vectorizer Worker] --> DB
    end

    subgraph Analysis Flow
        DB --> VA[Vector Analysis]
        VA --> TS[Trade Signals]
        TS --> AB
        VW --> |Embeddings| VA
    end

    style Agent BABA Core fill:#f9f,stroke:#333,stroke-width:2px
    style On-Chain fill:#bbf,stroke:#333,stroke-width:2px
    style Analysis Flow fill:#bfb,stroke:#333,stroke-width:2px

Loading

🚀 Quick Start

1. Clone & Install

   git clone https://github.com/yourusername/agent-baba.git
   cd agent-baba
   bun install

2. Configure Environment

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

3. Start Services

   docker-compose up -d        # Start database & vector store
   bun run dev                 # Start development servers

4. Open Dashboard

   • Core Service: http://localhost:3000
   • Dashboard: http://localhost:3001

🔧 Project Structure

agent-baba/
├── packages/
│   ├── core/              # Trading engine & services
│   ├── dashboard/         # Next.js frontend
│   └── types/            # Shared TypeScript types
├── package.json          # Monorepo configuration
└── turbo.json           # Turborepo config

🌅 Future Vision

Agent BABA is designed to evolve into a comprehensive trading platform:

• Multi-DEX support
• Advanced portfolio management
• Social trading features
• Custom strategy development
• Machine learning model integration

🛡️ Security & Risk Management

• Real-time risk monitoring
• Slippage protection
• Position size management
• Automated circuit breakers
• Vector-based anomaly detection

👥 Team

Built with 💜 by Team BABABILL for the Solana AI Agent Hackathon

Role	Member
🎨 Frontend	Fodé Diop
🔧 Backend	Fodé Diop
🧠 AI/ML	Fodé Diop
🔗 Blockchain	Fodé Diop

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

📄 License

MIT License - see LICENSE for details

---

Vote for Agent BABA in the Solana AI Agent Hackathon! 🗳️