Financial Advisor AI Agent

A comprehensive financial advisor AI system that uses LLMs, multiple specialized agents, and parallel processes to provide real-time cryptocurrency intelligence and advice.

Features

• Core LLM Agent/Orchestrator: Central brain that understands user queries and delegates tasks
• Portfolio Agent: Manages cryptocurrency portfolio tracking and Binance integration
• Query Agent: Handles data analysis, visualization, and natural language queries
• Vector Store: Stores and retrieves financial documents and knowledge
• SQLite Database: Stores structured portfolio data and technical indicators

Setup

1. Clone the repository
2. Install dependencies:

pip install -r requirements.txt

3. Create a .env file with the following variables:

OPENAI_API_KEY=your_openai_api_key
BINANCE_API_KEY=your_binance_api_key
BINANCE_API_SECRET=your_binance_api_secret

4. Create necessary directories:

mkdir data
mkdir chroma_db

Running the Application

Start the FastAPI server:

python main.py

The server will start at http://localhost:8000

API Endpoints

Query Processing

• POST /query: Process financial queries
• GET /portfolio/summary: Get portfolio summary
• POST /documents/upload: Upload financial documents
• GET /documents/search: Search financial documents

Example Queries

1. Portfolio Analysis:

{
    "query": "What's my average return in the last 6 months?",
    "context": {}
}

2. Technical Analysis:

{
    "query": "Show me technical indicators for BTC",
    "context": {}
}

3. Market Data:

{
    "query": "What's the current price of ETH?",
    "context": {}
}

Features

Portfolio Management

• Real-time balance tracking
• Transaction history
• Portfolio performance analysis
• Asset allocation visualization

Technical Analysis

• RSI (Relative Strength Index)
• MACD (Moving Average Convergence Divergence)
• Bollinger Bands
• Historical price data

Data Analysis

• Natural language queries
• Document processing
• Visualization tools
• Market trend analysis

Architecture

The system follows a hub-and-spoke model:

1. Core LLM Agent acts as the orchestrator
2. Specialized agents handle specific tasks
3. Shared memory maintains context
4. Vector store and SQLite database store knowledge and data

Contributing

Feel free to submit issues and enhancement requests.

License

MIT License