Bitcoin AI Agent

An intelligent Bitcoin assistant that provides real-time price information, answers questions about Bitcoin using the official whitepaper, and maintains conversation history. Built with FastAPI, PostgreSQL, and AI-powered RAG (Retrieval Augmented Generation).

What It Does

• Bitcoin Price Queries - Get current Bitcoin prices in INR with caching
• Knowledge Base Q&A - Answer questions using Bitcoin whitepaper and documents
• Multi-language Support - Understands queries in any language, responds in English
• Conversation History - Maintains chat history for each user
• User Management - Secure authentication and user accounts

How to Use

1. Setup

# Start database and Redis
docker-compose up -d

# Setup database schema
python -c "from bitcoin_agent.db.session import enable_pgvector; enable_pgvector()"
alembic upgrade head

# Load Bitcoin knowledge base
python knowledge_base/scripts/ingest_docs.py

# Start the API server
uvicorn bitcoin_agent.api.app:app --reload

2. Register and Login

# Register a new user
curl -X POST "http://localhost:8000/register" \
  -H "Content-Type: application/json" \
  -d '{"email": "user@example.com", "name": "John Doe", "password": "password123"}'

# Login to get JWT token
curl -X POST "http://localhost:8000/login" \
  -H "Content-Type: application/json" \
  -d '{"email": "user@example.com", "password": "password123"}'

3. Chat with the AI

# Ask about Bitcoin price
curl -X POST "http://localhost:8000/chat" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"message": "What is the current Bitcoin price?"}'

# Ask about Bitcoin concepts
curl -X POST "http://localhost:8000/chat" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"message": "How does Bitcoin mining work?"}'

4. Search Knowledge Base

# Search Bitcoin whitepaper
curl -X GET "http://localhost:8000/search?query=blockchain%20consensus&limit=3" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"

Environment Variables

TOGETHER_API_KEY=your_key
DATABASE_URL=postgresql://user:pass@localhost:5432/bitcoin_agent
REDIS_URL=redis://localhost:6379/0
SECRET_KEY=your_secret_key

Example Usage

Price Queries

User: "What's the current Bitcoin price?"
AI: "The current Bitcoin price is ₹8,19,67,071 INR. This price is updated every 5 minutes with caching to ensure accuracy."

User: "Bitcoin कीमत क्या है?" (Hindi)
AI: "The current Bitcoin price is ₹8,19,67,071 INR. I understand you asked in Hindi, but I respond in English for consistency."

Knowledge Base Queries

User: "How does Bitcoin mining work?"
AI: "Based on the Bitcoin whitepaper, mining works through a proof-of-work system where miners compete to solve cryptographic puzzles..."

User: "What is the purpose of the blockchain?"
AI: "According to the Bitcoin whitepaper, the blockchain serves as a public ledger that records all transactions in chronological order..."

API Endpoints

• POST /register - Create new user account
• POST /login - Authenticate user and get JWT token
• POST /chat - Send message to AI (requires authentication)
• GET /conversations - Get user's conversation history
• GET /search - Search Bitcoin knowledge base
• GET /docs - Interactive API documentation at http://localhost:8000/docs

Tech Stack

• Backend: FastAPI (Python)
• Database: PostgreSQL with pgvector for vector search
• AI Model: LLaMA 3.1 8B via Together AI
• Caching: Redis
• Authentication: JWT tokens
• Vector Search: Sentence Transformers + pgvector