🌐🧠 Crypto & World News AI Assistant with LangGraph

This project is an interactive AI assistant that can answer real-time questions about cryptocurrency prices and world news using tool-augmented language models and LangGraph. It integrates external APIs like GNews and CoinDesk to provide live data, powered by autonomous LLM agents.

---

📌 Project Overview

This AI assistant can:

• 📰 Fetch real-time news headlines for any country using the GNews API
• 💰 Retrieve the current Bitcoin price
• 🤖 Respond to user questions using LLM + tools via LangGraph's agent graph

Built using:

• 🔁 LangGraph (for agent workflows)
• 🛠️ Tool integration (getWorldNews, getBitcoinPrice)
• 🧠 LLM (Groq, OpenAI, etc.)

---

⚙️ Features

• 🔗 LLM-agent connected to real APIs
• 🌍 Country-specific news support (e.g., in, us, jp)
• 🔁 Agent reasoning with tool use
• 🧪 Notebook-based implementation for experimentation
• 📤 Streaming responses using graph.stream()

---

🧠 What is LangGraph?

LangGraph is a library that allows you to build stateful, multi-agent workflows for language models using a graph structure. It extends LangChain by enabling:

• Branching decisions (like if/else logic)
• Tool-based action steps
• Message memory tracking
• Streaming outputs

LangGraph is ideal for orchestrating tools + agents in real-time, as seen in this project.

---

🔧 Tools Used

getWorldNews(country_code)

Fetches latest headlines for a country using the GNews API. Example:

getBitcoinPrice()

Retrieves real-time Bitcoin price using a crypto API.

📥 Installation

Install the required libraries:

pip install langchain-core==0.3.13
pip install langchain==0.2.16
pip install langchain-huggingface==0.1.1
pip install langgraph==0.2.68
pip uninstall -y numpy transformers
pip install numpy transformers
pip install -qU langchain-groq

🔑 API Keys

You’ll need:

• ✅ GNews API Key for news

• ✅ LLM Provider Key (e.g., Groq or OpenAI)

▶️ How to Run

• Define inputs:

inputs = {
  "messages": [
    {"role": "user", "content": "You are good at cryptocurrency and world news"},
    {"role": "assistant", "content": "Ask me any question."},
    {"role": "user", "content": "What are the current situations in India?"}
  ]
}

• Run the agent with LangGraph:

print_stream(graph.stream(inputs, stream_mode="values"))