🔍 AI Research Agent

Hey there! Welcome to my AI Research Assistant project - a smart little helper that digs through the web and Wikipedia to answer your questions. Built with LangGraph and LangChain, it's like having a super-powered research buddy at your fingertips.

What's Cool About It?

• Searches multiple places at once:
  • Scours the web using Tavily
  • Dives into Wikipedia articles
• Uses GPT-4 to craft well-thought-out answers
• Built on LangGraph (because who doesn't love clean, organized code?)
• Comes with a neat visualization of how everything works

Before You Start

You'll need:

• Python 3.7 or newer
• An OpenAI API key
• A Tavily API key (don't worry, it's free!)

Getting Started

First, grab all the packages you need:

pip install -U langgraph tavily-python wikipedia langchain_openai langchain_community langgraph_sdk

Setting Up Your Keys

Make sure you've got these environment variables ready to go:

• OPENAI_API_KEY: Get this from OpenAI
• TAVILY_API_KEY: Grab this from Tavily's website

How It Does Its Magic

The assistant works in three simple steps:

1. Gathering Info:

   • Searches the web using Tavily
   • Pulls relevant stuff from Wikipedia

2. Processing Everything:

   • Takes all that info and organizes it nicely
   • Gets it ready for our AI friend to read

3. Creating Answers:

   • Uses GPT-4 to whip up helpful answers based on what it found

All of this is managed by LangGraph, which keeps everything running smoothly and efficiently.

Want to Try It Out?

Here's how simple it is to use:

# Get everything ready
graph = builder.compile()

# Ask away!
result = graph.invoke({"question": "What's on your mind?"})

# See what it found
answer = result['answer'].content

Technical Bits

For the technically curious, here's how the state management works:

class State(TypedDict):
    question: str
    answer: str
    context: Annotated[list, operator.add]

Under the Hood

The system uses three main components:

• search_web: Your gateway to web results via Tavily
• search_wikipedia: Your Wikipedia explorer
• generate_answer: The piece that puts it all together

Want to Help?

Found a bug? Have an idea to make it better? Jump right in! Issues and pull requests are always welcome.

---

Pro tip: Remember to keep those API keys secret when sharing your code!