LangGraph Agentic RAG System

An autonomous AI agent built with LangGraph that decides how to retrieve information, reasons over it across multiple steps, and responds — moving beyond fixed RAG pipelines into true agentic behaviour.

Basic RAG vs Agentic RAG

	Basic RAG	Agentic RAG (this repo)
Retrieval	Always retrieves from vector store	Agent decides whether to retrieve, search web, or answer from memory
Steps	Fixed: retrieve → generate	Dynamic: agent plans multi-step actions
Tools	Single retriever	Multiple tools: retriever + web search + memory
Control flow	Linear	Graph-based with conditional edges

What's Inside

File	What It Covers
basic_langgraph.ipynb	LangGraph fundamentals — nodes, edges, state, conditional routing
basic_agents.ipynb	Building agents with tool calling — how agents decide what action to take
agentic_rag.ipynb	Full agentic RAG system — agent chooses between retrieval and web search
qna_chatbot_using_langgraph.py	Stateful QnA chatbot with conversation memory using LangGraph
rag_agent.py	Script version of the agentic RAG pipeline
google_search_engine.py	Google search tool integration for real-time web retrieval

Agent Architecture

User Question ↓ LangGraph Agent Node ↓ Agent decides action: ├── Retrieve from vector store (for document questions) ├── Search the web (for current/external information) └── Answer directly (if sufficient context in memory) ↓ Tool execution ↓ Agent evaluates result → loops back if more steps needed ↓ Final Answer

Key Concepts Demonstrated

• StateGraph — how LangGraph manages agent state across multiple steps
• Conditional edges — how the agent routes between tools based on reasoning
• Tool binding — attaching retriever and web search as callable tools
• Memory — maintaining conversation context across turns in the QnA chatbot
• Agentic loops — agent re-invokes tools when initial retrieval is insufficient

Tech Stack

Tool	Role
LangGraph	Agent graph orchestration and state management
LangChain	Document loading, splitting, retrieval chains
OpenAI GPT	Reasoning and response generation
Groq LLaMA	Alternative LLM for faster inference
Google Search API	Real-time web retrieval tool
FAISS	Vector store for document retrieval

Setup

git clone https://github.com/Saiajaykumar12/langgraph-agentic-rag
cd langgraph-agentic-rag
pip install -r requirements.txt
cp .env.example .env  # add your API keys

Run notebooks in this order for best understanding:

1. basic_langgraph.ipynb
2. basic_agents.ipynb
3. agentic_rag.ipynb

Environment Variables

OPENAI_API_KEY=your_openai_key

GROQ_API_KEY=your_groq_key

GOOGLE_API_KEY=your_google_key

GOOGLE_CSE_ID=your_custom_search_engine_id