Agentic RAG Assistant | LangGraph + Pinecone + Gemini | Full Project Demo
In this video, I walk through my Agentic RAG Assistant — a production-grade AI system built with LangGraph, Pinecone, and Google Gemini.
The system reads documents (websites, PDFs, Word files, and text files), stores them as vectors in Pinecone, and answers questions using only the information inside those documents.
What makes it agentic is the LangGraph state machine that classifies each question and routes it through either a simple retrieval path or a complex multi-step reasoning path with a Planner Agent and Verification Agent.
Tech Stack:
LangGraph (agentic workflow)
Google Gemini (embeddings + LLM)
Pinecone (vector database)
FastAPI (backend)
Streamlit (frontend)
Python
Topics covered in this video:
LangGraph workflow architecture
Simple vs Complex query routing
Planner Agent for multi-step reasoning
Pinecone vector storage and retrieval
FastAPI backend integration
Streamlit frontend demo
You can find my GitHub, LinkedIn, and email address in the channel description.
#LangGraph #RAG #AgenticAI #Pinecone #Gemini #Python #FastAPI #Streamlit #AIEngineering