🤖 Smart FAQ System using LangGraph

A context-aware, LLM-powered FAQ assistant built using LangGraph, LangChain, and sentence-transformer embeddings. The system retrieves answers from a knowledge base and dynamically routes user queries based on confidence scores.

---

🚀 Features

• ✅ Smart question answering over a CSV FAQ file
• ✅ Semantic search with HuggingFace embeddings
• ✅ LangGraph-powered flow control based on confidence
• ✅ Follow-up prompt if LLM is unsure
• ✅ REPL-style CLI for continuous Q&A
• ✅ Modular code: retrieval, graph, and state are cleanly separated

---

📁 Project Structure

smart-faq-system/
├── data/
│   └── faq_data.csv              # Your FAQ dataset
├── vectorstore/
│   └── faiss_index/              # Saved FAISS index (generated once)
├── embed_store.py               # One-time script to embed and store FAQ data
├── retriever.py                 # Loads vectorstore for runtime retrieval
├── nodes.py                     # LangGraph nodes: retrieval, logic, answer, etc.
├── state.py                     # FAQState definition for typed LangGraph state
├── langgraph_faq.py             # Graph creation and routing logic
├── main.py                      # CLI loop: ask questions until "quit"
└── requirements.txt             # Python dependencies

---

🧠 How It Works

1. Vector Embedding Run embed_store.py once to convert your FAQ CSV into embeddings and save the vector index using FAISS.

2. Graph Routing with LangGraph LangGraph routes the user query based on the confidence of the retrieved answer:

   • If the score is high, return the answer
   • If the score is low, ask a follow-up question

3. REPL Loop Run main.py and interact with the assistant in a loop.

---

📦 Setup Instructions

1. Clone the repo

git clone https://github.com/ceodaniyal/smart-faq-system.git
cd smart-faq-system

2. Install dependencies

pip install -r requirements.txt

3. Add your FAQ data

Place your faq_data.csv file in the data/ directory. A sample is already included.

4. Generate embeddings (run once)

python embed_store.py

5. Start the Smart FAQ System

python main.py

---

📝 FAQ Data Format (faq_data.csv)

question,answer
What are your business hours?,Our business hours are 9 AM to 6 PM, Monday through Friday.
How can I reset my password?,Click "Forgot Password" on the login page and follow the instructions.
...

---

🔧 Tech Stack

• LangGraph
• LangChain
• HuggingFace Embeddings
• FAISS
• Python 3.9+

---

✨ Future Improvements

• Switch to Chroma or MongoDB vector store with scoring
• Add Streamlit or Gradio UI
• Live feedback + learning loop
• Web API with FastAPI or Flask

---

🧑‍💻 Author

Made with 💡 by @ceodaniyal

---

📄 License

MIT License