🤖 Production AI ChatBot

A production-level AI Assistant built using LangGraph, LangChain, Google Gemini, RAG, and Agentic AI architecture.

This chatbot intelligently decides the best way to answer user queries using:

• 🧠 Gemini LLM
• 📚 Retrieval Augmented Generation (RAG)
• 🌤 Real-time Weather Tool
• 📈 Stock Market Tool
• 🔎 DuckDuckGo Web Search
• ▶️ YouTube Video Summarization

The project follows a modular production architecture using Services, Graph Workflow, Tools, RAG Pipeline, and Streamlit UI.

---

🚀 Features

🧠 Intelligent AI Router (LangGraph)

The chatbot uses an LLM-powered router to decide which component should handle the user query.

Example:

User:
"What is the weather in Delhi?"

Router:
→ Weather Tool

User:
"Explain this PDF"

Router:
→ RAG Pipeline

User:
"Summarize this YouTube video"

Router:
→ YouTube Tool

---

✨ Main Features

1. Gemini AI Chat

Powered by:

• Google Gemini API
• LangChain ChatGoogleGenerativeAI

Capabilities:

• General conversation
• Coding assistance
• Explanation
• Reasoning
• Content generation

---

2. 📚 RAG Document Question Answering

Users can upload documents and ask questions.

Supported formats:

• PDF
• DOCX
• TXT

RAG Pipeline:

Document Upload

        ↓

Document Loader

        ↓

Text Splitter

        ↓

Embedding Generation

        ↓

Vector Database

        ↓

Retriever

        ↓

Gemini LLM

        ↓

Final Answer

Technologies:

• LangChain
• ChromaDB
• FAISS
• HuggingFace Embeddings
• Gemini

---

3. 🌤 Weather Tool

Provides real-time weather information:

• Temperature
• Weather condition
• Humidity
• Wind speed
• Forecast

Example:

"What is today's weather in Mumbai?"

---

4. 📈 Stock Market Tool

Provides financial information:

Supported:

• NSE
• BSE
• NASDAQ

Example:

"What is the current price of Tesla?"

---

5. 🔎 DuckDuckGo Search Tool

Used for:

• Latest news
• Current events
• Real-time information
• Web search

Example:

"Latest AI news"

---

6. ▶️ YouTube Summarizer

Features:

• Extract YouTube transcript
• Summarize videos
• Generate key points
• Answer questions from videos

Example:

"Summarize this YouTube video"

---

🏗️ System Architecture

                 User

                  |

                  |

            Streamlit UI

                  |

                  |

            Chat Service

                  |

                  |

          LangGraph Workflow

                  |

                  |

            AI Router Agent

                  |

 ------------------------------------------------

 |              |              |                |

Chat           RAG          Tools          Search

 |              |              |

Gemini     Vector DB      APIs

                  |

                  |

          Final Response

                  |

                  |

        SQLite Conversation Memory

---

📂 Project Structure

ChatBot/

│
├── app.py

├── README.md

├── requirements.txt

├── setup.py

├── Dockerfile


│
├── config/

│   ├── config.py

│   └── constants.py


│
├── core/

│   ├── logger.py

│   └── exception.py


│
├── graph/

│   ├── state.py

│   ├── router.py

│   ├── nodes.py

│   └── workflow.py


│
├── rag/

│   ├── rag_service.py

│   ├── loader_factory.py

│   ├── retriever_service.py

│   └── document_splitter.py


│
├── services/

│   ├── llm_service.py

│   ├── chat_service.py

│   ├── document_service.py

│   ├── conversation_service.py

│   └── checkpoint_service.py


│
├── tools/

│   ├── weather_tool.py

│   ├── stock_tool.py

│   ├── duckduckgo_tool.py

│   └── youtube_tool.py


│
└── ui/

    ├── styles.py

    ├── sidebar.py

    ├── chat.py

    ├── components.py

    └── utils.py

---

🛠️ Tech Stack

Programming Language

• Python 3.11+

Generative AI

• Google Gemini
• LangChain
• LangGraph

RAG

• ChromaDB
• FAISS
• HuggingFace Embeddings

Frontend

• Streamlit

Database

• SQLite

Monitoring

• LangSmith

Deployment

• Docker
• HuggingFace Spaces
• Cloud Platforms

---

⚙️ Installation

Clone Repository

git clone https://github.com/yourusername/ChatBot.git

cd ChatBot

---

Create Virtual Environment

python -m venv venv

Activate environment:

Windows:

venv\Scripts\activate

Linux/Mac:

source venv/bin/activate

---

Install Dependencies

pip install -r requirements.txt

---

🔑 Environment Variables

Create a file:

.env

Add:

GOOGLE_API_KEY="your_gemini_api_key"


LANGSMITH_TRACING_V2=true

LANGSMITH_API_KEY="your_langsmith_api_key"

LANGSMITH_PROJECT="ChatBot"

---

▶️ Run Application

Start Streamlit:

streamlit run app.py

Open:

http://localhost:8501

---

🐳 Docker Deployment

Build image:

docker build -t ai-chatbot .

Run container:

docker run \
-p 8501:8501 \
--env-file .env \
ai-chatbot

---

🧪 Example Queries

General AI Chat

Explain Transformers in simple words

Document Question Answering

Upload PDF:

Summarize this document

What are the important points?

Weather

Weather in Delhi today

Stock

Tesla stock price

Web Search

Latest AI trends

YouTube

Summarize this video:
https://youtube.com/...

---

📊 LangSmith Monitoring

LangSmith tracks:

• LLM calls
• LangGraph workflow
• Tool execution
• RAG retrieval

Flow:

User Query

      ↓

Router

      ↓

Tool Selection

      ↓

Execution

      ↓

Gemini Response

      ↓

LangSmith Trace

---

🔐 Security Features

Implemented:

• Environment based API keys
• File validation
• Exception handling
• Logging system
• Thread isolation
• Modular architecture

---

🚀 Future Improvements

• Authentication
• User dashboard
• PostgreSQL integration
• Redis caching
• Multi-agent architecture
• Voice assistant
• Image understanding
• Kubernetes deployment

---

👨‍💻 Author

Shravan Kumar Pandey

B.Tech Computer Science Engineering (Data Science)

Interested in:

• Data Science
• Machine Learning
• Generative AI
• LangChain
• LangGraph
• RAG Systems

---

⭐ Support

If you like this project, give it a ⭐ on GitHub.