Agentic Chatbot

A multi-usecase AI agent application built with LangGraph and Streamlit, demonstrating progressively complex agentic patterns — from a simple chatbot to an autonomous AI news aggregator.

---

Features

Usecase	Description
Basic Chatbot	Conversational AI with persistent message history
Chatbot with Web Search	Agentic chatbot that decides when to search the web using Tavily
AI News Aggregator	Autonomous pipeline that fetches, summarizes, and saves AI news

---

Tech Stack

• LangGraph — Graph-based agent orchestration
• LangChain — LLM orchestration framework
• OpenAI — LLM backend (gpt-4o, gpt-4o-mini)
• Tavily — Web search and news retrieval API
• Streamlit — Web UI
• Python 3.12

---

Project Structure

Agentic_Chatbot/
├── app.py                          # Entry point
├── pyproject.toml                  # Project metadata & dependencies
├── AINews/                         # Auto-generated news summaries
│   ├── daily_summary.md
│   └── weekly_summary.md
└── src/
    └── langgraph_agentic_ai/
        ├── main.py                 # Core app logic
        ├── LLMs/
        │   └── openaillm.py        # OpenAI LLM initialization
        ├── graph/
        │   └── graph_builder.py    # LangGraph graph construction
        ├── nodes/
        │   ├── basic_chatbot_node.py
        │   ├── chatbot_with_tool_node.py
        │   └── ai_news_node.py
        ├── state/
        │   └── state.py            # Shared graph state schema
        ├── tools/
        │   └── search_tool.py      # Tavily search tool wrapper
        └── ui/
            ├── uiconfigfile.ini    # UI configuration
            ├── uiconfigfile.py     # Config loader
            └── streamlitui/
                ├── loadui.py       # Sidebar controls & inputs
                └── display_results.py  # Result rendering

---

Architecture

The application follows a node-and-graph architecture powered by LangGraph. Each usecase is implemented as a distinct graph topology:

Basic Chatbot

START → Chatbot Node → END

Chatbot with Web Search (Agentic Loop)

START → Chatbot Node → (tool call?) → Tools Node → Chatbot Node → END
                     ↘ (no tool call) → END

AI News Aggregator (Sequential Pipeline)

START → Fetch News → Summarize News → Save to File → END

---

Screenshots

Basic Chatbot

Chatbot with Web Search

AI News Aggregator

---

Prerequisites

• Python 3.12
• An OpenAI API key — required for all usecases
• A Tavily API key — required for Chatbot with Web and AI News usecases

---

Installation

# Clone the repository
git clone https://github.com/souravnagesh/Agentic-Chatbot.git
cd Agentic_Chatbot

# Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate        # macOS/Linux
.venv\Scripts\activate           # Windows

# Install dependencies
pip install -e .

Note: This project uses uv for dependency management. If you have uv installed, run uv sync instead.

---

Running the App

streamlit run app.py

The app will open at http://localhost:8501.

---

Usage

1. Configure the sidebar:

   • Select an LLM (OpenAI)
   • Choose a model (gpt-4o-mini or gpt-4o)
   • Enter your OpenAI API key
   • Select a usecase

2. For Basic Chatbot: Type a message and press Enter.

3. For Chatbot with Web Search: Enter your Tavily API key, then chat. The agent will autonomously decide when to search the web.

4. For AI News: Enter your Tavily API key, select a timeframe (Daily / Weekly / Monthly), and click Fetch Latest AI News. Summaries are saved to AINews/.

---

Environment Variables

Create a .env file in the project root to pre-load API keys (optional — they can also be entered via the UI):

OPENAI_API_KEY=your_openai_api_key
TAVILY_API_KEY=your_tavily_api_key

---

Generated News Output Format

The AI News aggregator saves summaries in Markdown:

# Daily AI News Summary

### 2026-03-26

- GPT-5 released with multimodal capabilities (https://source-url.com)
- ...

---

Dependencies

Package	Purpose
langgraph	Graph-based agent orchestration
langchain	LLM orchestration
langchain-openai	OpenAI integration
langchain-community	Community tools (Tavily)
streamlit	Web UI
tavily-python	Web search API
python-dotenv	.env file support
pydantic	Data validation

---

License

This project is open-source and available under the MIT License.