Full-Stack LangChain Chatbot (Flat Local Setup)

A decoupled full-stack AI chatbot architecture built with FastAPI, Streamlit, and LangChain. This repository implements stateful chat memory across network boundaries and leverages dynamic model orchestration.

Architecture Overview

• Backend (FastAPI + LangChain): Manages conversational state tracking using InMemoryChatMessageHistory inside gpt3.py to bypass LLM statelessness. It translates incoming user strings into structured message payloads and communicates with OpenRouter API nodes.
• Frontend (Streamlit): Serves as an interactive interface inside Chatbot.py. It communicates asynchronously with the backend API over HTTP requests and handles visualization persistence across page refreshes via st.session_state.

Current Project Structure

FASTAPI/
│
├── __pycache__/
├── myvenv/                # Your Python virtual environment
├── Chatbot.py             # Streamlit chat user interface (Frontend)
├── gpt3.py                # FastAPI & LangChain orchestration logic (Backend)
├── secret_key.py          # Local OpenRouter API key credentials file
└── README.md              # Project documentation

Setup and Installation

Running Globally (Local Without Docker)

1. Clone the repository and navigate into the project directory:

   git clone https://github.com
   cd FASTAPI

2. Activate your existing virtual environment (myvenv):

   # On Windows (PowerShell):
   .\myvenv\Scripts\Activate.ps1
   
   # On Windows (Command Prompt):
   .\myvenv\Scripts\activate.bat

3. Install the required dependency libraries if you haven't already:

   pip install fastapi uvicorn langchain-openrouter langchain-core streamlit requests

4. Verify your secret_key.py contains your token:

   openrouter_api_key1 = "your_actual_openrouter_api_key_here"

5. Run your FastAPI backend server using your local environment python runner:

   .\myvenv\Scripts\uvicorn gpt3:app --reload

6. Open a separate terminal split window, ensure your virtual environment is active, and launch the Streamlit frontend client app:

   .\myvenv\Scripts\streamlit run Chatbot.py

• Frontend UI Workspace: Open http://localhost:8501 inside your web browser.
• Backend API Endpoints: Open http://localhost:8000/docs to view the interactive FastAPI Swagger documentation.

System Workflow

1. The user types a query into the Streamlit interactive chat box (Chatbot.py).
2. The user interface logs the message locally into st.session_state and fires an HTTP GET network request containing the prompt string to the local address http://127.0.0.
3. The FastAPI endpoint inside gpt3.py captures the string parameter, wraps it into a HumanMessage schema object, and saves it into InMemoryChatMessageHistory.
4. The historical conversation log list is bundled and pushed to OpenRouter via the universal model router openrouter/free to maintain conversational context.
5. The API parses the text answer, updates the backend memory map with an AIMessage data object, and returns a JSON dictionary back across the network to render the text bubble visually in Streamlit.