Completed
Full Stack

WebChat AI

A RAG chatbot that makes ChatGPT get context of any URL or website, built with Streamlit and LangChain.

Overview

WebChat AI is a browser content-aware chatbot that lets users paste a URL and chat with the content of that webpage. It extracts text from any site, processes it with LangChain embeddings, and provides contextual AI responses through a simple Streamlit UI.

What Does it Do?

  • Take a website link as input and scrape its textual content.
  • Split the scraped text into chunks for embedding.
  • Store text embeddings in a vector database for RAG retrieval.
  • Accept user queries and perform semantic search over the page content.
  • Generate context-aware answers using an LLM based on the retrieved website content.
  • Display chat history and responses within a Streamlit interface.

Why I Built this

  • To build a tool that lets ChatGPT understand and answer questions using the actual content of any URL.
  • To learn and apply LangChain and RAG techniques in a real project.
  • To provide an easy UI for contextual website Q&A without manual data prep.
  • To explore how vector embeddings can enhance chatbot relevance.

Tech Stack

  • Python
  • Streamlit
  • LangChain
  • BeautifulSoup
  • Embeddings and Vector Database
  • OpenAI / GPT-4

Designed and Developed by Shahu Kor