Real Estate Research Tool

An AI-powered research assistant that scrapes real estate and financial pages, indexes them into a local vector store, and answers context-aware questions using Groq's Llama models.

Features

• URL scraping: Extracts text from web pages using unstructured.
• Vector storage: Uses ChromaDB for local embeddings and persistence.
• Embeddings: Uses the thenlper/gte-base embedding model for semantic search.
• LLM answering: Uses Groq/ChatGroq (Llama) for responses with contextual retrieval.
• Streamlit frontend: Simple UI to process URLs and chat with indexed documents.

Quick Start

1. Create and activate a virtual environment (recommended):

python -m venv .venv

On Windows:

.venv\Scripts\activate

On macOS / Linux:

source .venv/bin/activate

2. Install dependencies:

pip install -r requirements.txt

Or install the full stack:

pip install streamlit langchain langchain-community langchain-huggingface langchain-chroma langchain-groq unstructured python-dotenv sentence-transformers

Note (Windows): if unstructured reports file-type or libmagic issues, run:

pip install python-magic-bin

Configuration

1. Create a .env file in the project root with your Groq API key:

 GROQ_API_KEY=your_groq_api_key_here

Project Structure

The expected layout for the minimal app:

• real_state_tool/
  • main.py or app.py — Streamlit frontend
  • rag.py — RAG processing & indexing code
  • requirements.txt
  • resources/ — ChromaDB persistence (created at runtime)

Usage

Run the Streamlit app:

streamlit run app.py

Workflow:

• Enter up to 3 URLs in the sidebar.
• Click Process URLs to scrape, split, embed, and index content.
• Ask questions in the chat — the app retrieves relevant context and uses the LLM to answer.

Requirements

Add these to requirements.txt (example):

streamlit
langchain
langchain-community
langchain-huggingface
langchain-chroma
langchain-groq
unstructured
python-dotenv
sentence-transformers

Notes & Tips

• First run may download the gte-base embedding model (~200MB).
• Indexed data persists under resources/; you can reuse it instead of re-processing URLs.
• If you need robust HTML scraping, consider adding beautifulsoup4 or trafilatura as optional helpers.