InsureRAG Agent — 60-minute Hackathon MVP

What it does

1. Upload an insurance policy PDF.
2. Extract and chunk the policy.
3. Build a local TF-IDF retrieval index.
4. Detect question intent (coverage, exclusions, claims, documents, waiting period).
5. Retrieve relevant policy pages.
6. Send only retrieved evidence to Groq Llama.
7. Return an evidence-grounded answer with page references.
8. Optionally fetch current public health-insurance guidance from the official IRDAI site.

Run in VS Code (Windows)

python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
copy .env.example .env

Open .env and paste your Groq API key.

Then:

python app.py

Open http://127.0.0.1:5000

Important

• Use a text-readable PDF for the fastest demo. Scanned/image-only PDFs need OCR.
• The uploaded policy is the primary source for policy-specific answers.
• IRDAI context is general regulatory guidance and does not replace the policy contract.
• Do not put your Groq API key in frontend code or GitHub.