📊 FinXtract: LLM-Powered Annual Report Insight Tool

FinXtract is an LLM-powered tool that transforms annual reports into structured, searchable, and insightful data — with no manual rule-writing.

🚀 What It Does

Upload a full annual report (PDF or text), and FinXtract will:

• 🗂 Detect Document Structure
  Identify key sections (e.g. Chairman’s Statement, Risk Disclosure) and map them to page numbers

• 📄 Extract Section Text
  Retrieve and display full text for each detected section

• 🧾 Named Entity Recognition (NER)
  Automatically annotate entities like names, roles, organisations, and locations

• ❓ Question Answering
  Ask natural-language questions about the company or report
  e.g. “Who is the chairman?”, “What risks are mentioned?”

• 📈 Keyword Tracking & Chart Generation
  Analyse keyword trends and visualise them across the report

🧠 Powered By

• 🔗 Large Language Models (LLMs)
• 💬 Prompt Engineering
• 🐍 Python + Streamlit (or Flask)
• 📄 PDF/Text parsing (e.g. PyMuPDF)

💼 Use Cases

• Financial and ESG analysts
• Regulatory and audit teams
• NLP researchers and students
• Anyone working with complex corporate disclosures

🏁 Getting Started

git clone https://github.com/yourusername/finxtract.git
cd finxtract
pip install -r requirements.txt
python app.py