FinSightAI

🧠 FinSightAI - Finance Advisor

Your AI-powered assistant for smarter financial planning, document analysis, and portfolio recommendations.

🚀 Overview (TL;DR)

A web app that:

Answers general financial questions using a RAG-powered AI chatbot 💬
Analyzes uploaded documents like invoices/contracts and flags financial risks 📄
Provides personalized financial strategy based on user profile 💸

🎯 Problem Statement

Financial literacy is low, and most people struggle with:

Understanding complex documents like balance sheets and contracts
Knowing how to start investing based on their personal income/expenses
Getting trustworthy, AI-driven financial guidance

💡 Our Solution

Personal Finance Advisor is a 3-in-1 tool that:

Acts as a financial chatbot using RAG (Retrieval-Augmented Generation)
Classifies and analyzes uploaded documents with a risk flag system
Offers investment strategies tailored to your income, goals, and risk appetite

🔍 Features

📚 Module 1: RAG Chat for Financial Queries

Gemini based AI chatbot
Trained on curated financial knowledge base
Explains terms like SIP, tax-saving options, loans, etc.

📄 Module 2: Financial Document Inspector

Upload PDFs, DOCX, or images
Auto-detects type (invoice, contract, balance sheet)
Extracts key terms: amount, date, interest rate, penalties
Flags risks (hidden fees, lock-ins)
Generates a structured summary + risk report

📊 Module 3: Personal Strategy Generator

Takes user input (age, income, expenses, loans, risk level)
Suggests diversified portfolios
Gives actionable financial planning advice

🗂️ Project Directory Structure

project/
│
├── main.py                    # Unified Streamlit app integrating all 3 modules
├── advisor_app.py             # Standalone Streamlit UI for financial advisor
├── rag_chatbot_app.py         # Standalone Streamlit UI for RAG chatbot
├── pdf_ai_app.py              # Standalone Streamlit UI for document analysis
│
├── financial_advisor_api.py   # Backend logic for financial strategy module
├── rag_chatbot_api.py         # Backend logic for RAG chatbot
├── pdf_ai_api.py              # Backend logic for document analysis
│
├── advisor_model.pt           # Saved PyTorch model for strategy module
├── rag_model.pt               # Saved model for RAG chatbot
├── pdf_model.pt               # Saved model for PDF understanding
│
├── requirements.txt           # Python dependencies

🛠️ Tech Stack

Backend: FastAPI (Python)
Frontend: Streamlit
AI: Gemini + LangChain
ML: Transformers + scikit-learn + PyTorch
Document Embedding: FAISS
PDF Parsing: PyMuPDF
OCR (optional): Tesseract

📸 Sample Workflow

User launches main.py and sees integrated tabs: Ask Chat, Upload Document, Get Advice
Uploads a contract → system flags lock-in clauses + suggests negotiation tips
Asks chat: “How to save tax under 80C?” → AI responds with examples
Inputs age/income/expenses → gets personalized portfolio suggestion

🧪 How to Run Locally

git clone https://github.com/shivankursharma018/FinSightAI.git
cd FinSightAI

# Backend
python -m venv venv
.venv\Scripts\Activate.ps1

pip install -r requirements.txt
uvicorn app:app --reload

# Integrated Streamlit App
streamlit run main.py

🧠 Sample Inputs & Outputs

Upload: “Invoice_ABC.pdf” → Output: Document type = invoice, Total = ₹22,500, Risk = Late fee
Chat: “What is a SIP?” → Output: “A Systematic Investment Plan (SIP) allows investors to invest a fixed amount…”
Advice Tool: Age 27, ₹50k income, ₹20k expenses, medium risk → Output: “30% large cap funds, 20% FDs, 10% gold, 40% emergency + debt”

👨‍💻 Team

📜 License

MIT License

🔮 Future Scope

Add OCR for image-based invoices
Enable live market integration for portfolio suggestions
Export advice and reports as downloadable PDFs

🙌 Acknowledgments

Gemini, LangChain, Streamlit
HuggingFace (FinBERT)
Hackathon organizers + mentors

This site is open source. Improve this page.