FinGenie: Your AI Accountant

Screenshot of the working chatbot
Screenshot of the backend code of the chatbot model

Inspiration

The idea for FinGenie came from the complexity of Indian tax laws and the challenges individuals face in understanding financial regulations. Many taxpayers struggle with compliance, deductions, and filing requirements, leading to financial inefficiencies. By leveraging AI, I wanted to create an accessible and intelligent assistant that can provide instant, accurate, and personalized guidance.

What it does

FinGenie is an AI-powered chatbot designed to assist individuals and businesses with Indian tax laws and financial queries. The project aims to simplify tax compliance, provide real-time financial advice, and ensure accurate information retrieval from tax documents.

How we built it

AI Model Development:

Initially, I experimented with Pickaxe to create a chatbot trained on Indian tax laws.
I later aimed to develop a custom ML model that could read PDF tax documents and extract relevant insights.
1. Technology Stack:
Frontend: Vercel-hosted website for user interaction.
Backend: Render-based API for chatbot integration.
Database: To store chat history and enhance user experience.
Machine Learning: Custom NLP model for document comprehension and intelligent responses.

Integration:
- API-based chatbot deployment for smooth interaction.
- Plans to build a browser extension or background app for real-time tax-related query assistance.

Challenges we ran into

Building an ML Model from Scratch: Avoiding dependency on pre-trained models like Ollama while ensuring high accuracy.
Processing Large-Scale Tax Documents: Efficiently extracting data from complex PDFs.
Deployment & Scalability: Ensuring smooth integration of the chatbot on a website with real-time query handling.
Maintaining Chat History: Implementing a structured database system to store user interactions for a better experience.

Accomplishments that we're proud of

Successfully built an AI chatbot capable of answering Indian tax-related queries.
Designed a structured system for handling and storing chat history.
Developed an initial prototype with API-based integration.
Created a roadmap for expanding the chatbot’s capabilities, including a browser extension.
## What we learned
Advanced NLP techniques for document processing and chatbot development.
API integration for seamless chatbot deployment.
Efficient ways to structure and manage real-time databases.
The importance of user-friendly AI-driven financial tools.
Strategies for improving AI accuracy and response time.