FinsightAI: Empowering Financial Insight with AI

Problem Statement: Revolutionizing Information Access and Decision-Making with Large Language Models and Retrieval-Augmented Generation

In today's data-driven world, decision-making is hindered by inefficient access to real-time, accurate information. Traditional systems are limited in their ability to process vast amounts of data contextually. Businesses need more intelligent systems that can analyze real-time data, synthesize domain-specific knowledge, and offer actionable insights.

Our challenge was to create a cutting-edge solution that leverages Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) to transform enterprise decision-making. The goal was to provide a dynamic, context-aware system that empowers stakeholders with data-backed insights and innovative strategies.

How We Overcame This Problem: Our Solution

FinsightAI addresses this challenge by integrating a robust AI-powered engine that uses LLMs like GPT-4 and RAG technology. Our solution offers actionable insights, strategic recommendations, and real-time data analysis in the financial domain.

• RAG Integration: We enhanced LLM capabilities by implementing RAG to incorporate domain-specific knowledge, such as real-time market trends and company-specific data. This enables businesses to make more informed decisions with up-to-date insights.
• Interactive Q&A System: We built an intuitive, context-aware Q&A system that empowers users to ask specific questions about their business and receive dynamic, AI-driven responses tailored to their needs.
• Data Visualization: To help users quickly interpret AI-generated insights, we implemented real-time data visualization using D3.js and Plotly, ensuring that key information is easily accessible and understandable.

Overview

FinsightAI leverages the power of artificial intelligence to provide in-depth financial insights, analyze market trends, and predict stock movements. It's designed to help financial analysts, investors, and businesses make data-driven decisions with greater accuracy and efficiency.

Why FinsightAI?

1. Real-time Adaptation: The financial world is constantly evolving, and FinsightAI allows for rapid model updates with a lightweight adaptation process, minimizing costs and maximizing accuracy.

2. Accessibility: Unlike proprietary solutions with limited API access, FinsightAI is fully open-source, allowing for democratization of financial insights using publicly available data.

3. Reinforcement Learning: FinsightAI implements RLHF (Reinforcement Learning from Human Feedback), ensuring personalized insights that align with user preferences like risk tolerance and investment style.

How It Works

FinsightAI integrates financial data from multiple sources, analyzes historical and real-time data, and uses AI models to:

1. Predict stock prices for the next week.
2. Analyze sentiment from financial news and social media.
3. Provide personalized portfolio advice using risk profiles.

Try the Latest Demo

Check out our live demo at FinsightAI Demo to test our financial prediction and analysis features.

Screenshots

Prerequisites

For Machine Learning Model (Python)

• Python 3.10+
• Pip: Install pip if you haven't already.
• Virtual Environment (optional but recommended): To keep dependencies isolated, use venv or conda.
• Dependencies:
  • transformers==4.40.1
  • peft==0.4.0
  • sentencepiece
  • accelerate
  • torch
  • datasets
  • bitsandbytes
  • flask
  • flask_restx
  • numpy
  • pandas
  • scikit-learn
  • matplotlib
  • seaborn
  • requests

For Website (Next.js)

• Node.js: Make sure you have Node.js installed. Download here.
• npm or yarn: You will need a package manager like npm (comes with Node.js) or yarn.
• Next.js: Install globally with npm install next -g.

Installation

For Machine Learning Model (Python)

1. Clone the repository:

   git clone https://github.com/Prisha-Mordia/finsightai.git
   cd finsightai

2. (Optional) Create a virtual environment:

   python3 -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

4. Run the model:

   python run_model.py

For Website (Next.js)

1. Navigate to the frontend directory:

   cd finsightai/frontend

2. Install dependencies:

   npm install

3. Run the development server:

   npm run dev

   The website will be available at http://localhost:3000.

4. Build for production (optional):

   npm run build

Models and Benchmarks

Model Name	Accuracy	Performance	Cost
FinsightAI Predictor V1 (RTX 3090)	89.5%	High	$10
FinsightAI Sentiment Analysis (A100)	85.3%	Moderate	$15
LLAMA-3.2	87.8%	High	-

Team CodeStorm

• Team Leader: Prisha Mordia
• Members:
  • Punit Choudhary
  • Anmol Sharma
  • Sanskar Shrivastava

License

This project is licensed under the MIT License - see the LICENSE file for details.

Disclaimer: This project is for educational and research purposes only. Nothing here constitutes financial advice. Always consult a professional before making investment decisions.