This project is an AI-powered Financial Advisor specifically designed to assist elderly users with their financial queries and concerns. It aims to provide accessible, personalized financial advice using natural language processing and machine learning techniques.

1. Personalized User Profiles: The bot creates and maintains individual user profiles, storing relevant financial information securely.
2. Sentiment Analysis: It analyzes the user's emotional state to provide empathetic responses.
3. Financial Context Recognition: The bot identifies the specific financial context of each query (e.g., spending, saving, investments).
4. Synthetic Spending Data: For demonstration purposes, the bot generates realistic spending data to provide insights.
5. Persistent Storage: User profiles and conversation histories are securely stored for continuity across sessions.
6. Natural Language Interaction: Users can communicate with the bot using natural language, making it accessible for those who might struggle with complex interfaces.
7. User Authentication: Implements a login system to securely manage multiple users.

• Language: Python
• AI/ML Libraries: OpenAI's GPT-3.5, Hugging Face Transformers
• Data Storage: MongoDB
• User Interface: Text-based interface (can be extended to voice interface for increased accessibility)
• Backend Framework: FastAPI
• Frontend: Streamlit

Our AI Financial Advisor application is built using a modern, modular architecture that separates concerns and allows for scalability and easy maintenance. Here's an overview of the key technical layers:

• Technology: Streamlit
• Purpose: Provides an intuitive, web-based user interface for interacting with the AI Financial Advisor.
• Key Features:
  • User authentication and profile management
  • Chat interface for communicating with the AI
  • Display of financial insights and advice
  • Visualization of spending data and financial trends

• REST API: Built with FastAPI

  • Provides endpoints for user management, profile operations, and AI interactions
  • Handles request validation and response formatting
  • Implements JWT-based authentication for secure API access

• Database: MongoDB

  • Stores user profiles, financial data, and conversation histories
  • Provides efficient querying and updating of user information
  • Ensures data persistence across sessions

• Conversational AI: OpenAI GPT-3.5-turbo

  • Generates human-like responses to user queries
  • Provides personalized financial advice based on user context
  • Handles a wide range of financial topics without extensive domain-specific training

• Sentiment Analysis: DistilBERT

  • Analyzes the emotional tone of user inputs
  • Allows the AI to provide empathetic responses
  • Fine-tuned on the SST-2 (Stanford Sentiment Treebank) dataset for efficient performance

• Architecture: Transformer-based models

  • Both GPT-3.5 and DistilBERT use the Transformer architecture
  • Excellent at capturing long-range dependencies in text
  • Enables understanding of complex financial queries and generation of coherent, contextually relevant responses

1. Language Model - GPT-3.5 (OpenAI): Chosen for its advanced natural language understanding and generation capabilities. It produces context-aware, human-like responses, crucial for providing empathetic and personalized financial advice.

2. Sentiment Analysis - DistilBERT (Hugging Face): A lightweight version of BERT, fine-tuned on the SST-2 (Stanford Sentiment Treebank) dataset. It's suitable for real-time sentiment analysis in our application.

3. Transformer Architecture: Both GPT-3.5 and DistilBERT are based on the Transformer architecture, ideal for understanding complex financial queries and generating coherent, contextually relevant responses.

• Python 3.7+
• pip (Python package manager)
• MongoDB installed and running

git clone https://github.com/alipala/ai4elderly1.git
cd ai4elderly1

python -m venv env
source env/bin/activate

pip install -r requirements.txt

Replace your_secret_key_here with a secure random string and your_openai_api_key_here with your actual OpenAI API key.

MONGO_URL=mongodb://localhost:27017
DB_NAME=financial_advisor_db
SECRET_KEY=your_secret_key_here
ALGORITHM=HS256
ACCESS_TOKEN_EXPIRE_MINUTES=30
OPENAI_API_KEY=your_openai_api_key_here
API_URL=http://localhost:8000

cd backend
uvicorn main:app --reload

cd frontend
streamlit run streamlit_app.py

Open your web browser and go to http://localhost:8501 to access the Streamlit interface

1. Open the Streamlit app in your web browser (at http://localhost:8501).
2. Log in with test admin user
3. Create a new user profile(or load an existing one)
4. Generate syntetic spending data to have accurate advice based on this
5. Interact with the Financial AI Advisor by typing your financial queries or concerns.
6. View personalized advice, spending insights, and conversation history