🧾 Fintrack – AI-Powered Personal Budget Tracker

💡 What Inspired Me

I’ve always believed that budgeting shouldn't be complicated—but most tools either overwhelm users with data or lack personalization. I wanted to build something that not only simplifies expense tracking but also provides actionable financial insights powered by AI. The idea was to create a clean, intuitive, and intelligent experience that anyone—especially students and early professionals—could use daily.

This inspired Fintrack, a lightweight full-stack web app that combines traditional expense tracking with real-time AI analysis.

🛠️ How I Built It

The project uses a modern Node.js + React-based architecture:

Frontend:

• Built using Bolt (low-code/no-code) to quickly develop UI components.
• Includes:
  • A landing screen where users enter budget categories (Transport, Food, Groceries, Others).
  • A home dashboard that displays:
  • Total income and expenses.
  • A dynamic list of transactions.
  • A floating "+" button to add new entries.
  • An AI Assistant tab for financial tips and insights.

Backend (Node.js):

• Uses Express.js for API routing.
• A custom POST endpoint (/api/analyze) receives user budgets and transactions.
• Integrates with OpenAI's GPT-4 Turbo to analyze user data and generate personalized insights.
• Includes /health endpoint for health checks.
• Error-handling and debug logging implemented throughout.

AI Logic:

• Calculates:
  • Total income, expenses, and balance
  • Expense breakdown by category
  • Budget utilization percentage
• Sends structured data to OpenAI, which returns tailored suggestions like:
  • “You're spending 85% of your food budget. Consider meal prepping to save.”
  • “Groceries and transport are under-budget—great job!”

🚧 Challenges I Faced

• Silent server exits: The backend initially started and stopped immediately due to silent promise rejections. I resolved this with proper try/catch blocks and process.on('unhandledRejection') handlers.
• OpenAI API behavior: Handling large prompt objects and optimizing for clarity and token usage took tuning.
• Data flow: Passing and structuring transaction data from Bolt’s frontend into the Express server required extra care with serialization and validation.
• Debugging in a virtual environment: Running Node inside a Python virtual environment (venv) caused initial confusion as the terminal showed misleading behavior.

📚 What I Learned

• How to architect a full-stack budgeting app from scratch using both low-code UI tools (Bolt) and custom APIs.
• How to structure AI prompts for financial data analysis and use GPT models effectively.
• How to handle asynchronous flows and prevent silent crashes in a Node.js server.
• The importance of good UX in fintech—keeping the experience clean, responsive, and reassuring.

🚀 What's Next

• Adding authentication and persistent storage (Firebase or Supabase)
• Real-time charts with category breakdowns
• A smarter AI assistant with chat support and goal-based tracking
• Exporting reports as PDF
• Notifications when a user is nearing their budget

🧠 "Fintrack" was built not just to record transactions—but to help users reflect, adjust, and grow financially smarter, one insight at a time.

Built With

• express.js
• openai
• typescript