Inspiration
Most banking apps tell you how much money you have, but they don't always help you understand how much you can actually afford to spend.
Imagine having $2,500 in your checking account and transferring $1,000 to savings, only to realize your $1,800 rent is due in five days. The transfer seemed fine at the time, but it left you short on an important payment.
We wanted to make banking more conversational, accessible, and proactive. What if you could simply text your bank, ask questions about your finances, and understand the consequences of a decision before making it?
That's what inspired ChatterBank.
What it does
ChatterBank is an AI-powered banking assistant that lets you talk, type, or text your bank while helping you plan ahead.
- Smart Transactions: Identifies potential financial risks before a transfer or purchase and suggests safer alternatives.
- What-If Simulator: Compares financial decisions through conversations and interactive charts.
- Safe to Spend: Estimates how much money you can safely use without affecting upcoming bills.
- Financial Forecasting: Visualizes 30 days of expected income, bills, and balances.
- Financial Health Score: Analyzes five indicators to provide an overview of financial health.
- Text Banking: Supports iMessage conversations through Photon, with live dashboard updates.
- Secure Confirmation: Requires explicit approval and backend validation for simulated transfers.
How we built it
- React, Vite, Tailwind CSS & Recharts: Interactive banking dashboard and financial visualizations.
- Node.js & Express: Backend services and a custom financial simulation engine.
- Gemini API: Conversational AI, function calling, and financial explanations.
- Capital One Nessie: Simulated banking accounts, transactions, and transfers.
- MongoDB Atlas: Conversation history, saved scenarios, and transaction records.
- Photon Spectrum: Text-based banking through iMessage.
Challenges we ran into
- Refining our idea: Combining conversational banking and financial analysis into one focused, practical solution.
- Reliable calculations: Separating financial calculations from AI-generated responses to ensure accuracy.
- Text messaging setup: Configuring multiple phone numbers and connecting Photon messaging to the banking workflow.
- API reliability: Handling Gemini overload errors with a fallback system that keeps essential features working.
- Service integration: Connecting Gemini, Nessie, MongoDB Atlas, and Photon into one seamless experience.
Accomplishments that we're proud of
- Built a complete ask → simulate → compare → confirm transaction workflow.
- Integrated Gemini, Nessie, MongoDB Atlas, and Photon.
- Enabled iMessage banking with live dashboard updates.
- Developed Safe to Spend and a five-indicator financial health score.
- Designed an AI assistant that cannot execute simulated transfers without explicit user confirmation.
What we learned
- Think beyond balances: Forecasting helps users understand the impact of financial decisions before spending.
- Clarity over complexity: Bringing accounts, bills, and expenses together makes finances easier to understand.
- Conversation simplifies banking: Natural language makes financial information more accessible.
- Trustworthy AI matters: Gemini handles conversations while our backend manages calculations and transaction validation.
- Actionable warnings matter: Explaining risks and offering alternatives is more useful than simply displaying alerts.
What's next for ChatterBank
- Expand iMessage banking into real phone conversations.
- Introduce proactive alerts for upcoming bills and cash shortages.
- Add stronger authentication and multilingual support.
- Improve forecasts using spending patterns and explore secure real-bank integrations.
