UPI Smart

• 

  Adding phonePay screenshot

• 

  Ai generate analysis and suggestion to reduce expenses

• 

  Report analysis

Inspiration

With the rapid adoption of UPI, we noticed a common problem—people make dozens of small daily transactions but rarely have a clear understanding of where their money goes. Most expenses are scattered across screenshots, WhatsApp chats, and different payment apps. Existing expense trackers either require manual entry or depend on structured bank data, which doesn’t reflect real user behavior. This inspired us to build a solution that works in the messy, real-world environment of digital spending.

What it does

UPI Smart is an AI-powered expense intelligence tool that converts payment screenshots into meaningful financial insights. Users simply upload a screenshot, and the system: Extracts transaction details (amount, merchant, date) Automatically categorizes expenses Displays spending patterns through simple visualizations It eliminates manual tracking and helps users instantly understand their spending habits.

How we built it

Used Gemini API for image understanding and data extraction Applied OCR techniques combined with prompt engineering for accuracy Developed a clean frontend interface to visualize expenses and insights Focused on handling unstructured inputs like varied screenshot formats

Challenges we ran into

Handling inconsistent formats across apps like GPay, PhonePe, and Paytm Extracting accurate data from low-quality or partial screenshots Designing reliable categorization beyond simple keyword matching Ensuring a smooth user experience despite AI uncertainty/errors

Accomplishments that we're proud of

Successfully converting raw screenshots into structured financial data Building a working prototype with minimal manual input Creating a system that works even with fragmented and unorganized data Shifting the idea from a basic tracker to an AI-driven insight tool

What we learned

Real-world data is noisy and unpredictable compared to test datasets Prompt engineering is critical for improving AI output quality Simplicity and automation matter more than adding too many features Users value insights and awareness, not just data collection

What's next for UPI Smart

🔐 Secure user authentication and cloud sync 📊 Automated weekly/monthly expense reports 🚨 Smart budget alerts and anomaly detection 🤖 AI insights like “Where did you overspend this month?” 📱 Integration with SMS, WhatsApp, and email for auto-tracking 📈 Personalized financial recommendations

Built With

• gemini
• github
• nextjs
• rechart
• tailwindcss
• typescript
• vercel