Landing Page of App
Initial Summary Statistics Information
Conclusion From Our AI Coach
In Depth Insight Generation
AI Analysis and Recommendations

SafeSpend: Your Personal AI Financial Coach for Smarter Paytm Spending

Inspiration

India’s rapid adoption of digital payments, especially Paytm UPI, has transformed how millions pay and transfer money daily. However, many users struggle to understand and manage their spending from extensive transaction lists, leading to overspending and missed saving opportunities.

This inspired us to build SafeSpend—a personal AI financial coach that helps users make sense of their Paytm spendings and empowers smarter money management.

What it Does

SafeSpend allows users to upload their Paytm UPI monthly transaction statements and automatically categorizes expenses, compares spending against income, and identifies risky spending patterns.

Leveraging AI, it delivers personalized, encouraging advice and visual insights, helping users manage budgets, avoid surprises, and build healthy financial habits.

How We Built It

We used Python and Streamlit for the interactive front-end app. Pandas handles data cleaning and transaction categorization. For AI-driven financial coaching, we integrated a local llama3.1:8b language model via Ollama to generate friendly, actionable advice based on spending summaries.

The app takes Excel uploads, processes data, then outputs spending breakdowns and AI recommendations.

Challenges We Ran Into

Handling different formats and noisy transaction data required robust cleaning and normalization. Mapping Paytm’s emoji-based tags into consistent expense categories was tricky.

Running a large LLM locally with limited resources posed performance considerations. Crafting prompts to elicit clear, helpful AI advice without jargon demanded multiple iterations.

Accomplishments We’re Proud Of

Building a fully self-contained app combining real-world Paytm transaction data with advanced AI, running entirely on open-source, hackathon-friendly tools. Empowering users with clear financial literacy insights from complex raw data.

Delivering personalized coaching that motivates proactive money habits, not just raw numbers.

What We Learned

The importance of clean data input for meaningful AI outputs. How to leverage open-source LLMs for domain-specific coaching tasks. Best practices for user-friendly Streamlit apps with seamless file uploads and instant feedback. Fundamentals of personal finance budgeting to translate rules into AI prompts.

What's Next for SafeSpend - Personal Finance Risk & Literacy AI Coach

Expand spending trend visualizations with interactive charts and month-over-month comparisons.
Add user-configurable budget goals and alerting.
Integrate more data sources beyond Paytm UPI, including bank accounts and credit cards.
Improve AI advice with more context-aware and motivational coaching.
Prepare for public deployment and wider testing.