Inspiration

I wanted to build something that actually solves a real problem people face every day. Most people have no clue how healthy their finances are. They know roughly what they earn and spend, but they have no score, no benchmark, no clear picture of what risks they are taking or where they will end up in 20 years. I wanted to change that with AI.

What it does

FinSight AI analyzes your income, expenses, savings, debt, and investments and gives you a financial health score from 0 to 100. It then breaks everything down across 5 features: a full overview with charts, a 20 year wealth projector, a budget optimizer based on the 50/30/20 rule, an automatic risk detector, and an AI chat advisor that knows your full financial profile and answers your specific questions.

How I built it

I built it using Python and Streamlit for the frontend, Plotly for the charts, and the Groq API running Llama 3.3 70B for the AI analysis and chat. The financial health score is calculated from a custom algorithm that weighs savings rate, expense ratio, debt to income ratio, and investment activity. The AI receives the user's full financial data and generates personalized assessments, action plans, and debt strategies in structured JSON which the app then renders cleanly.

Challenges I ran into

Getting the AI to return consistent structured JSON without breaking the parser took some work. Deployment on Streamlit Cloud also had some dependency issues that needed fixing. Building a financial scoring algorithm that felt accurate and fair across different income levels was harder than expected.

What I learned

I learned how to combine a custom scoring system with generative AI to create something that feels genuinely intelligent and useful. I also got much better at deploying Streamlit apps and working with the Groq API.

What's next for FinSight AI

Bank API integration to pull real transaction data automatically, a mobile version, personalized investment recommendations, and a goal setting feature where users can track progress toward

Built With

• github
• groq
• llama-3.3-70b
• plotly
• python
• streamlit
• streamlit-cloud