Inspiration
I wanted to explore how blockchain-style ideas could make everyday things like saving money with friends more transparent. Most savings apps are private and don’t show you much detail. This project flips that by making every transaction visible and verifiable.
What it does
Users can add deposits or withdrawals into the pot, and the app records it to a hash-linked ledger. There’s an option to sign transactions with a crypto wallet, but it also works without it. The app shows the full ledger, calculates totals, tracks progress toward a savings goal, and generates simple charts of who contributed what. At the end, there’s a coach summary that explains the pot’s status in plain language.
How I built it
I built the whole thing in Python with Streamlit for the interface. Pandas and Altair handle the data and charts, and I used eth-account for optional wallet signatures. The AI coach summary can run locally or connect to Google AI Studio if the user wants.
Challenges I ran into
I came into this with very little experience in Python or the tools I used, so a big challenge was just getting everything to work together. It took a lot of trial and error, digging through resources, and leaning on AI help to get the program functional. Setting up the MetaMask connection for signing was tough — it took a lot of testing and debugging before it worked reliably. The same was true for the Google AI Studio API; integrating it so the app returned clear summaries wasn’t straightforward. The MLH workshop on Google AI Studio was absolutely great and helped me understand how to connect the API properly, which made a huge difference. Another challenge was handling different currencies in the interface and making sure the totals and charts updated correctly whenever users switched them. Each of these felt like small roadblocks, but solving them gave me a much better understanding of how everything fits together.
Accomplishments that I'm proud of
I was able to build a working demo that ties everything together: a chain of transactions, charts, and a coach summary, all in one app. Doing this solo and learning new tools along the way was something I’m proud of.
What I learned
This project pushed me to try out a bunch of new tools and concepts. I learned how to use Altair to make charts that are interactive and actually explain the data in a clear way. I also got experience working with the Google AI Studio API, which showed me how to bring in AI-generated summaries and connect them to real app features. On the coding side, I leaned on GitHub Copilot to help speed up development and fill in gaps in my Python and Streamlit knowledge, which was super helpful while working solo. Finally, I learned a lot about how to connect to MetaMask in real time for wallet signatures, which gave me a better understanding of how blockchain-style attestations can plug into a normal web app.
What's next for FinSight
I’d like to connect the app to real financial data and possibly put the ledger on a test blockchain to see it working live. Other ideas include better rules for withdrawals and making the coach smarter with more detailed insights.
Log in or sign up for Devpost to join the conversation.