Inspiration
A lot of us struggle to keep track of where our money goes. We wanted a tool that could help us clearly understand the most responsible financial decisions to make based on our current situation. Monetra came from our own experiences trying to save for things like moving-out, buying a car, or just having an emergency fund—without feeling overwhelmed.
What it does
Monetra connects to your financial data (currently simulated using the Plaid Sandbox), automatically categorizes your transactions into Needs, Wants, and Savings using AI, and generates clear visual breakdowns so you can understand what’s realistic—not just ideal. You can set financial goals like “Save for a car,” enter how much you need and by when, and the app calculates how much you need to save each month based on your actual spending habits. On top of that, Monetra analyzes your spending patterns to recommend financial tools—such as credit cards with high cashback in your top spending categories or high-yield savings accounts when you're focused on saving.
How we built it
We built Monetra with Next.js, React, and Tailwind for the frontend. Authentication is done with Supabase Auth, and transaction data is simulated using Plaid’s Sandbox API. We process that data with Gemini AI, which helps categorize spending into Needs, Wants, and Savings. The Needs, Wants, and Savings pages all pull from this categorized data. For now, data like goals are stored locally for speed, but we're planning to move this to Supabase/PostgreSQL.
Challenges we ran into
- Getting Plaid Sandbox to return data properly and learning how public tokens, access tokens, and accounts actually connect.
- Parsing and cleaning Gemini’s responses so they were usable and consistent JSON.
- Handling merge conflicts and keeping branches organized when multiple people were working on different features at the same time.
- Making sure the UI didn’t crash when no accounts were linked or when data wasn’t loaded yet.
Accomplishments that we're proud of
- We built a working dashboard that updates based on categorized transactions.
- Our app doesn't just track spending—it calculates savings goals and checks if they're realistic with your current budget.
- We fully set up Supabase authentication and a working login/signup flow.
What we learned
- How to securely send financial data to an AI model (Gemini), format prompts, validate responses, and handle incorrect or malformed JSON from the model.
- How to work as a team with GitHub—pull requests, upstream repositories, branching workflows, rebasing, and resolving merge conflicts.
- How Plaid actually works—public tokens, access tokens, accounts, and how transaction syncing happens behind the scenes.
What's next for Monetra
-Replace hardcoded storage with a real database (Supabase/Postgres). -Let users edit or reclassify transactions manually if AI gets them wrong. -Track goal progress based on real savings activity instead of just manual updates. -Add monthly summaries and notifications like “You’re overspending on eating out” or “You only need $50 more to stay on track.” -Make it mobile-friendly so people can check their money on the go.
Log in or sign up for Devpost to join the conversation.