SpeedSense

Inspiration

We wanted a single place where everyday spending meets long-term investing. Students and early-career folks told us they track expenses in spreadsheets but never get clear next steps. That inspired SpendSense: a tool that explains where money goes, finds savings, and turns those savings into simple investment moves.

What it does

SpendSense is a smart wallet coach with AI spending analysis. It imports bank data from the Capital One Nessie API, auto-categorizes transactions, and explains patterns in plain English. It highlights recurring subscriptions, flags anomalies, and suggests quick actions to save more. It also provides personalized financial advice and investment recommendations powered by Gemini.

How we built it

  • Frontend: React + MUI for a responsive, accessible dashboard and a clean transactions table.
  • Backend: Node/Express with routes that proxy the Nessie API. We merge Nessie and CSV data into a unified schema.
  • AI spending analysis: Rules handle simple categories. A background job calls Gemini for hard cases and returns a short human-readable reason for each item. Gemini also writes concise spending insights and investment guidance tied to user goals.
  • Data and auth: MongoDB for persistence, JWT for secure sessions, and environment-scoped keys.

What we learned

  • Small prompts beat large ones when we constrain the output format.
  • Normalizing data from different sources is the key to reliable insights.
  • Short, specific advice helps users act. Clear reasons build trust in AI decisions.

Challenges

  • Handling edge cases in CSV mapping and keeping category names consistent.
  • Designing prompts that produce stable, short explanations.
  • Keeping the system fast while calling external APIs.
  • Ensuring secure key management while demoing live.

What’s next

  • Bank-grade CSV mappers for popular institutions.
  • Deeper portfolio planning with risk budgets and simple glide paths.
  • Smarter subscription tracking with cancellation playbooks.
  • Goal-based simulations that connect cut expenses to investment growth.

Built With

Share this project:

Updates