CreditPick

Home screen
voice input
card details
compare cards
why this card

\section*{\Credit Pick}

Inspiration

Every one of us has stood at checkout holding four cards and a blank stare, silently doing cashback math while the cashier waits. Reward apps exist, but none of them just tell you — out loud, in the moment — and none of them notice that using the "best" card might quietly wreck your utilization. We wanted a wallet that talks back.

What it does

Say what you're buying. CreditPick listens, figures out the merchant and category, and scores every card in your wallet as reward minus risk — real cashback rate, minus the FICO-utilization cost that purchase would cause on that card, tuned to your own credit profile. Then it just tells you which one to use, out loud, with the receipts to back it up.

How we built it

Voice → Gemini → scoring backend engine → ElevenLabs. One Gemini call handles transcription and structured extraction at once; a dependency-free Python engine does the actual math; ElevenLabs speaks the verdict. Nessie supplies saved account data, VectorMint supplies real reward rates, Expo/FastAPI/Postgres hold it all together. Four of us, a day and a half, four parallel tracks — held together by locking the schema and API contracts before anyone wrote a line of feature code.

Challenges we ran into

Nessie does not provide credit limits, so we had to store those separately to calculate utilization. We also had to make sure the AI could not invent reward figures: the scoring engine calculates the numbers, and the app checks the AI’s response against them.

Accomplishments that we're proud of

We built a working voice-to-voice flow: a user can describe a purchase and hear a card recommendation. We’re also proud that each recommendation comes from calculations we can explain, rather than an AI guess.

What we learned

We learned how much data a useful card recommendation needs beyond the reward rate. We also learned that AI works best here when it helps people ask questions and understand answers, while the financial comparison stays grounded in calculations.

What's next for CreditPick

Next, we want to test the voice experience on native devices, make it easier to add real cards, and improve how purchases are matched to reward categories. Also, have the detailed summary of which card benefits the most in overall catagories.