CrediWise - Use the right card, every time.

Chrome Extension
detects checkout pages, suggests the best credit card via popup
Overall savings using the tool
Dashboard
Initial setup to let the tool know what cards you use

Inspiration

Every checkout feels like a guessing game: “Which card should I use?” Most of us don’t know which of our cards gives the best rewards or cashback at that moment. Studies show that people miss out on millions in rewards because they don’t optimize card usage. We wanted to solve this problem in a way that’s effortless — no manual lookups, no entering merchants. That’s how Crediwise was born: a Chrome extension that’s always there at checkout, making sure you swipe smart and save more.

What it does

Auto-detects checkout pages across domains (Amazon, Spotify, Delta, etc.).
Instantly recommends the best card to maximize rewards/cashback, factoring in base rates and public offers.
Provides a savings summary dashboard: How much you’ve saved so far, What other cards would have offered, Category breakdowns and overall performance
Tracks lifetime savings and generates AI-powered insights about spending patterns.

How we built it

Extension (MV3): Detects checkout flows, extracts domain/total, and injects a non-intrusive banner with the best card recommendation.
Backend (Flask, Python): Scrapers (Bright Data MCP) pull public offers and rotating categories, normalize into structured data, and refresh every few hours via GitHub Actions.
Database (Supabase, Postgres): Stores offers and user profiles. Redis (Upstash) used for fast caching.
Dashboard (React + Tailwind): Shows savings summary, category breakdown, card comparisons, and AI-powered insights. Designed for auto-visualization: line charts for trends, pie charts for proportions, bar charts for comparisons.
LLM Integration (LLama): Used for parsing unstructured offer terms & conditions into structured rules (merchant, min spend, expiry). Also powers AI-generated insights in the dashboard (e.g., “Most of your savings this month came from streaming — consider adding a dining rewards card next”).

Challenges we ran into

Offer scraping: Card issuers format offers inconsistently. Normalizing them into one schema required careful parsing.
CORS and extension security: Communicating between extension, backend, and UI meant carefully managing Chrome MV3 permissions.

Accomplishments that we're proud of

Built a full-stack system in a short timeframe: Chrome extension + backend + scrapers + dashboard, all working together seamlessly.
Ensured the system is privacy-first — no sensitive card data stored, only issuer/rewards profiles.

What we learned

Using Bright Data MCP effectively for real-time scraping of public offers and handling differences in site structures.
How to work with Chrome Extension MV3 APIs (content scripts, background service workers, messaging, and permissions).

What's next for CrediWise - Use the right card, every time.

Student-focused mode: Help college students optimize spending on tuition, rent, and subscriptions
Issuer integrations: Connect directly with card networks for real-time, user-specific offers.
Mobile app version: Extend functionality beyond desktop Chrome.