Inspiration

Credit cards can be incredibly valuable tools when used responsibly. Some people, often called credit card hackers, go to great lengths to maximize rewards through sign-up bonuses, perks, and benefits. But not everyone has the time (or patience) to dive into that level of research. That’s why we built OptiCard—a tool designed to help everyday users quickly find the best Capital One card for their needs.

How it is Built

*Disclaimer: In real-world use, OptiCard could be integrated directly within Capital One, using real customer transaction data to match users with the most suitable credit card offerings. OptiCard’s backend is powered by Python and leverages ADK to connect with our AI agent. While we would have loved to use real Capital One API data, access is restricted. To work around this, we built a simulator that generates randomized purchase data, categorizes transactions, and outputs them into a JSON file. This JSON is then processed by the AI agent, which analyzes the spending history and recommends the top three card options tailored to the user. On the frontend side, we used React to create a clean, interactive user experience. This React page shows the three most optimal cards for your spending habits. It shows each one at a time, showing the name of the card, how it looks, and the reasons our AI agent figured out. The page includes arrows to swap over and look at the other cards recommended by our AI agent. The front end interacts with the back end by reading JSON files and presenting this information. This allows the page to change to fit any person's spending habits, given that we have access to that information.

Biggest Challenges

The biggest challenges we faced came from the frontend and integration layer—specifically, ensuring smooth communication between the agent and the backend, as well as perfecting the JSON formatting. The backend logic itself was relatively straightforward.