Flightopolis

screenshot of the interface

Inspiration

We were inspired by websites like Skyscanner, which look for the cheapest price of a flight.

What it does

Flightopolis will ask users for their current location, desired destination and a range of dates they wish to fly out on. It will then respond with a date within the range that is predicted to have the cheapest flight. Additionally, it will also provide the flight details, such as flight time, airline, and the estimated cost.

How we built it

We used React to create the user interface and included Gemini in the backend with an API taking in live flight pricing from Google Flights and history data. The UI takes in the user input and sends it to Google Gemini, which will then process the information, turning it into a more computer-friendly language. The processed information will then be used alongside the API to determine the cheapest flight. Google Gemini will then again take the flight information, format it into a more human-like response, and finally send the response back to React to display on the UI.

Challenges we ran into

Integration of the front and back end with the API was our biggest struggle. Individually they were not hard to implement, but when we tried to put them together, many issues arose. React was written in Java, while the rest was in Python, so we did spend most of the time trying to get them to interact with one another. Another issue was connecting the API to the backend; it was especially difficult since we had limited to no experience with APIs before.

Accomplishments that we're proud of

We were very surprised to successfully connect the two interfaces to one another; it was a big step in the project when Google Gemini was finally receiving data from React and vice versa.

What we learned

The entire project was a learning experience for us all; it was the whole team's first hackathon, and everyone wasn't very experienced in coding. However, we went to the workshops and got to learn more about different frameworks, which we found very helpful. This was also our first time working with an AI such as Google Gemini, which we found to be extraordinarily helpful.

What's next for Flightopolis

Moving forward, we're hoping to polish the program a lot more. The core function of the program was achieved; however, there are many potential issues that we haven't gotten a chance to touch upon. We're also hoping to improve our chat responses to be more human-like, and also expand it to receive more than just flight input.