travyl | Devpost

Inspiration

Humans are born explorers - always drawn towards exploration and discovery, especially when it comes to activities they are already passionate about. However, the biggest thing keeping us from dropping everything and backpacking through Europe is often the sheer lack of knowledge on where to go, what to do, and when to do it.

What it does

Our solution begins by surveying the user of their interests with a quick quiz. With careful prompt engineering we make an API call based on their selections to OpenAI, leveraging their LLM GPT-3. The user then has the option to select one of the five generated AA location options, listed from most to least relevant. Once again, the chosen location along with a carefully engineered prompt is used to make a call to OpenAI to generate a day-to-day itinerary for the given number of guests, on the specific days. To make it most convenient for the user, necessary flights can be instantly booked for this trip right through this app by making a call to the Flight Engine API. And there you have it, a highly personalized itinerary matching the interests of your group without having to do anything more than just answer a few quick questions.

How we built it

We gathered information about the user’s preferences through a quick quiz and utilized OpenAI to curate 5 possible travel locations for the user. Based on the travel locations, we made another call to OpenAI to generate a detailed itinerary of the entire trip. We then found departing and retuning flights to and from the user's inputted location using the AA Flight Engine API and presented them to the user.

Challenges we ran into

Determining the budget for the trip based on the flight distance, due to inconsistencies in the Flight Engine API Steep learning curve due to this being the first time many of our team members are working with the Next.js Framework

Accomplishments that we're proud of

Being able to successfully complete the project despite it being the first time with the tech stack for some members
Integrating an LLM to curate results
Plan an entire travel itinerary and not just book flights
Decreasing time to first booking for American Airlines

What we learned

Familiarizing ourselves with the tech stack
Integrating LLMs into a full stack solution

What's next for travyl

Implement additional API’s in our application such as Hotels.com or Uber in order to provide an exhaustive travel itinerary for the user and leverage cross promotional marketing for AA
Create an additional filter to generate trips in a specified budget
Integrate travyl to smart home assistants so customers can book an entire trip by just saying “Hey, Google plan me a trip through travyl”
Share-me feature to easily send your friends the itinerary for your next trip and to share straight on social media