planeIt

PlaneIT helps groups find the perfect trip by matching personalities and budgets with destinations using AI—no stress, just smart, affordable travel everyone loves.

Comment

Inspiration

The inspiration for this project was skyscanners challenge (as you might have noticed). Planning group trips has always been chaotic. So I wanted to make this easy, and even fun like a game.

What it does

planeIt lets you form travel groups that can be shared via QR code or a written code. Once a group is created you can see every participant and if he or she has already answered the quiz. The quiz consists of five very light questions about traveling and personality, no need to think too much. After a user takes the quiz planeIt generates a persona for them and selects the most appropriate cities for them to visit. When everyone has their personas, the intersection of the recommendation list of each user is calculated, and all the users play a Tinder-like game where they can like or dislike cities based on the description and price of flight from where they are at. At the end a podium with the three most liked cities is showed.

How we built it

To calculate which city a user is most prone to like I generated a user persona and got its embedding (using openAis API) and did the same to each city, then I used the Cosine Similarity to calculate how similar the embeddings were and went from there. To obtain flight prices, city pictures and somethings else I used various APIs.

Challenges we ran into

Every challenge possible really. To begin with I came all alone, since the people that signed up with me were rejected. Furthermore, I am only on the second year of university and right now competing with people studying majors. Since I was alone I had to code every possible moment to be able to deliver the project, and sadly sacrifice the quality of my code for deadlines. It is also the first time I build a full stack project by myself and implement the AI techniques mentioned before. I faced many more challenges in this hackathon, but I am happy with my results.

Accomplishments that we're proud of

The UI looks very good in my opinion, and I really think my project is useful. I plan on deploying it as soon as I can.

What we learned

As I said before the techniques I used with AI were new to me, so I learned a lot about RAG and embedding comparison. Also for the first time I connected a backend made by me to a frontend made by me, that was really exciting. Lastly, I learned what it is to live a Hackathon since this is my first one, I will be coming back next year for sure.

What's next for planeIt

As mentioned before the goal is to polish some minor details and deploy it for general use. In the state it is right now it can be used and useful, however I do recognise that this is a MVP and still need to evolve, so I'll see what I can do about that ( ;

Built With

Share this project:

Updates