TripTailor

Inspiration

Often, we find that trip planning takes very long, especially because our sources are very scattered. We decided to make TripTailor because we want to simplify this process so that users can easily create trip plans while exploring different destinations they can go to in a specific location.

What it does

It is an interactive, intuitive, and intelligent trip explorer and planner. The explore and trip plan features work together to allow you to efficiently build a trip plan for any country in the world. Once you have logged in and selected a country of interest, we present you with the Explore page which is an interface that visualizes paths you can take as filters for narrowing down your choice of attraction. For instance, if you choose Canada, it may at first provide you with popular Canadian cities to choose from so attractions can be narrowed down accordingly. This occurs until the results are specific attractions, which you can select and add to your personal travel plan, located and managed on the Travel Plan page. The Travel Plan page allows you to get information about and delete selected attractions.

How we built it

Firstly, to make a high performance backend, we engineered an efficient SQL database structure. We used a ReactJS frontend and a Python Flask backend. We also used the PostgreSQL database to store queries. The frontend and backend communicate with each other to extract the information needed as requested by the users. We heavily utilized Postman to visualize the data that was being transferred to and from the database. Notably, we used Cohere’s generate API to incorporate AI to have it generate lists and descriptions for locations and attractions. Finally, we hosted the client and server separately using render platforms.

Challenges we ran into

Some challenges we ran into include finely tuning the settings in Cohere API to satisfy our requirements. For instance, adjusting the temperature and selecting the Cohere Generate API we wanted to use. In our case, we needed concise results, so we experimented with lower temperatures. We also had to make sure the organization of the database/queries was finely tuned by manipulating data by trial and error. Furthermore, interacting and connecting the Client and Server also had its challenges with synchronizing both interfaces and restructuring the data in such a way (ex. through JSON) that both interfaces can effectively interact with one another. Of course, we had to learn many new technologies very quickly which was certainly challenging, yet fun and rewarding!

Accomplishments that we’re proud of

We are proud that we were able to finish a project of this complexity within the time frame given. For some of us, this is the first time we made a website, let alone the fact that this is our first time meeting each other and working as a team! It was also amazing to be able to deploy our product and launch it on the web. Learning a new tech stack, using AI-driven tech, and applying it in less than 36 hours was an awesome experience, and we hope to continue to build on what we learned by implementing improvements to our product after UofTHacks!

What we learned

We learned new technologies such as SQL, Postman, Flask, and Cohere’s API through the backend work. We had the opportunity to engineer a SQL database structure from scratch for the first time. We also learned about UI/UX design by using HTML/CSS/JavaScript in React to build the interface. We learned how difficult simple tasks like making arrow visuals and working with SVGs are. We learned how to use unique tools like Render and ElephantSQL. But most importantly, we learned how to take on new challenges as a team through effective communication and task delegation!

What’s next for TripTailor

To continue TripTailor, we plan on adding more attraction categories to explore (e.g. Museums). We also plan on including more informative media (e.g. videos and images) to go along with our recommendations for a more informative user experience. We plan on using more APIs such as Google Maps API, to improve the geographical accuracy of our recommendations. We also would like to use the Cohere API in more ways, for instance with Co.Detect_language, where we can allow multiple languages to be used with the software. Finally, we also plan to make the platform highly responsive so that anyone can use it anywhere.

Built With

• co:here
• elephantsql
• flask
• git/github
• html/css
• javascript
• postgresql
• postman
• python
• react
• render