Inspiration
Our team has had many experiences that have gone something like this: We look for fun and cool locations for a place we are going to, but when we just search online, all we could find are Yelp reviews for boring locations. However, when we just add "reddit" to the end of our search results, we get a wealth of quality recommendations of cool locations from real people.
Thus, TravelBulb was born. We wanted a way to look for quality travel locations from Reddit, without having to sift through a bunch of Reddit posts and comments ourselves. We wanted a simple search engine where we could specify where we are going and what we want to do, and have it tell us the best locations itself.
What it does
TravelBulb acts like a search engine, but with some extra features. Users can select a location and some activities they want to do in that location. We will scour Reddit for the top posts and comments and by the end of the process, users will have a selection of recommendations. These recommendations could include shopping centers, trails, restaurants, and more.
How we built it
In terms of front-end, we utilized Bootstrap to streamline our creation process while building on top of a horizontal scrolling template. To make the design more eye-catching, we created a custom, uniquely TravelBulb color palette. Improving the user experience meant we needed a simple and easy-to-use interface.
Challenges we ran into
Throughout the project, we had troubles regarding front-end designs. We had finalized a draft which involving sleek styles and a scrollable user interface. Additionally, we had struggles trying to find the best framework which works with our web design alongside our backend code.
Accomplishments that we're proud of
The fact that our backend was finalized is our strongest point. Due to the bottleneck in our team being front-end, we were luckily able to fully implement our data scraping pipeline. In the future, this data will finally get used inside of our web interface.
What we Learned
We learned the front-end development has its challenges and that getting consistent outputs from LLMs can be a challenge, especially with its tendency to hallucinate.
What's next for TravelBulb
We want to compile these Reddit posts into a living knowledge base in the future to enhance search functionality and allow for easy updating of new information from new Reddit posts.
Built With
- langchain
- python


Log in or sign up for Devpost to join the conversation.