Mosaic

Kai Black posted an update — Oct 26, 2025 02:19 PM EDT

Partial demo link (internet sucks): https://media.oregonstate.edu/media/1_aytq6w63

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

Kai Black posted an update — Oct 26, 2025 02:18 PM EDT

Inspiration

Mosaic transforms group travel planning from chaos to consensus. Friends create a trip, invite their crew with a code, and everyone swipes through local restaurants, activities, and attractions, like Tinder for places. Our AI analyzes the group's collective preferences to generate personalized, day-by-day itineraries that satisfy everyone's tastes. No more endless group chats debating where to eat or what to do. Whether you're vegan, adventurous, or need the perfect coffee spot, Mosaic finds the hidden gems your entire group will love. Stop compromising. Start exploring together.

What it does

Mosaic provides a way for users to form groups with their friend and like locations that are based off geo spatial coordinates. Users can like these places and our custom recommendation system will generate a selection of locations based on the entire groups preferences.

How we built it

Mosaic is powered by modern day technology like ElasticSearch, OpenAI, and Supabase. Behind the scenes Mosaic utilizes OpenAI’s embedded vector generator to generate embedded vectors for users. When users like a post our backend goes through and updates these vectors to ensure that our system properly personalizes each feed for each user and group. This is all done by utilizing ElasticSearch’s cloudless infrastructure to host our vector database to allow for easy querying and vector manipulation. We also utilize Supabase to allow our system to store user information provided from Google APIs for authentication as well as location services.

Challenges we ran into

This was the first time any of us really handled a project that dealt this heavily with vector databases and AI powered technology. There was definitely a learning curve, but having the opportunity to work with technologies like ElasticSearch provided us all with valuable experience in the technology that dominants the modern day world. Other than that we ran into big issues with outdated documentation from google APIs, making it difficult to perform some actions that our app relied on such as gathering images for each location. (Also Calhacks wifi sucks)

Accomplishments that we're proud of

In general, we can confidently say that each member of our team tackled something new in this hackathon. Whether it be working on mobile development for the first time, creating a custom recommendation system, or even just participating in the hackathon, we each experienced something new and our proud of what we made at the end and the challenges we overcame.

What we learned

In general a lot of us learned about many different technologies, but the thing we all took away was how to produce a powerful application that utilizes AI to create something interesting. (Also learned that if you book a venue for 3000 people, you better hope the wifi works)

What's next for Mosaic

We heavily implemented Mosaic using ElasticSearch with Agent Builder in mind to allow for users to create a whole itinerary in one application, including booking hotels, flights, and finding fun events. Whilst we didn’t manage it in time during the Hackathon, we believe Mosaic has the foundation built already to truly create the final all encompassing travel planning application. Mosaic already has all the information needed for such an application, with more time, it can truly transform into something incredible

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

Kai Black started this project — Oct 26, 2025 12:29 PM EDT

Leave feedback in the comments!

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