DeepCut

Challenges:

• Tim Connors
• John Berryman

FUNCTIONALITY

Our proof of concept integrates with both Apple Music and Spotify. We grab the entire music catalogue of the user and create a weighting/reranking system based on the type of music the user prefers. We then weigh in the prompt from the user into the data stream (enhanced with an LLM), the location data, the weather data, the time of day data, and the variable of any other users on the platform they may want to collaborate with and provide a built out playlist back to the user (it currently is saved in their music provider of choice). Upon clicking the CTA it redirects to the connected music player. Users can switch between the two players, share their playlists/contextualized music preferences with other users. Users can blend their sound profiles together to make more nuanced, music informed decisions.

CUSTOMER VALIDATION

Potential users really loved our idea. They saw it as a much better pandora that empowered them to find music they typically wouldn’t come across. The natural intuitive design was very approachable and aesthetically pleasing for them. The ability to collaborate with their friends and make more nuanced playlists was a rich feature to them. The fact that the application was free to use and platform agnostic were also big value adds. Users of course preferred having the music player native in the application.
Design & UX

Our app leverages the traditional context window format, but is Siri/AI native. Leveraging voice as the user input is a much more seamless experience for a music curating experience. It’s like having a friend who’s a music aficionado in your pocket. The entire application was built around gestures, using as few button clicks as possible. Our proof-of-concept is less intuitively derived, as opposed to our MVP. Which is fully scoped.

FEASIBILITY

We see this as a free product to offer in the market. Being the first mover in the space built upon Apple Intelligence (which means we have no LLM API costs), and by leveraging the preexisting API’s for music streaming, our operating costs are negligible. We think our best path forward is to attack the market quickly and onboard users. We do see an opportunity to monetize by offering promotional services to local bands that have shows coming up in the market (Hey, this artist is playing at Basement East next week). It’s also a wonderful platform for indie artists to be discovered.

TEAM EXECUTION

We successfully have a proof-of-concept IOS app deployed. We have a backend webapp built with full functionality (which backend devs used internally). And we have the entire MVP frontend designed. We needed more time to scope out the music streaming from the API’s to have the player work in the app. Steven ate all the oreos.

TECH STACK

React Native, Expo, Figma, Gemini, Cline, Cursor, GPT-5, Veo, Vercel, Supabase, Claude Code, DoorDash, Sharpies, Paper, and Cd’s and Cline.
User Research Details
We presented the prototype to users to see how they would prompt the LLM. We then used those prompt cadences to create pretemplates for the application. User’s also expressed that they needed support for Apple Music and Spotify (platform agnostic).

NOTES

Once we made our repo public to submit it cleared our api keys so we’ll get it working about for the live demo

Ross used the Vivy frontend repo to create an opinionated React Native + Expo boilerplate project (but scrapped out all of the proprietary stuff) beforehand