WTF IS MONAD

Inspiration

We love indie night, so we wanted to make something they would use or find cool. Haskell is an unconventional tool, especially when used as a web app's backend and fulfils the hackathon's theme of unconventional connection.

What it does

In the words of Claude

a proper unconventional RAG system built in 24 hours using Haskell of all things

We take the official WTF is indie playlist and a live webcam stream from a DJ's laptop to compute a real-time crowd energy score %, and then suggest/autoplay the songs that are most likely to please the crowd and increase crowd energy!

How we built it

We split it up into three sections:

React JS frontend and Spotify Client Control API
Fast API orchestrator connecting Spotify Metadata API (getting song lists and metadata( to rest of project
Haskell RAG backend to compute a heuristic using a CSP and pipes the top 10 to Gemini API

Challenges we ran into

RAGging with Haskell and connecting the Gemini API to Haskell. We think it speaks for itself.

Accomplishments that we're proud of

Laying out an initial MVP, building it fully + extra features

What we learned

Never use the Spotify API or Haskell for REST APIs (not sure which is worse)

What's next for WTF IS MONAD

Transitions between songs (DJ announcements using ElevenLabs)
Improve heuristics for picking songs (e.g. statistics not just from Spotify but also stats.fm)
Use better LLM models for picking top 10 (maybe host locally to save costs?)
Get verified on Spotify to allow other users to use our live demo
This could be made into a study tool to suggest songs to keep you engaged and being productive (e.g. LoFi playlists)
Actual Indie night??