Inspiration
A great instrument invites you to touch it. We wanted that feeling on a phone: the immediacy of a hardware sampler, the freedom to experiment, and a faster path from an idea to a beat.
CRATE starts with a simple question: what if describing a sound helped you find it, and folding your phone became part of playing it?
What it does
CRATE is an AI-assisted sampler and groovebox designed for iPhone Duo. Type a vibe, load a playable kit from the sample library, tap the pads, and shape the result into your own performance. The AI helps choose and arrange sounds; the performer stays in control.
In laptop mode, the lower display becomes a pad controller and the upper display keeps musical information in view. An audience-facing outer display gives the performance a visual presence. Hinge movement controls effects, turning a fold into tension and an opening into a musical release.
How we built it
Aradhya, Mahin, and Steven built CRATE in Swift and SwiftUI, with dedicated components for audio playback, sequencing, sample retrieval, AI orchestration, and the interface.
Our AI system splits the work: TypeSafe's Jev handles fast, structured musical decisions, while OpenAI models support more involved arrangements and variations. Local retrieval supplies the samples. We designed playback to stay responsive while AI requests run, with musical changes applied at bar boundaries.
RevenueCat powers CRATE PRO's subscription flow. The app checks the pro entitlement, loads the current offering, displays monthly and annual packages with store-provided prices, and supports purchases and restores. The custom black-and-orange paywall matches the instrument. After the initial three-DIG allowance, the app checks access before presenting the upgrade flow.
Challenges we ran into
Music makes latency obvious. A delayed pad hit or a change arriving halfway through a bar can break a performance, so we separated audio timing from AI requests and UI updates.
The Duo also required us to rethink layout. Controls have to remain reachable while the upper display stays readable, and hinge motion has to feel useful rather than decorative. We gave each display a clear role and treated the device's physical movement as an expressive control.
Accomplishments we're proud of
We turned an unusual form factor into a musical interaction. CRATE's fold is part of the instrument, and its displays serve the performer and the audience.
We're also proud of building a student-team project that connects native audio, a playable interface, AI-assisted sample selection, and a RevenueCat subscription flow in one app. CRATE is a tool for making and performing music, with an engineering foundation that goes beyond a generated-song demo.
What we learned
The most useful AI feature is the one that keeps a musician in the flow. Fast retrieval, predictable timing, and clear controls matter as much as the model itself.
We also learned that monetization belongs inside the product design. RevenueCat lets us keep subscription state, offerings, purchases, and restores together while designing a paywall that fits CRATE.
Try CRATE
TestFlight: https://testflight.apple.com/join/NTU9XppA
Install Apple's TestFlight app on your iPhone, open the invitation, and install CRATE if the beta accepts your device. Launch the app, use DIG to describe a musical vibe, play the loaded pads, and explore the sequencing and performance controls. The demo shows the Duo-specific fold and display experience; these interactions require the corresponding Duo environment.
Demo video: https://www.youtube.com/watch?v=tbkqIPaX97s&t=2s
Why CRATE fits the Next Gen Award
CRATE shows what a student team can build when ambitious hardware interaction meets careful software engineering. Aradhya, Mahin, and Steven built a native music instrument that connects sample playback, sequencing, AI-assisted sound selection, and RevenueCat subscriptions. Its strongest idea is physical: folding the Duo changes the performance, while the displays give the performer and audience different roles. The project combines creative product design with demanding timing and state-management work, and gives judges a TestFlight invitation and a working video demonstration rather than just a concept pitch.
What's next
We want to expand sample-pack import, stem separation, and performance resampling, refine the hinge controls on physical hardware, and learn from producers using CRATE in their workflow.
Built With
- avfoundation
- bitrig
- openai
- revenuecat
- swift
- swiftui
- typesafe
Log in or sign up for Devpost to join the conversation.