Amazon Music

Design at Berkeley ran its annual designathon on April 13, 2025, with Amazon Music as the partner and the judge. 55 teams and more than 200 participants made it the largest designathon in the Bay Area.

The prompt was broad on purpose. Amazon Music asked us to design a global experience that closed the gap between listeners and deepened how people engage with the platform. We had 12 hours to go from a blank brief to a defended concept, and the people scoring us were the same people who would decide whether the winning concept was worth building into the real app.

Before we could design anything, we had to figure out what was actually broken.

Streaming platforms have gotten good at taste and bad at timing. They can tell you your favorite genre, but not that it’s 7 AM on a Monday and you have a commute, not a party.

That missing layer, context, is where personalization actually breaks down. The same song lands completely differently at a workout, a study session, and a wind-down, and none of these platforms read the difference.

So we reframed the brief into a question we could actually design against.

How might we help users explore global music and content through personalized, context-aware recommendations?

With 12 hours on the clock, we had no time for original user research, so we grounded the concept in market and competitive evidence instead. I ran that research.

My first decision was what signal to build on. I set one bar: it had to already exist, and come from real behavior, not a habit we’d have to invent.

1.42B

people actively use Google or Apple CalendarStatista

$103B

global music streaming market by 2030Grand View Research

The calendar cleared the bar instantly: a billion-plus people already mapping their days hour by hour, and no platform touching it. In a $103B market where the entire fight is over relevance, the richest context signal in a user’s life was sitting unused.

But unused was a hunch. Before I’d build on it, I mapped every major platform against the contextual features they’d actually shipped.

Calendar was the only column no one had touched. Smart discovery was universal. Social was split between Spotify and YouTube. Location was Spotify’s alone. Calendar was the one contextual signal no platform had claimed.

Knowing the opening was calendar-shaped, we ran a round of ideation on more ways music could plug into a user’s real-world context. We came out with eight concepts and plotted them on two axes: how much engineering each would take, and how delightful and unexpected it would feel.

The top-right was where the fun ideas lived. AI trip soundtracks, live audio drops from local artists at your destination, all ideas that demo well. Calendar syncing sat in the opposite corner: higher effort, lower novelty, the least delightful thing on the board.

We picked it anyway.

The flashy ideas were tempting. But delight you can’t defend is a demo, not a product. Calendar was the only idea sitting on a data signal a user already owned, and the only one that got stickier over time. Connect it once, and it gets more useful every week.

We plan trips, schedule workouts, mark the dates that matter, but discovery ignores all of it. The playlist on the morning of a long flight is the same one you’d get on a random Tuesday. The calendar is the one place people already record their real-world context, in detail, without being asked, and no platform was reading it.

Least exciting, yet the most defensible, and the signal closest to how people actually live.

Everything up to here was a case for the calendar. The design is where that case had to turn into something a person could open, tap, and understand in a second.

The architecture

The sync is the whole product. Nothing works until a calendar is connected, so the Calendar tab doesn’t exist until it is. The feature only takes a slot once the user is synced.

Lo-fi screens

My teammate and I took charge of the product design. We worked in grayscale first, mapping out where the event sits, where the playlist attaches, and how the drawer opens. Structure had to hold up before any branding made it look finished.

Branding

I kept Amazon Music’s existing visual language, the signature teal, Amazon Ember, their existing components, so the concept read as a native feature and not a rebrand.

Calendar-aware listening inside Amazon Music. Connect a calendar once, and every event on it comes with a soundtrack.

Connect once. The first screen is the entry point and only does one thing, syncing Google or Apple calendar. Nothing else exists until a calendar does.

Then your day fills itself in. Football practice at 1pm gets a hype mix. An economics study session at 5pm gets a study mix. Tap any event to open its playlist, then tap again for the tracks. Three taps from a calendar entry to music that fits it.

And it finds you where you already are. A lock-screen prompt before practice starts, and a card for a hype playlist when you tap it. A list of paradise-y tracks when you’re vacationing in Bali. You don’t have to go looking, because the recommendation arrives when the event does.

Or even where you left off. The feature tracks more than events. Stopped an audiobook mid-chapter? It comes back as a resume card, a second, smaller behavior running on listening history rather than schedule. Music was the obvious case for calendar-aware listening. Audiobooks were the proof it didn’t have to stop at music.

Every suggestion shows its work. Generated from your calendar. Created from the event FOOTBALL PRACTICE. One tap views the calendar entry behind it. Personalization that reads someone’s schedule should be able to say where it got the idea.

2nd

of 55 teams

200+

participants

12 hrs

brief to pitch

Our team placed second at the largest designathon in the Bay Area, judged by the Amazon Music team we’d spent twelve hours designing for. The people who’d implement it into their product sat through fifty-five pitches and put ours second.

This was my first designathon. Placing second in a room that size, judged by the company itself, is a result I’m still proud of. I couldn’t have gotten there without my teammates, Sarah Mou, Arshia Narula and Akshat Parikh.

I never designed the permission ask

The first screen offers Google or Apple and assumes yes. But a calendar is the most personal data on someone’s phone, and I built the entire product on top of an integration whose trust moment I skipped entirely. How much does Amazon Music actually see? Can you connect one calendar and not another? That flow is the highest-risk unsolved piece of this concept, and under twelve hours I chose the payoff screens over it.

I validated a concept, not an experience

Nothing shipped and nothing was tested. We validated a concept with stakeholders, not an experience with users. If Amazon ran it, the number I’d watch is session-start rate on calendar-linked events: how many of the events the sync attaches a playlist to actually get played. That’s the leading indicator for whether context-aware recommendations beat the home feed, and it’s measurable in a week.

I answered half the brief

Amazon asked how they could build global experiences for creators and listeners, and I only built for listeners. The creator side, artists surfacing music into moments people are already planning for, is a whole product I never touched, and one I’d be eager to explore.