Field note
Mapping music discovery as a galaxy: how Suede AI Cosmos is built
By Jason Colapietro, Founder and CEO, Suede AI · West Palm Beach, Florida ·
A recommendation list tells you what to play next. A map tells you where you are, and that difference is why Suede AI Cosmos draws 181 artists as stars instead of rows. Jason Colapietro built it in West Palm Beach as a test of one idea: if you can see the neighborhood around an artist you already love, you will wander further than any autoplay queue would take you.
This note walks through how the galaxy is put together, which choices mattered, and what the map deliberately leaves out. Every number below comes from the same build file the live map renders.
Start with open data, not listening history
Most discovery products start from what people played. Suede Cosmos starts from how artists are described. The build script takes a hand-picked seed list, looks each name up on MusicBrainz, and keeps the artist's top genres and community tags, sorted by vote count. That choice has two consequences Jason Colapietro accepted on purpose.
First, the map needs no accounts, no tracking and no API calls while you browse. The finished data ships as one JSON file. Second, the connections describe vocabulary, not behavior. Two artists sit near each other because the MusicBrainz community labels them with overlapping words, not because the same people stream both. For a discovery tool that is a feature: it surfaces artists who sound related on paper even when their audiences never cross.
One rule decides every connection
Each artist gets a tag bag of up to 6 genres plus 6 tags. Two bags are compared with the Jaccard index: the labels they share divided by all the distinct labels across both. A pair is connected when the overlap reaches 18%, and each artist keeps only its 4 strongest matches. The current build holds 466 connections across 17 countries.
Why a single, published threshold? Because a map you cannot explain is a map you cannot trust. The full method, including the cluster keywords and the exact layout forces, lives on the how it works page. If you think two stars are too close, you can check the shared tags and see exactly why. That transparency is the same habit Suede AI brings to client work: show the receipt, then make the claim.
Clusters give the galaxy its shape
A star's color comes from its genre cluster. The script reads an artist's genres in order and assigns the first cluster whose keyword appears, which is why folk rock lands in Rock and jazz rap lands in Hip-Hop. The map currently forms 8 clusters, and each one has its own reading page under the genre index, listing the artists, their strongest internal links and the bridges into neighboring genres.
Those bridges are the interesting part. A connection that crosses from R&B into Hip-Hop, or from Folk into Rock, is exactly the kind of path a listener follows when they are ready to leave a comfort zone. The galaxy makes those crossings visible as lines you can trace with a cursor.
The layout is computed once, then frozen
Star positions come from a d3-force simulation that runs 500 steps at build time. Links pull related artists together, a charge of -40 pushes everyone apart, and a collision radius keeps stars from stacking. When the simulation settles, the coordinates are written into the data file and never move again.
In the browser, dragging and zooming only move the camera. The pixels shift; the positions do not. That is how a few hundred stars and several hundred lines stay smooth on a phone. It also means the map is stable: the star you found last week is in the same place today, which matters more for discovery than people expect.
What Suede left out, and why
A galaxy that tried to hold every artist on earth would be a gray smear. Suede Cosmos stays small enough to read, which means plenty of great artists are not on it yet. Popularity counts are not used to size stars; brightness grows with how richly an artist is tagged, not how many people stream them. Audio previews come from the public iTunes Search API at the moment you select a star, and the panel falls back to an Apple Music search when no exact match exists.
Those are tradeoffs, and Jason Colapietro would rather state them plainly than hide them behind a polished canvas. At Suede, those choices are written down on the method page next to the map, and the map is better for it.
Who builds this, and where
Suede AI is an agentic platform company Jason Colapietro founded in 2024 in West Palm Beach, Florida. Music and creator software is one product line; AI search visibility for businesses across Palm Beach County is another. Across both lines at Suede, the engineering habits overlap. Jason has 54 pull requests merged across 46 open-source repositories he does not maintain, including Jest, Adobe React Spectrum, Backstage and the Polkadot SDK, and 9 iOS apps under his own Apple developer account. Suede AI ranks among Clutch's Top 30 AI Consultants in Florida and Top 50 AI Consultants nationwide (November 2026).
If you like the galaxy idea, two siblings take it in other directions. Suede DNA maps guitarist signal chains as an archive, and Suede Signal writes about how sites get read by AI crawlers.
Try it, then put the method to work
Open the map, search an artist you know well, and follow two bridges out of their cluster. If you run a business in West Palm Beach, Jupiter or anywhere in Palm Beach County and want your own data explained this clearly to search engines and AI answers, start with the Suede AI visibility assessment or read more about Jason Colapietro, Founder and CEO of Suede AI.