How to Use AI for Your Music Business (Not Your Music)

The line: human for the art, AI for the ops

Let me be clear about what this is and isn’t, because the word “AI” in music has earned a lot of suspicion, most of it deserved. I’m a producer. I have no interest in a model writing my songs, and if you’re reading this you probably feel the same. The flood of generated tracks clogging up playlists is a real problem, not a feature.

That’s not what this is about. A music career has two halves. One is the music: writing, playing, producing, the stuff you got into this for. The other is the business: metadata, royalty statements, release checklists, pitch descriptions, the same bio retyped for the tenth blog, the spreadsheet of who gets what. Most artists are drowning in the second half, which leaves less time for the first. That second half is exactly what AI is good for. Keep the art human. Hand off the admin.

One rule that runs through everything

Before any specific task, there’s a single rule that keeps AI useful instead of embarrassing: you supply the true facts, AI shapes the work, and you cut anything that isn’t you or that you can’t prove. Treat it as a drafting partner you direct, and never as a ghostwriter you trust blind.

The reason this matters so much for an independent artist is credibility. You don’t have a label’s name behind you, so the honest, accurate version of your story is the whole asset. A model will happily invent a stream count, a press quote, or a festival you never played, and it’ll sound completely confident doing it. If that lands in your bio or your pitch and someone checks, you’re done. So every number and every quote AI hands you gets verified against something real before it goes anywhere.

The admin and money side, where it saves the most time

This is the half AI is best at, because it’s reading and organizing dense information rather than inventing anything. A royalty statement is a wall of rows: per-track, per-territory, per-DSP, gross and net, three months late. Ask AI “what changed since last quarter, what looks off, and where is the money really coming from” and you get a plain answer in seconds instead of squinting at a spreadsheet. Same with streaming dashboards, metadata checks, and release admin: turning a mess into a checklist is exactly its strength.

The catch is the same as always: it can only work with the numbers you give it, and it can misread or round things, so you treat its summary as a fast first pass, not gospel. The full walkthrough, including the prompts that surface useful answers and the mistakes to watch for, is in using AI to read your royalties and streaming data.

The writing side: where AI drafts and you finish

The other half is words: your bio, your EPK, pitch descriptions, outreach emails, captions, a newsletter. Here AI is a cure for the blank page, and a trap if you stop there. The generic, hype-heavy, em-dash-ridden voice that screams “a machine wrote this” is the default it reaches for, and an editor or a fan can smell it.

The move that fixes it: feed it the real facts plus a couple of samples of how you talk, get the draft, then cut hard. Delete every sentence that could belong to any artist. Keep the specifics only you would write. What you’re left with reads like you, written faster. The bio and EPK version of this, with a before-and-after on the same artist, is in writing your artist bio and EPK with AI.

start your one-sheet from the real facts with the free artist one-sheet generator →

Vee vs a general assistant: which to reach for

There are two kinds of AI you’ll use, and the difference is what they can see. A general assistant like ChatGPT only knows what you paste in, and it fills the gaps by guessing. That’s fine for open-ended writing or brainstorming. It’s risky the moment the question is about your specific catalog or money, because it doesn’t have your numbers and will invent plausible ones.

Vee, the assistant built into Velveteen, is the other kind. It reads your real catalog inside the platform: your releases, tracks, pitches, earnings, and streaming data. So “what did my top track earn last quarter” or “what’s my next release and is it ready” comes back from your actual data, not a guess. It can also draft and update releases, build a pitch, and submit for distribution, with you approving each step. It won’t write your newsletter or read an outside contract, so for those you’re back to a general assistant. Match the tool to the job.

For the full picture of what the built-in assistant can and can’t do, see what Vee can do.

Where to start

Pick the task that’s eating the most of your time and start there. If it’s the writing, the bio is the usual first win, because you write it once and reuse it everywhere. If it’s the money, get AI to read your last royalty statement and tell you what moved. Either way, keep the rule in your head: you bring the truth, it brings the draft, you keep only what’s really yours.

And if a tool ever drifts toward writing the music for you, that’s the line. The whole point of handing off the admin is to spend more of your time on the part a machine can’t do.

try the kind of AI that does a real business task: draft your Spotify pitch from the facts →