What if the whole operation ran on one system?
We built MusicPeaks to run our own label — releases, content, spend, royalties and the daily list of what to do next, in one record that AI agents can read and act on. People keep the taste, the relationships and the approvals; automation carries the busywork. Release operations is running today; the rest is coming out of the workshop. Now opening conversations with labels and management companies who want to work this way.
We built this to run our own label.
MusicPeaks started inside Slaight Music, a working record label, as the system we needed and couldn't buy: every platform, every task, every dollar and every ID in one place, with automation doing the chasing. MCMXVI was the second team on it. The first thing we automated was release marketing, because that is where the platforms decide who gets heard — and three things are true about that fight.
The wall
Over a hundred thousand new tracks reach the platforms every day, on top of every song ever recorded. Nobody hears yours by default.
40% and growing
Roughly forty percent of streams on the major platforms are algorithmically driven — personalised playlists, autoplay, radio, the AI DJ — and the share grows every year. Even the “intentional” sixty percent is shaped by what the algorithm served first.
Everything is a signal
Delivery timing, pitch text, artwork, saves, skips, the gap to the next release — everything the team does, or doesn't do, tells the platform something. Reading how it sees you is the first job. Shaping what it sees is the second.
Every release sends a signal. Send the right one.
A black box decides who hears the music, and everything the team does — or doesn't do — is a signal to it. The release-operations module reads how the platforms see a release and runs it, from setup to wrap, so the signals you send reinforce the ones already working. The seven stages below are screens from the system as it runs today, on our own releases.
Before you can be heard, you have to be legible.
Eight platforms, one readiness check. Artist IDs matched, pages connected to the ad account, pixels firing, catalogue imported, metrics flowing — each one verified against the platform, not ticked in a spreadsheet. Chartmetric is wired the same way, so the algorithmic picture we build next starts from the real accounts. Everything else reads from here.
Hear the song the way the algorithm does.
After a track is delivered, the platform analyses it and places it among the songs and artists it already knows — sonic similarity first. We run the same kind of analysis before release: genre, mood, character, energy, key and tempo from the audio itself, then the similar tracks, artists and playlists around it. Positioning text, pitches and bios are drafted from that analysis and the artist's own notes, then edited and approved by a person. Our working theory: when the submission text matches what the platform already heard, it confirms the placement. It costs nothing to test.
- positioning drafts approved, 2 awaiting review
- 3 of 5
- similar tracks and playlists surfaced per song
- 20 + 20
Find the fans who are already halfway there.
Artists that recur across the analysis become candidates. We add “fans also listen to” two hops out from the artist's own profile, up to three targets the team names, and Chartmetric's numbers for all of them — then prune the set into one or two clusters that stand for a real fan community, labelled by the micro-genres that keep surfacing. For each release the team picks the active cluster and we add the social signals that reveal its demographics, geography and lane.
- cluster confidence
- 76%
- retained artists · graph links · growth targets
- 6 · 12 · 3
A plan that knows its budget, its assets and its deadlines.
Once the positioning is approved, we generate a Marketing Context — a diagnostic snapshot of what's true about the release right now: strengths, opportunities, what's in motion, what needs attention. From it comes a plan with options for the budget and the assets each option needs. The team approves the items it wants; approved spend is tracked against the budget, and a release checklist keeps the tasks, owners and dates current, with automations nudging whatever is blocking the next step.
- items approved, 16 superseded
- 4
- of a $2,000 budget committed
- $1,300
Everything you've made, and everything you could make from it.
A shared cloud folder is scanned once a day, so every asset the artist and the team upload is ingested, analysed and ready to use — and the social promo tab tracks what actually got posted. Short-form video is where the volume goes: through Flowstage, one performance clip becomes a run of vertical edits with different hooks, captions and lyric styles. A person reviews every render — keep, adjust or discard — before anything is scheduled.
Human-made music. AI on the busywork. The artist's footage, the artist's words, a person's taste.
Popularity is the number that moves the others.
Streams, campaigns and smart links, read from the platforms daily — popularity with its full history, spend against results down to the ad set in the currency it was spent, and the smart-link funnel split paid versus organic. Of all the numbers, Spotify popularity at the track and artist level is the one we watch most closely, because in our experience it is the best read on whether the algorithm is taking notice. What we've observed: front-load the spend, and have the next track ready before this one peaks.
- Spotify streams by day 27
- 32.2K
- of smart-link clicks became service clicks
- 55.6%
Every release should make the next one smarter.
When the marketing phase ends, the cycle closes with a sourced record rather than a status change: what ran, what it cost in the currency it was spent, what the platforms reported, and what stays unknown. AI compiles and compares; it may not erase provenance, silently reconcile conflicting systems, or turn missing evidence into fact. The reviewed lessons are what the next release's plan starts from.
- campaigns reconciled against the source
- 4 of 4
- spend kept in the currency it was spent
- CAD + USD
One system, rolled out in modules.
Release operations is the first module out of the door. The rest of the chain runs inside our own label today, at different levels of polish, and each module ships when it has earned it. Every one reads and writes the same record, so a release, its content, its spend and its results are never re-keyed and never lost between stages.
Release operations
Running today. Setup, Strategy, Audience, Plan, Make, Measure, Wrap — the seven stages above, on real releases, with a sourced record at the end of every cycle.
Content generation
In daily use, being polished. Short-form video batch-produced from the artist's own footage, positioning and pitch copy drafted from the analysis, assets ingested from a shared folder — with a person approving every piece before it goes anywhere.
Spend and finance
In use. Campaign spend recorded in the currency it was spent, reconciled against the platforms and tracked against the budget the team approved — so you can see what a release is costing while it's still running.
Royalty accounting
In use inside our own label, still being polished. Royalty statements brought into the same record as the spend, so income and cost sit side by side per release. Offered when it's boring — the standard for anything that touches money.
The morning brief
In development. The system reads everything above and tells you what to do today: what's blocked, what's due, what's drifting, what changed on the platforms overnight — and chases the loose ends itself.
Bespoke modules
By conversation. The same infrastructure, built around a workflow that's specific to your company. If you run something repetitive that no product fits, that's a conversation we want to have.
Every repetitive operation, captured.
Underneath every module is the same discipline. Each repetitive operation is written down as either a deterministic automation — the rules — or a probabilistic LLM skill — the judgment calls — and connected to the platforms by API wherever one exists. Every task, ID and piece of metadata is tracked in one record, so agents can act on it and people can see what they did.
Deterministic automations for the rules — syncs, checks, reconciliations, reminders
LLM skills for the judgment calls — drafting, tagging, comparing, summarising — reviewed by a person
APIs wherever a platform offers one; a scheduled read wherever it doesn't
Every task, owner, date, ID and metadata item tracked in one record
Agents with access to the whole record, and an approval gate on anything that spends money or speaks for an artist
Provenance on every automated action, so a person can see what ran and why
One track. Dozens of posts.
Short-form content is the most labour-intensive part of a modern release. MusicPeaks batch-produces vertical edits from the artist's own footage and brand, so a release's worth of content is made in an afternoon — and a person curates every batch before anything posts.
Your footage, your look
A mood board per facet of the artist — their sound, influences, world — built from their own clips, B-roll and photos. The artist's real content, edited at speed, not synthetic footage.
Volume from one track
Each release is auto-transcribed and cut into sections, then crossed with hooks and visuals — so one song produces dozens of distinct vertical edits.
Hooks from the brand layer
On-screen text hooks are drafted from the artist's own talking points and brand voice — the same layer that powers their pitches and bios.
A human keeps the taste
An operator reviews and curates every batch before anything posts; posting itself stays manual. The tooling removes the grunt work; the team keeps the judgment.
An audience that's actually yours.
Streaming and social rent you access to your fans. A mailing list is the one audience no algorithm can take away, so we set up the artist's own site, newsletter and memberships on Ghost — the list, the relationship and the revenue belong to them. The same goes for the operating record: the intelligence, the plans and the results are yours to export.
One platform, three jobs
Website, newsletter and paid memberships in one place — no stitching together Mailchimp, Substack and a separate site. Publish a post to your site and email it to your whole list in one action.
The list is yours, exportable
Your subscriber list lives in your account and exports any time. No platform holds your audience hostage; if you ever leave, you leave with everyone.
You keep the money
Paid tiers and subscriptions run through your own Stripe account — Ghost takes 0% of your revenue.* Compare that to the cut a Patreon or Substack takes.
* 0% Ghost platform fee — standard Stripe processing fees still apply.
Twelve tools and a spreadsheet, or one system that knows the whole operation?
Most labels and management companies run on a stack: a dozen subscriptions, a shared drive, the spreadsheet that reconciles them, and one person who holds it all in their head. An agency will run the release from their inbox. MusicPeaks is the operation itself — one record, automation on the busywork and, when you want it, our team — on infrastructure you can leave with.
Setup. Plan. Make. Measure. Wrap.
Every track moves through the same five stages, and every stage sends a signal. MusicPeaks runs the cycle — software and, where you want it, our team — as one connected record rather than a pile of features you assemble yourself.
Setup
- Artist, accounts & access
- Platform readiness
- Context & source data
Plan
- Acoustic analysis
- Positioning & strategy
- Marketing plan & approvals
Make
- Creative assets
- Short-form content
- Campaign & smartlink preparation
Measure
- Streaming & fan signals
- Campaign performance
- Smartlink & conversion results
Wrap
- Sourced closeout
- What ran and what it cost
- Rules for next time
Questions we get asked
What does MusicPeaks do?
MusicPeaks is the operating system we built to run our own label. It brings the whole operation — releases, content, spend, royalties, tasks, IDs and metadata — into one record that automation and AI agents can read and act on, and it's being opened to other labels and management companies one module at a time. The first module, release operations, runs a release from setup to wrap on how the platforms actually see an artist; the rest of the chain is listed above with its current status. Part software, part our team, while we prove each module.
Is this just another dashboard?
No. A dashboard shows you a status field. MusicPeaks checks whether the work actually happened — whether the smart link was provisioned, the pixel is firing, the campaign was recorded with its real spend — and turns the gap into the next action. Then it keeps the record, so the next release starts with what the last one taught you.
How is this different from a tool like Symphony or Chartmetric?
Those are single tools you still have to operate and stitch together. We read them — Chartmetric is one of our sources — and coordinate the whole release across them. Fewer logins, less manual chasing, and a record that survives the release.
Is this like intellijend, ToneDen or Hypeddit?
No. Those are self-serve tools for an artist running one lever — Meta ads pointed at a smart link, with a popularity graph — and if that's what you need, use one. MusicPeaks is built for the company around the artist: labels and management companies with a roster, a budget and an operation to run. It isn't self-serve, and it isn't trying to be.
Does the system run itself?
No, and we don't think it should yet. AI compiles, compares, drafts and monitors; a person approves anything that spends money, speaks for the artist or changes the plan. Our rule: AI may not erase provenance, silently reconcile conflicting systems, or turn missing evidence into fact. As each workflow proves itself, more of the routine work moves to automation.
What does this mean for the artist's music?
Nothing, unless the artist wants it to. The artist supplies the music, the meaning, the taste, the references and the art direction. AI works on the business around it — coordination, analysis, transformation and repetitive execution. Human-made music. Agent-ready operations.
Do I own my data and my audience?
Your accounts, assets, audience relationships and source data stay yours. What MusicPeaks builds on top — the release record, the decisions, the learning — is yours to export in documented formats, with a defined exit path. We don't hold your operational memory hostage. Your website, mailing list and memberships run on your own Ghost site and your own Stripe account, so the list and the revenue belong to you.
Can you really make videos at scale?
Yes. We produce and export vertical videos in volume through Flowstage, built from each artist's own footage and brand — one track becomes dozens of edits. It's a batch editor, not AI-generated footage, and a person reviews and curates before anything posts. It's one module inside Release Operations, not the whole offer.
What integrations are available?
Spotify for Artists, Chartmetric, Cyanite, Meta, TikTok, Feature.fm, Music Tomorrow, Ghost, YouTube and Flowstage — read automatically into one release state.
What about finance and royalties?
Spend tracking is in use: campaign spend recorded in the currency it was spent and reconciled against the platforms. Royalty accounting runs inside our own label and is still being polished; we'll offer it when it's boring, which is the standard for anything that touches money. Both are listed above with their current status, and we'll keep those labels honest.
Can you run releases for us, or just set things up?
Either. We can build the system and hand it over, or run releases, campaigns and content with you. While we open up, most teams do a mix — and we record how much of the work still needs a person, because that's what we're proving.
Can you build something bespoke for us?
Yes, within reason. The infrastructure underneath MusicPeaks — the record, the integrations, the automation and skill layer — was built to be shaped around a workflow. If your company runs something repetitive that no product fits, we can build a module for it on the same foundation. That's a conversation, not a form.
Who is this for?
Labels and management companies with more than one release moving, a marketing budget, and an appetite for AI- and automation-driven ways of working — but no interest in building an in-house operations department to get there. It's not built for a solo artist's first release.
When is MusicPeaks not the right choice?
A solo artist with one release and no team; anyone who only wants to run Meta ads; anyone who needs to self-serve today; or a company that isn't ready to give a system access to its platforms and its numbers. We'll say so on the call rather than sign you up.
How do I get access?
By conversation. We're opening slowly and deliberately, with a small number of labels and management companies, one module at a time — and we'd rather be honest about what's ready than sell you a roadmap. Start a conversation, tell us about your roster and how you operate, and we'll tell you what fits now, what's coming, and when we could start.
Can it fit how we already work?
Yes. We build around your roster, your platforms and your workflow rather than forcing you onto a fixed product — and we tell you up front which accounts we need access to and why.
Built for where music is going.
The next few years belong to small teams coordinating people, platforms and agents around one durable operational memory. Humans keep authorship, taste, relationships and approval. Agents carry more of the research, monitoring, administration and execution. Every completed release should leave the team — not just the software provider — better equipped for the next one.
The artist stays the author
The artist supplies the music, meaning, references and art direction. AI works on the business around it, and never quietly becomes the creative principal.
Your memory works wherever your team works
We're designing the release record so it can be read by your team and, when you authorise it, by your own agents and tools — with scoped permissions, provenance and audit trails. Exports first; API and agent access as they're proven.
More automation as it earns trust
Every workflow starts with a person approving. As a workflow proves dependable, more of it moves to automation — measured, not assumed.
Start a conversation.
We're not selling a self-serve product, and we're not in a hurry. We're talking to labels and management companies who want to run this way — on the system as it stands, on the modules still coming, or on something built for how you work. Tell us about your roster and your operation; we'll tell you honestly what fits now and what doesn't yet.