Building the agentic
music industry.
A research lab and implementation partner. We build custom agents into your stack — and you own every one.
Your data stays yours — we never train on it. See our data boundary.
Used by teams at 8 labels, distributors & platforms
The gap
Your team has Claude.
It still can't do the work.
General AI knows everything about the world and nothing about your business. So the work that actually moves money — diligence, royalties, A&R — still lands on your team.
It doesn’t know you
Never seen your roster, your catalog, or your deal terms — so every answer is generic.
It can’t reach in
Can’t open a royalty statement, query a distributor, or touch Drive without you driving every step.
It can’t do the job
Catalog diligence, royalty ops, A&R, content batches — none of the real workflows are built in.
11
Open-source skills
19
Research pieces published
8
Labels, distributors & platforms
1
Label we run ourselves
How we work
We research it, build it, and run it in your stack.
Resources
Learn what's working.
Research, playbooks, and field notes from working with labels, catalogs, and platforms — shared in public, not behind a sales call.
Read our researchProducts
Build it yourself, or use the hosted version.
Recoup OS and open skills to run in your own stack — or Chat, the hosted workspace with nothing to install.
Explore productsConsulting
We implement it with you.
When you want us in the room: strategy, rollout, and custom agents built for your roster and workflows.
Work with usYou could wire this up in-house. Our bet is you won't.
Not because your team isn't good enough — because they're shipping releases today. The model team, the data plumbing, the eval loop: it's a project that never reaches the top of the list when there's a record to put out this week.
That's the work we do. We research what's working, build it in the open, and run it on our own label first — so the agents that land in your stack are already proven on a real roster.
Research
What's actually working in AI for music.
Catalog diligence, label operations, agent rollouts — we publish what actually works (and what doesn't) as we build with labels and catalogs.
Read the latestGatsby Grace
indie-pop / bedroom pop
Monthly Listeners
45,200
+12% MoM
Followers
28.1K
+8% MoM
Avg. Saves / Release
3,200
+22% MoM
Top Cities
Playlists
14
editorial placements (90 days)
Build
One engine behind every workflow.
This is the engine behind every implementation. Chat is a hosted workspace; the API and open skills let your team — or ours — embed music intelligence into Claude, Cursor, and your own stack.
MCP is the open standard that lets any agent — Claude, Cursor, or one you built — securely call Recoup's music tools without custom glue code.
Research Billie Eilish's streaming data and create a 12-piece content batch for her next single.
Recoup Research
Billie Eilish — 98.2M monthly listeners, +3% MoM. Top markets: US, UK, AU, DE, FR. 847 editorial playlists.
Recoup Content
Generated 12 assets — 8 short-form videos, 3 covers, 1 press kit. Optimized for Reels + TikTok.
Research Billie Eilish's streaming data and create a 12-piece content batch for her next single.
Recoup Research
Billie Eilish — 98.2M monthly listeners, +3% MoM. Top markets: US, UK, AU, DE, FR. 847 editorial playlists.
Recoup Content
Generated 12 assets — 8 short-form videos, 3 covers, 1 press kit. Optimized for Reels + TikTok.
Ask anything
The open skills it runs — free to install.
Open-source skills our own agents run every day. Install them into Claude, Cursor, or your own stack in seconds. Your private work stays yours — review our data boundary.
Music Research
View on GitHubArtist analytics, people search, competitive analysis, and web intelligence — the groundwork for catalog and deal research.
$ npx skills add recoupable/skillsOne command installs the whole open repo. Browse all skills
Prefer it done for you? The same skills run inside Chat, or our team sets them up in your stack.
Plugins marketplace
Going deeper? The native plugins add API-backed research and catalog-diligence workflows.
Data-room ingestion, royalty normalization, IC memos, and valuation analysis — for Claude Code, Cowork, and Codex.
Proof
We run our own labels.
Recoup Records and our artist Gatsby Grace use these same tools every day — to write release plans, audit catalogs, pitch playlists, and run the business. Every skill we publish earns its keep on a real roster first.
See how we dogfoodPartner
Most music teams are still planning their AI strategy. Start executing it.
Training and implementation from operators who build the tools — not slide decks. We work with labels, catalogs, and platforms inside their own stack.
Talk to us“Catalog diligence is one of the biggest pain points I have. Cut it down to minutes and it changes how we buy.”
Questions worth answering first.
How is this different from ChatGPT?+
ChatGPT is a general chatbot. Recoup is the music layer — skills, an API, and MCP integrations built for music work like research, release ops, and catalog analysis. Point your own agent at it, or have us wire it into your stack.
Who owns what we build — and how is our data handled?+
You do. The agents, skills, and workflows we build are yours, and they live in your stack or a repo your organization controls. We scope data access and reuse terms before work starts, and your private work is not folded into public skills without approval.
Do I need technical people on my team?+
No. We translate between the technical and business sides — that's the point. You leave with a plan your team can actually run, and the same skills work whether your team lives in spreadsheets or in a terminal.
What does it cost to start?+
The open skills are free, the hosted Chat workspace starts free, and consulting sessions start at $500. Implementation and partner deals are scoped to your roster and volume.
The next decade of music belongs to the teams that own their data.