It was close to midnight, and I was trying to hold three releases in my head at once.
One was due at the distributor that week, which meant the final master, the artwork, the metadata, and the credits all had to be locked, because a delivery deadline three weeks out from release day doesn't move. Another was already out and mid-promo, waiting on me to line up the next round of posts. A third had just been signed, and the artist was waiting on contract details and a list of assets I still hadn't sent. Each one sat at a completely different stage, and each one needed something different from me, right then.
None of it was hard. That's the part worth sitting with. Not a single one of those tasks required skill or talent. They required someone to remember them, in the right order, at the right time, across releases that were all moving on different clocks. And there was only one someone. Me.
If you run an independent label, you already know this feeling. The romantic version of the job is discovering music and shaping how it reaches people. The actual version, most days, is coordination. It's dozens of small, low-drama tasks strung across weeks and across several releases at once, where the whole game is simply not dropping any of them. Miss a distributor deadline and the release slips by weeks. Forget to follow up and an artist feels forgotten. The thing you actually care about quietly suffers because of the thing you never signed up to care about.
I started building software to fix that. But to explain why, I have to back up a little.
The question that never left me
I've been pulled toward emerging technology for as long as I can remember. Not the hype around it, which I've always found a bit exhausting, but the machinery underneath, and the questions that machinery raises. I ended up writing both my bachelor and master thesis on distributed ledger technologies, and I fell for both halves of the subject at once. I loved the mechanics: how a chain of hashes makes tampering visible, how strangers who trust no one can still agree on a shared truth, how a few elegant cryptographic primitives combine into something genuinely new. And I loved the questions sitting right on top of that machinery. Who does this technology serve? What does it quietly assume about how people should trust each other? What does it encode about the way we want to live together?
I've never been able to separate those two loves, and I've stopped trying. The technical detail is where the real design decisions live, and the human consequences are why those decisions matter. You have to hold both to see clearly.
That framing stuck with me, and I've come to believe it applies to all technology, not just the buzzy kind. A tool is never neutral. It carries the values of whoever built it, whether they meant to put them there or not. Software that optimizes for engagement encodes a belief about what your attention is worth. Software that optimizes for the number of paying seats encodes a belief about who the customer really is. These choices are cultural before they are technical.
So when I sat down to solve my own late-night coordination problem, I wasn't only trying to save time. I was, without quite realizing it at first, encoding what I believe a label should be.
Building the boring, important thing
The first version was not impressive to look at, but I loved building it. It started as an Airtable base that treated every release as a structured record, with the artists, tracks, deadlines, and deliverables all linked together instead of scattered across my memory and a dozen chat threads. Then I wired in automation, so that when a release crossed a certain date, a scenario would fire and send exactly the right message through SendGrid without me having to remember to do it. It was held together with tape and duct-tape logic, and honestly that was half the fun. There is a specific pleasure in watching a system you built quietly do a thing you used to do badly by hand.
The deeper it went, the more I enjoyed it. Turning a messy human process into clean data models, deciding what should be a field and what should be a rule, figuring out which moments deserved an automated nudge and which needed a human touch. This is real engineering, even when it looks humble, and I find it genuinely satisfying. But something more interesting happened as it grew.
Every time I added a feature, I had to make a small decision about what mattered. When a release is running late, should the system nudge the artist, or nudge me? When an old track suddenly starts getting attention, whose job is it to notice? How should the label talk to an artist when it reaches out, and in whose voice?
Each of those is a values question wearing a technical costume. And because I was building the tool for a label I actually run, for artists I actually care about, the answers came out differently than they would have if I were building generic software for a market I'd never met. The tool started to reflect a particular idea of care. Not because I wrote that idea into a mission statement, but because it lived in a thousand tiny defaults.
That's the moment this stopped being a side project in my head and started being something I genuinely wanted to think out loud about. Because if a modest tool for my own label already encodes values this quietly, then the much bigger wave that's arriving now, software that doesn't just remind you but acts on your behalf, is going to encode them far more powerfully. And almost nobody in music is talking about that yet.
What this series is about
This is the first post in a series, so let me be honest about where it's going and where it isn't.
I'm going to write about building Labelflow, the software I make for running my label. Not as a product pitch, but as a way into a bigger set of questions I find genuinely fascinating. How does artificial intelligence actually change the work of a label, once you strip away the noise? What happens when the software stops waiting for instructions and starts taking initiative? And what does all of this tell us about where music technology is heading, and what we should want it to protect?
I'll try to keep two promises. The first is to explain complex ideas as simply as I can, because I think most technology writing hides fairly small ideas inside unnecessarily large words. The second is to stay honest about what I don't know yet. I'm figuring a lot of this out in real time, while running a real label, which I think is the only honest vantage point to write from anyway.
I don't believe technology is going to save music, and I don't believe it's going to ruin it. I believe it's going to encode whatever we put into it, more and more directly, and that this makes the human choices behind it more important, not less. That's the thread I want to pull on, one post at a time.
If that sounds interesting to you, I'm glad you're here. Let's get into it.