When Every Demo Is Generated: The A&R Ear in the Age of Infinite Music
Every A&R team in electronic music is living through the same strange paradox. The tools have never been more powerful, the barrier to entry has never been lower, and yet the job of finding a record that actually matters has never been harder. The reason is simple: when anyone can generate a thousand grooves in an afternoon, making music stops being the bottleneck. Deciding what deserves to exist becomes the entire discipline.
We have watched this shift arrive in stages. First came sample packs and loop libraries that collapsed the distance between idea and arrangement. Then came AI-assisted tools that could finish a pattern, suggest a chord, or sketch a mix in seconds. Now the conversation has moved to generation itself — systems trained on the genre’s patterns, capable of producing plausible techno, dark progressive, and melodic techno on demand. As a label, we do not see this as a threat to the music. We see it as a stress test for curation.
The demo inbox has changed forever
The volume of submissions we receive no longer tracks human hours. A single producer, armed with the right stack, can prototype more music in a weekend than a studio artist could finish in a season. That is not a complaint — it is a description of the new terrain. The result is that the signal-to-noise ratio has collapsed, and the skills that once made an A&R person valuable are being forced to evolve.
Three things now separate a record we sign from a record we pass on:
- Intention. A generated track can be perfectly competent and say absolutely nothing. We listen for a point of view — a decision the artist made that the machine would not have.
- Texture. The ear still hears the difference between a plugin that recreates a room and a room that was actually there. Imperfection is becoming the tell.
- Context. A track does not live alone. We ask what it does on a floor, what it sits next to in a set, and what it will mean in six months — not just today.
Where AI actually earns its keep
None of this makes us Luddites. Quite the opposite. Inside our own release pipeline, AI has become a quiet collaborator with a narrow remit. It helps us prototype references quickly, test arrangement ideas against a tempo map, and surface patterns in the flood of demos that deserve a second listen. It is an accelerator of attention, not a substitute for it.
The distinction we draw is deliberately strict: generation is cheap, curation is not. A machine can propose ten variations on a bassline. Only a human decision — informed by years of floors, labels, and listening — can tell you which one belongs in a catalog that will still be relevant in five years.
The demo inbox used to measure talent by how much you could produce. Now it measures taste by how much you can refuse.
The label’s real moat
This is the uncomfortable truth of the current moment. When the cost of production approaches zero, the only scarce resources left are taste, trust, and time. A label is not a distributor of files. A label is a promise: that if this record carries our name, someone listened, someone argued, and someone chose it for a reason.
That promise is the thing AI cannot generate. It is also, we suspect, the thing listeners are beginning to crave again. After a stretch where abundance felt like freedom, a new appetite is emerging for filtering — for someone to say “not everything deserves your attention, but this does.” That is, and always was, the job of an A&R ear.
We are not worried about the machines. We are worried about the labels that forget why they exist — the ones that will let the algorithm do the listening. Those catalogs will feel thin, interchangeable, and eventually silent. The ones that survive will treat AI as a tool and curation as a craft.
We intend to be in the second group.