For thirteen years I built software for others.
I run an agency —Agencia Digital Manager— and my job was always that: understand a client's problem, translate it into code and ship it working. I did it well and I enjoyed it. But there was an idea that kept coming back, always the same one: at some point I wanted to build something of my own. Not a commission. Something that really moved me.
The problem is that I never found that something. No idea obsessed me enough to drop everything and chase it. Until music got in the way.
The hardest thing I tried to learn
Three years ago I decided to study seriously. First DJing, then music production, at the Casa Sonido academy. I thought it would be the usual struggle —I'm an engineer, I'm used to learning complex systems— and I was completely wrong.
Producing music was the hardest thing I've ever had to learn. And not for a technical reason. Music feels almost infinite: every decision opens ten more decisions, there's no correct answer, and the craft that separates an idea from something that actually sounds good is enormous. I could spend a whole afternoon chasing a feeling I had crystal clear in my head and never get there.
I came from a world where almost every problem has a solution you can search for, read and apply. Music production doesn't work like that. There it's just you, the open session and a ton of intuition you don't have yet.
The moment the two things crossed
Through my other job, I was already using AI copilots to code. And one day, while writing code with one of those assistants by my side, the obvious thing happened: I thought "this should exist inside Ableton".
Not a generator that spits out a finished song. That didn't interest me —and still doesn't. What I imagined was a copilot. Something that understands your session, that walks you through decisions, that explains why something isn't sounding the way you want, that proposes and lets you say yes or no. The same kind of assistance I already had while coding, but for the craft that was giving me the hardest time.
I gave it a ninety percent chance of being just another one of my delusions.
Three weeks locked away
The only way I know to find out whether an idea is any good is to build it. So I locked myself away for three weeks, got obsessed, and put out a minimal viable version. Enough to look at it and decide whether it was worth landing the full idea or filing it away with the rest.
It worked. Not perfectly, but well enough to give me the green light I was looking for. That MVP is the origin of what Deloy is today.
What Deloy is, in one sentence
Deloy is an AI copilot that lives inside Ableton Live and accompanies you while you produce. It doesn't replace the producer: it assists them. It understands your music, it proposes, and you keep the authorship and the final word.
That last part isn't a marketing detail. It's the reason I built this instead of joining the line of AI music generators. The producer remains the author. Always.
Why I'm writing this
These "Notes from the studio" are going to be exactly that: what I'm learning at the crossroads between engineering and music, the product decisions I make and why, and what I think is coming for AI in music.
I write it from the inside out. Not from the podium of someone who already knows everything, but from the desk of someone who is building and producing at the same time, and gets it wrong at both, often.
If you produce music, if you care about where AI is heading in this craft, or if you simply want to see how a tool like this gets built from scratch —stick around. This is just beginning.
— Cristian
