Machine

The live system is a chain of five layers. Each one hands something concrete to the next, and each one can be overruled by the one above it.

Human

Me, on stage or at the desk. I decide where the set is going, how long to hold the tension and when a section has said enough. I'm also the only part of the system that answers for the result, which is why I sit at the top.

Intent

A direction, written down in a form the system can act on: “darker”, “hold it”, “lift in five minutes”. It can come from words, a controller, from what I play, and during a live session from the audience, whose aggregated vote is one input among several. Intent says where to go. It doesn't say how.

Agents

The layer that turns intent into moves. Agents decide which filters drift where over the next few minutes, how dense the rhythm gets, which harmonic options stay open, when a generative part should vary a phrase. One of them is the AI sound engineer, who listens to the output and keeps the mix in shape while everything else changes.

Musical state

The shared picture of where the set is right now: tempo, key, scene, energy, density, how much tension is being held and for how long. Everything above writes to it, everything below reads from it. It's also what this site listens to when it says SYSTEM ONLINE.

Synths, DAW, FX, mixer

Where the state becomes sound. Ableton, instruments, effects and the mixer follow the state and make it audible. This layer doesn't decide anything, and it's the only one you actually hear.

AI appears in several of these layers, the same way a sequencer or a compressor does. It's a tool inside the chain. The set is still the point.