- What it does
- Listens continuously to a live mix, separates speech from music, adjusts as the room changes, and coaches whoever is operating in plain language on a phone or a projector.
- The control
- Every action it can take is safety-classified. Destructive changes ask first, and a kill switch, a rollback and a manual override sit above all of it.
- Failure
- Output quality that depends on whichever volunteer is at the desk that day, going to FM, Zoom and YouTube at once.
- Unit of work
- Competence that travels with the system, rather than with the individual.
- Authority
- Named systems
- Soundcraft Ui24R · FM · Zoom · YouTube
- Claims
- Deployed across four venues in three countries, with the same safety classification, operator consent and override hierarchy at every desk.
- Non-claims
- A measured time saving or audio-quality uplift. The deployment footprint is confirmed; a comparable cross-venue performance series has not been supplied.
- Evidence held
- Four venues in three countries.
- Status
- In production
The operator always outranks the AI
A foundation running multilingual services for congregations across global venues faces a quality problem that does not scale: each hall depends on whichever volunteer is at the desk that day, and the output goes to FM, Zoom and YouTube at once.
The deployment footprint is evidence of operation, not evidence of a particular quality uplift. Those are different claims and the page keeps them separate.
The same ten fields, with the arguments behind them.
Commercial work makes room for selected public-benefit work.
This engagement was delivered pro bono for a charity foundation. We reserve some capacity for projects introduced by clients, boards and partners where the mission, need and technical fit are unusually strong.
That is a case-by-case commitment, not an automatic benefit attached to a commercial engagement. Selection, scope and availability are agreed separately.
Classified before it was capable.
The safety classification came before the capability, not after it. That ordering is the whole design: a system that can already act, being retrofitted with a list of things it may not do, is a different and much weaker object than one that could never do them.
Classify every action before building any.
Safe, destructive, or forbidden, decided for each action the desk exposes and fixed before a line of control code existed.
Speech against music, continuously.
The room changes through a service, and a mix set at the start is wrong by the middle.
Plain language on a phone.
The volunteer at the desk is told what is happening and what to do, rather than watched by a system that acts around them.
Coaching was built alongside the listening, because a system that adjusts silently teaches the room nothing
Roll out across four venues in three countries.
The same safety classes, operator consent, kill switch and rollback travel with every installation.
Operate under the local desk.
The system listens and coaches continuously, but the person in the venue remains above every action it can take.
Before and after.
| Measure | Before | After |
|---|---|---|
| Operating footprint | One controlled development environment | Four venues in three countries |
| Broadcast comparison | Recorded feeds used during rehearsal | Each installation compares the desk output with the live broadcast signal |
| Who holds authority | Whichever volunteer is at the desk that day | The local operator, with coaching, rollback and a kill switch above the system |
The table shows deployment and control changes only. Deliberately excluded: any time-saving or audio-quality figure that has not been measured consistently across the four venues.
What this case does not claim.
No hours saved or quality improvement. Four venues in three countries establishes a live operating footprint, not the size of an outcome.
No claim that it replaces the operator. It coaches one, and the whole authority design exists to keep the person at the desk in charge of the room.