Glossary

Conductor

A conductor is a coordination layer that sits above a group of AI agents and watches what all of them are doing, not just what any single one is doing. Instead of running tasks itself, it looks across every agent's output for the kind of problem only visible from above: two agents working at cross purposes, two agents quietly doing the same job, or a job that nothing is covering at all. It writes what it finds as findings, for a person to read and act on, rather than acting on its own.

The more AI agents an organization runs, the more this kind of problem grows, not because any one agent is broken, but because none of them can see the others. A conductor is built to look at that whole picture, the way a manager reviewing everyone's work catches an overlap or a dropped ball that no single person on the team would notice from inside their own task.

In Glitch Team Zone, every org gets its own conductor, scoped to that org's agents and pills alone. It reads across pill runs and their outcomes and writes what it notices, conflicts between two pills, redundant work, a gap where nothing is covering a job that needs one, to the conductor's desk for an owner or admin to review. It doesn't act on its own. A person decides what happens next.

Checking business rules is part of the same daily rhythm: the conductor a customer sees is the same one running those checks and filing what trips a rule. Whether it's watching agents for conflicts or checking a rule against yesterday's activity, everything it finds lands as a finding for a human to review, never a change it makes on its own.