Purpose-built agents you schedule and wire together into a score that keeps running while you don't. Not one general assistant — a roster of narrow ones for outreach, inbox triage, lead research and rank monitoring, each handing work to the next.
One assistant that does everything does most of it badly. Maestro goes the other way: a roster of narrow agents — cold outreach, inbox triage, lead research, ranking monitoring — that run on a schedule and hand work to each other.
The metaphor did the explaining
An orchestra: a roster, a score, and you conducting. That framing carried the whole site, which meant the architecture could be explained without leaning on a system diagram.
What it took to land
The roster section shows concrete agent cards — the job, the tools it is wired into, live status, and real numbers. 86 sent today. 1,204 enriched last week. 14 drafts waiting. One screen, and you understand both what an agent is here and that these are running right now.
The handoff is the actual product. Any one of these agents is a script somebody could write in an afternoon; the thing that makes it a platform is that the outreach agent’s replies land in the triage agent’s queue without anyone moving them. So the score — how work passes between agents — got its own explanation rather than a bullet in a feature list.
What the project included
Agent roster with live-state cards, the score and handoff explanation, the integration surface, docs, blog and changelog, and the beta application flow.