Maestro runs teams of specialised AI crews that plan, build, test and ship software — on your own machines, never ours. And Maestro Hub is your mission control: a web dashboard where you watch every project, review every plan, and approve every step. No terminal required.
Three ideas explain the whole system — no jargon needed.
A crew is an AI specialist with one job: a Planner writes the plan, a Builder writes the code, a Tester verifies it, a Security reviewer inspects it, a Releaser ships it. Up to five crews work in parallel — each hands off to the next only when quality checks pass.
A voyage is a sprint: a batch of related work that travels together from idea to shipped code. Every voyage moves through clear phases — from backlog, through active work, to merged — so you always know exactly where things stand.
Maestro Hub is the web dashboard over it all. Watch every project live, see which voyages need you, read and approve plans in the browser, and track spend — without touching a terminal.
One screen across every project. The Hub shows what the crews are doing right now, flags what needs a human decision, and puts that decision one click away.
Voyages are grouped into five plain-English phases — Backlog, Underway, Needs Attention, Final Review and Merged. The two phases that need a human are always at the top, in the brightest colours. One tap on the Needs you filter hides everything the fleet is handling on its own.
Before the crews build anything significant, a plan lands in your review queue. Read it in the Hub, then approve it, send it back for revision with your notes, or reject it outright. The fleet waits for your call — human review gates are built into the workflow, not bolted on.
Split checkout into 4 orders: payment form rebuild, address validation, order summary, e2e verification. Estimated 2 crews, 1 voyage.
Every crew slot shows what it's doing right now — which task, which specialist role, how long it's been at it. A live activity feed streams every meaningful event across the fleet: work dispatched, steps completed, anything that needs eyes. If a crew stalls, you see it; no work disappears silently.
Every unit of AI usage is tracked per project, per voyage and per task, with estimated cost and day-by-day history. You pay Anthropic directly for the AI — Maestro just shows you exactly where it went, so budgeting is a report, not a guess.
The Hub is where leaders live. The CLI is where engineers live. Same fleet underneath.
Govern the fleet from the Hub: watch delivery across every project, approve plans, keep the audit trail, and see the spend — with nothing leaving your infrastructure.
Everything is driven by a pip-installed CLI: scaffold a workspace, queue orders, dispatch up to five parallel Claude Code squads, and monitor them from your terminal.
Plane Sales Global rebuilt its aircraft marketplace — three portals, Stripe billing end to end, advertising inventory, data migration — with one human operator running Maestro fleets, for roughly $2,300 of tracked model spend.
Maestro runs entirely on your machines. There is no Maestro cloud, and your code never crosses the network to us.
Everything — crews, dashboard, records — runs where your code already lives. Nothing is hosted by us.
Crews use your own Anthropic key and your billing relationship. We never see your code or your tokens.
Every action, decision and approval is recorded locally — with air-gapped operation supported for regulated environments.
Talk to us about running Maestro in your organisation — or start solo with the free Community edition today.