Works with
+33 more

Stop babysitting agents. Start merging real work.

Run a fleet of coding agents while keeping branches, reviews, and CI failures manageable.

Used by engineers from

Features

orchestrator

describe the outcome, the orchestrator runs the fleet

Every project gets a main agent. Tell it what you want built and it plans the work, starts a worker for each piece in its own worktree, and comes back only when something needs you.

board

the whole fleet, at a glanceevery session sorted by what it needs from you

Sessions place themselves from live facts: what the agent is doing, its pull request, CI, and reviews. You open what needs you and leave the rest alone.

  • placed by facts

    Lanes come from the agent, its PR, CI, and reviews, not a status someone has to remember to update.

  • one card per session

    Task, agent, branch, and pull request together. Open it for the chat, terminal, changed files, and preview.

  • needs you, first

    Questions, failed checks, and requested changes land in one lane, so nothing waits on you unseen.

feedback loop

CI fails, the right agent fixes itreview comments go back to the session that wrote the code

Orchestrator watches every pull request it opens. Failed checks, review comments, and merge conflicts go to the worker that owns the branch, with what it needs to fix them, until the PR is approved.

  • failed checks

    The failing check, its name, and a link to the log go straight to the agent that owns the branch.

  • review comments

    Comments from people and from reviewer agents return to the same session, thread by thread.

  • agent reviews

    Have a reviewer agent read every PR before a person does, and send its findings back in one click.

agents

use the agent you already trust38 coding agents, in chat or in their own terminal

Pick the agent and model for each task. Work with it in structured chat, or open its native terminal UI. The task, branch, and feedback stay attached either way.

  • structured chat

    Plans, commands, and file edits in one readable timeline, with approvals right where you need them.

  • native terminal UI

    Open the agent's own TUI in the same worktree, or move a session between chat and terminal.

  • a browser it can drive

    Every worker gets its own browser to preview its app, and the agent can click through it to check its work.

mobile

your fleet goes where you goanswer an agent from your phone, while agents run on your machine

Pair the free iPhone, iPad and Android app with your desktop by QR code. Get a notification when an agent needs you, reply in chat, or open its terminal.

Free on the App Store for iPhone and iPad and on Google Play for Android. How pairing and remote access work

  • notified when it matters

    An alert when an agent asks a question, a pull request is ready, or a check fails.

  • reply from anywhere

    Answer in chat, approve a plan, or start a new task without opening your laptop.

  • runs on your desktop

    Agents run on your machine. The phone reaches it on the same network, through the managed HTTPS tunnel, or over Tailscale on Android (iPhone and iPad can't use Tailscale).

See it in action

Pricing

Free and Open Source for local. Pricing applies to Cloud only.

Free

For running agents on your own machine. Open source, no account needed.

$0/month

What's included

  • Claude Code, Codex, Cursor and more
  • Orchestrator, workers and PR reviews
  • Your own API keys, no proxying
  • Open source under Apache 2.0

Starter

Recommended

For one developer whose agents keep going after the laptop closes.

$20/month

2 at once8 sessions

What's included

  • Agents keep working while you sleep
  • Instant VMs or Coder dev workspaces
  • Pull requests, CI fixes and reviews in the cloud
  • Up to two hosts on the same network
  • 1 orchestrator, 20 session hours a week

Pro

For a bigger fleet, spread across all your machines.

$40/month

10 at once25 sessions

What's included

  • Everything in Starter
  • Unlimited hosts over a relay connection
  • 5 orchestrators at once
  • 80 session hours a week, 24 every 6 h
  • 1 usage reset a month

Enterprise

For teams that need scale, control and support.

Custom

25+ at once100+ sessions

What's included

  • Everything in Pro
  • 10+ orchestrators, no session-hour caps
  • Bring your own Coder deployment
  • SSO and team management
  • Priority support with an SLA

In the wild

Real feedback from builders using Orchestrator.

Frequently
asked questions

I already use an IDE like Cursor, is this for me?

Orchestrator is designed to work with your existing tool. We natively support deep-linking to IDEs like Cursor so you can open your workspaces and files in your IDE. Orchestrator sits above individual tools; use whatever agent you like, and it keeps the workflow the same.

Which AI coding agents are supported?

Orchestrator supports 38 coding agents, including Claude Code, OpenAI Codex, Cursor, OpenCode, Aider, Goose, GitHub Copilot, and Gemini CLI, with per-project agent choice. Each project can pick its own worker, orchestrator, and reviewer agents.

How does the parallel agent system work?

An orchestrator agent plans the work and spawns worker agents, each in its own isolated Git worktree, so they can work on different branches or features at the same time without conflicts. Orchestrator watches every pull request they open: CI failures, review comments, and merge conflicts are routed back to the agent that owns the branch. A built-in AO reviewer agent can review the PR and send requested changes to the same worker, and you follow the whole fleet from one board.

Is Orchestrator.inc the same as Agent Orchestrator?

Yes. Orchestrator.inc is the new name for Agent Orchestrator (AO). It is the same open-source app and the same ao CLI, and the source code lives at https://github.com/OrchestratorInc/agent-orchestrator.

Is Orchestrator free to use?

Yes. The local app is free and open source under the Apache-2.0 license, with unlimited local agents on any supported harness, the orchestrator, workers, and PR reviews. Agents work in Git worktrees on your computer, Orchestrator doesn't proxy your API calls, and no account or credit card is needed to run agents locally. Your phone or another computer can connect on the same network; on Starter and up, the managed HTTPS tunnel connects from anywhere. Tailscale doesn't work on iPhone or iPad and is not yet verified on Android, and there is no built-in device relay yet. Paid Cloud plans start at $20/month and add cloud projects and cloud agents that run in hosted sandboxes.

Can agents run in the cloud?

Yes. Orchestrator Cloud is available now: sign in to Orchestrator Cloud from the desktop app, choose Clone to the cloud to pick a GitHub repository, and run worker and orchestrator sessions in hosted sandboxes. From the desktop you get Chat, the remote terminal, the Files inspector, and the session's pull request and reviews. Cloud agents come with the Starter, Pro, and Enterprise plans. Local projects keep working without Cloud.

Can I use my own API keys?

Yes. For local projects, Orchestrator runs the coding-agent CLIs already installed on your machine with their existing sign-ins, whether that is an API key or a provider subscription. It doesn't proxy API calls, so usage is billed by each provider and you keep full control over costs.

Can I manage my agents from my phone?

Yes. The free Orchestrator app for iPhone, iPad and Android pairs with the desktop app by QR code. From your phone you can see every session on a board, get a push notification when an agent needs you, reply and approve, start new work, follow PRs and CI, and open a live terminal, across every computer you've paired. Agents run on your computer. On the same Wi-Fi it connects directly. Away from home, Starter and up can install the managed HTTPS tunnel from Connect Mobile, which works from any network, or pair over Tailscale on Android (not yet verified on a device; iPhone and iPad can't use it). There is no built-in device relay yet.

Does it work across multiple devices and machines?

Yes. Run the desktop app on each computer, then pair them all to one phone: the mobile app lets you name each desktop, choose a default and filter work across machines. The Free plan covers 2 devices total, in any mix of phones, computers and remote VMs. A signed-in desktop can also add Orchestrator on another computer as a remote host under Settings → Remote hosts, a preview you turn on with Settings → General → Developer mode.

Which platforms are supported?

The desktop app runs on macOS (Apple silicon and Intel), Windows and Linux (AppImage, .deb and .rpm). The mobile app runs on iPhone and iPad (iOS 16.4 or later) and Android.

Is Orchestrator open source? What license?

Yes. Orchestrator is open source under the Apache-2.0 license at github.com/OrchestratorInc/agent-orchestrator.

Still confused?

Read the documentation for a detailed usage guide.

Read the docs