aq.
PricingStartupsDocsCompareGuidesLaunchesSign in
  • Claude Code
  • Codex
  • Kimi Code
  • Grok
  • Cursor
  • Antigravity
  • Command Code
  • Devin
  • Muse Code
  • OpenCode
  • Pi

Run every coding agent
in the cloud, with your team.

Claude Code, Codex, Cursor and every other coding agent, in a browser tab, on a machine that never sleeps. Teammates join the same live session with a link.

Start freeSee how it works

Free sandbox. No card. Your subscriptions stay unmetered.

An AQ workspace in the browser: an AQ thread running two agents in parallel with a queued follow-up, Claude and Codex tabs beside it, the team sidebar with live presence, and the live preview of a launch site with a T-minus countdown
5,300+workspaces opened
700+pull requests merged
14,000+agent sessions run

Counted across every team on AQ, as of September 2026.

What AQ is

One workspace per task. Every coding agent you already pay for. Your whole team in the same live terminal, on infrastructure that is yours.

Live multiplayer

Watch the same terminal, live.

Teammates open your workspace and see the agent type, not a replay. Hand over the keyboard with one click and take it back any time.

A teammate requests terminal control and the session owner grants it for five minutes
Parallel workspaces

Give every task its own worktree.

Each workspace is an isolated git worktree, so parallel agents never collide. Close the laptop; the session keeps running on the machine.

The team sidebar: five active work items with live presence, Linear identifiers, and pull-request badges
Live previews

Review the running app, not the diff.

The dev server streams right next to the agent. On the Pro plan, share it with a link, pin a comment on the page itself, and send that comment to the agent as its next prompt.

A live preview of a launch site inside the workspace, annotate mode active, with a teammate's comment pinned to the countdown
Linear in, PR out

Label an issue. Get a pull request.

Tag a Linear issue ai-task and a workspace appears in the team sidebar, with status syncing both ways. Agents commit and open pull requests with each person's own GitHub account.

Work item rows created from Linear issues, showing issue identifiers and pull-request badges
Any device

Pick it up from your phone.

The same session on macOS, Windows, Linux and mobile. Sessions are exactly where you left them, from any device.

The same AQ thread open on a phone, with the mobile workbench navigation
Where your code runs

Your machine, your accounts, your boundary.

  • Code stays on a machine dedicated to you. Your checkout, worktrees, shell sessions and agent credentials live on a VM in your own cloud or one AQ runs for you. There is no shared execution infrastructure.
  • Agents sign in with your accounts. Claude Code, Codex and the rest use the subscriptions your team already has, and those stay unmetered.
  • Model traffic goes where you point it. Prompts go to the provider you chose. Only AQ's built-in agent, on AQ's models, routes through AQ's gateway.
Your teamBrowsersPhonesAny OSaq.devSign-in, presence,streamingYour machineYour cloud, or a VM dedicated to youCheckout and worktreesTerminals and dev serversAgent credentialsClaude Code, Codex, Cursor...Not shared with any other teamModel providers, via your own accounts
The browser sees the session. The machine holds the code. Providers see only what the agent sends.
How it works

Up and running in three steps.

  1. 01

    Connect an environment

    Start free in a sandbox AQ creates for you, or run in your own cloud with an always-on machine for your team.

  2. 02

    Sign in to your agents

    Claude Code, Codex, Cursor and every other coding agent, with the accounts you already have. Or skip this: AQ's built-in agent works with nothing connected.

  3. 03

    Build together, live

    Every task gets its own workspace: a shared live terminal, a code editor, and an app preview your whole team opens with a link.

Questions

Before you start.

What is included in the Free plan?
A private, AQ-managed sandbox for one person with nothing to install, and no seat charge. Use your own agent subscriptions, or use AQ's built-in agent at published per-token prices. The Pro plan adds teammates, sharing, and your choice of machines in your own cloud or a dedicated VM managed by AQ. Details are on the pricing page.
Do I need my own Claude or OpenAI subscription?
No. Connect the subscriptions your team already has and they stay unmetered, or use AQ's built-in agent on AQ's models. Every model shows its own price before you pick it. See pricing.
Where does my code run?
On the Free plan, on a private AQ-managed VM dedicated to you. On the Pro plan, on machines you connect from your own cloud account or on a dedicated AQ-managed VM. Your checkout, worktrees, shell sessions and agent credentials stay on that machine; AQ does not use shared multi-tenant execution infrastructure. Model inference is separate: prompts go directly to the model provider, or through AQ's gateway when you use AQ's built-in agent.
What happens when I click Start free?
You accept the Terms of Service and continue with a work Google account. If you are not already approved, AQ adds you to the review queue. Requests are reviewed personally, usually within a day or two, and AQ emails you when your workspace is ready. Then sign in, connect an agent, and add your first project.

Bring your whole team into the session.

Start free in a sandbox AQ creates for you. Move to your own cloud when the team is ready.

Start freeSee pricingTalk to sales