Everything AQ does

AQ is one place for the AI work a team does all day: chat with agents about your code and data, build with any coding agent in an isolated worktree, and give agents standing jobs as bots, on a machine that is yours and with the subscriptions you already pay for. Here is all of it, with links to the details.

Three ways to work

ChatAsk about your code, data and tools. Answers come back as tables, charts and pages. BuildA worktree per task, any agent, a live preview, review and merge. BotsA standing job that wakes on a schedule or an event, with a monthly cap.

Together

CrewsA Lead plans, builders build, and a separate agent reviews the work. Review and mergeDiff scopes, line comments that reach the agent, checks and a merge that explains itself. Live multiplayerTeammates open the same live terminal and preview, and take the keyboard when you hand it over.

Everywhere

GenieYour personal assistant: a morning briefing, mail, meetings and the work waiting on you. PhoneEvery thread, preview and terminal on any phone. The iPhone app is in beta. IntegrationsLinear, GitHub, Slack, Gmail, Calendar, Granola, Sentry and Mercury, per-tool permissions. Models and AutoYour own subscriptions unmetered, AQ models priced up front, Auto per turn.

Live multiplayer

Every work item is a shared place. Teammates who open it see the agent working in the same terminal, not a replay, and can request the keyboard; the owner grants it in one click and takes it back any time. Previews of the running app sit beside the conversation, and on the Pro plan you can share them with a link, pin comments on the page and send those comments to the agent as its next message. See Live Links and previews and comments.

A teammate requests control of a live terminal and the owner grants it for five minutes — Hand over the keyboard for a few minutes, and take it back any time.

Memory across every agent

AQ reads the memory each coding agent keeps on your machines (Claude Code, Codex, Grok, Muse Code, Command Code and Antigravity) and makes it searchable from any agent, so what one learns the others can use. Edit a note once, share project notes with your team when you choose, and keep one set of rules in a shared AGENTS.md that AQ mirrors into every agent's home. See memory.

AQ's memory page with notes from Claude Code and Codex, team notes and a Needs a look tab — Notes from every agent in one place, personal by default.

Secrets, kept out of transcripts

Save API keys and connection strings as personal or organization secrets. Agents, terminals and previews get them as environment variables, an agent can ask you for a missing one with a secure card, and every saved value is replaced with its name in transcripts, commit messages and pull request titles. See secrets.

Visual replies and pages

An agent can answer with a page above its text: a chart, a table, a diagram or a mockup. Publish it at a stable link your team can revisit, keep it fresh with a bot, and share it with people outside the team through a guest link on the Pro plan.

An agent reply rendered as a chart in the thread — A chart, rendered by the agent, in the thread.

Linear in, pull request out

Label a Linear issue ai-task and a work item appears in the team sidebar with the issue moved to In Progress. Agents commit and open pull requests with each person's own GitHub account, and finishing the work item can move the issue to Done. See the Linear integration.

API, webhooks and MCP

Drive AQ from your own tools on the Pro plan: a versioned REST API, signed webhooks when agents open or merge pull requests, and an MCP server for Claude, ChatGPT and Cursor. See the API and MCP.

Machines

Every workspace runs on a private machine in its own isolated network: a sandbox AQ creates for you on the Free plan, or on the Pro plan an always-on machine AQ runs for you or VMs in your own cloud. On Pro, make your machine bigger or add machines from Settings, and pick a boot disk up to 2 TB. Prices are per month, billed with your plan; an added machine starts at $61.86 a month.

Size	Shape	Good for	Your machine	Each added machine
Standard	2 vCPU · 8 GB	One person, everyday work	Included	$61.86 / month
Performance	4 vCPU · 16 GB	Good for 2 to 4 people	+$51.36 / month	$113.22 / month
Power	8 vCPU · 32 GB	Big builds, many agents at once	+$154.08 / month	$215.94 / month
Max	16 vCPU · 64 GB	Many people, many agents	+$359.52 / month	$421.38 / month
Ultra	32 vCPU · 128 GB	Large builds, a whole team's agents	+$770.48 / month	$832.34 / month
Mega	48 vCPU · 192 GB	Big monorepos, parallel suites	+$1,735.20 / month	$1,797.06 / month
Mega Memory	48 vCPU · 384 GB	Memory-hungry builds and data work	+$2,358.75 / month	$2,420.61 / month

More on runners and infrastructure and security and isolation. Plans are on the pricing page.

Frequently asked questions

What is AQ?

AQ is where engineering teams chat, build and run bots with every AI coding agent: Claude Code, Codex, Cursor and the rest, plus AQ's built-in agent, in shared live workspaces on machines that are theirs.

Which coding agents does AQ support?

Claude Code, Codex, Cursor, Kimi Code, Grok, Antigravity, Devin, Muse Code, OpenCode, Command Code and Pi, each signed in with your own account, plus AQ's built-in agent on AQ models.

Is AQ free?

Yes, the Free plan is a personal sandbox for one person. The Pro plan adds your team, sharing, your own cloud, always-on and bigger machines, and the API. See pricing.