AQ vs Superconductor: two takes on multiplayer coding agents

Published July 19, 2026 · by the AQ team

Choose AQ for multiplayer coding agents done right: teammates and agents in one shared live workspace (real terminals your whole team can watch and type into), running the CLIs you already subscribe to, on infrastructure you control. Choose Superconductor if you want a hosted take on the same idea, with native mobile apps and a flat monthly price, and you are comfortable with your code running in the vendor's cloud sandboxes. The one-line difference: both are multiplayer platforms for coding agents, but AQ runs in your own cloud while Superconductor is a hosted product.

This is the rare comparison where the two products agree on the thesis: coding agents are a team activity, not a single-player tool. They disagree about where the work should live. All claims about Superconductor below are as of July 2026, from its public site, docs, and pricing pages.

AQ vs Superconductor at a glance

What is Superconductor?

Superconductor (superconductor.com) calls itself "the multiplayer workspace for your team and AI coding agents." It is built by Volition, Inc., a small team led by the founders of Gradescope. You run many Claude Code, Codex, and other agents in the company's cloud sandboxes, teammates join shared agent sessions to observe or take control, and the product layers guided code review, live previews with automated QA checks, and GitHub and Slack integrations on top. It ships native iOS and iPad apps for driving agents from a phone, and its automations can triage email and draft tickets from meetings. A distinctive feature: it can build a personal benchmark from your own repository's past pull requests and score different agents against it.

Its docs are explicit about the philosophy: agents belong in the cloud, at scale, and the company argues that local, single-machine agent tools are a dead end. Pricing follows that bet. Rather than charging per seat, Superconductor charges a flat monthly fee per workspace tier (free for small use, $128 per month for up to 32 members as of July 2026) and meters the compute the agents consume as "sandbox hours."

What is AQ?

AQ is the multiplayer coding harness where engineering teams run AI coding agents like Claude Code and Codex together: shared live terminals, a code editor, and app previews, in your own cloud.

Every AQ workspace is an isolated git worktree on a VM you control: bring your own, or provision an AQ-managed dedicated VM in a few clicks. The agents inside it are the real CLIs you already subscribe to, running in tmux and streamed live to the browser, so a teammate who opens your workspace sees the same terminal scrollback, the same editor, and the same running preview you do, and can type into the session or pin a comment on the preview and send it to the agent as a prompt. Agents authenticate with each user's own Claude, OpenAI, and GitHub sign-ins, and AQ does not mark up that usage; OpenCode and Pi can alternatively run on AQ's built-in inference gateway. Work arrives by hand, from a prompt, or from Linear: label an issue "ai-task" and a workspace appears in the team sidebar with status syncing both ways.

Where Superconductor wins

• Native mobile apps. Real iOS and iPad apps for launching agents, swiping through previews, and getting push notifications. AQ works in mobile browsers but has no native app as of July 2026.
• Flat pricing for larger teams. $128 per month covers up to 32 members (plus metered compute). For a big team where per-seat math stings, that is genuinely cheaper than AQ's per-user plan.
• Personal benchmarking. Scoring agents against your own repository's past pull requests; AQ does not have an equivalent today.
• Bundled automations. Email triage and meeting-to-ticket automations ship in the product; AQ's intake is Linear, GitHub, and manual prompts.
• Zero infrastructure thinking. Everything runs in Superconductor's sandboxes with per-second billing and idle pausing. You never think about a VM at all.

Where AQ wins

• Your code runs on your infrastructure. AQ executes agents on your own VM or a dedicated AQ-managed VM, on every plan including Free. On Superconductor, execution in your own infrastructure is an Enterprise feature; everyone else runs in the vendor's cloud. If a security review asks "where does the code live while agents work on it," AQ's answer is "wherever you decide."
• The real CLIs, in a real terminal. AQ workspaces are actual tmux sessions running the unmodified CLIs. Anything the CLI can do in a terminal, you and your teammates can do together: watch, scroll back, interrupt, take over, run shell commands next to the agent.
• Sharing that crosses team lines. Workspaces can be shared with specific people, the whole team, or outside guests through revocable links, and preview comments can be sent back to the agent as prompts. Design and product can review an agent's work from a link, with no seat on the team and no local checkout.
• Linear-native intake. A label on a Linear issue creates the workspace, ownership follows the assignee, and status syncs back automatically.
• No new runtime to trust. AQ is a harness around tools and infrastructure you already trust: your CLIs, your subscriptions, your GitHub auth, your VMs. Leaving AQ means keeping all of them.

The deeper difference: whose cloud is it?

Superconductor and AQ agree that the interesting problems start when a whole team works with agents: visibility, handoff, review, and shared context. Superconductor's answer is a hosted platform: polished, integrated, and running in its own cloud, with your code inside its sandboxes. AQ's answer is a harness over your own infrastructure: the same multiplayer workflows, but execution stays on machines you control, running CLIs you already pay for. Which answer is right depends on how your team weighs convenience against control. Teams that have been burned by vendor lock-in, have compliance requirements about where source code executes, or simply want the agents they already use rather than a new platform's abstraction tend to land on AQ's side of that line.

Pricing compared

Superconductor's Pro plan is $128 per month flat for up to 32 members as of July 2026, plus metered sandbox hours (billed per second, with an overage rate past the included allotment) and a free tier capped at 4 members and modest compute. AQ is $100 per user per month at promotional launch pricing (standard $200) with a managed, always-on VM included, and a Free plan if you bring your own VM. The honest math: for a large team using modest compute, Superconductor's flat fee is likely cheaper; for a small team that wants always-on dedicated compute it owns, or any team that wants the Free-plan-on-your-own-VM route, AQ's model is simpler and keeps the infrastructure yours. On both products, model usage comes from your existing Claude and OpenAI subscriptions.

Bottom line

These are the two products in this category that treat coding agents as a team sport rather than a solo tool, and the difference between them is conviction about where your code belongs. Superconductor packages the idea inside its own sandboxes. AQ gives teams the multiplayer workflow on infrastructure they control (their own VM, or a dedicated AQ-managed one), running the unmodified CLIs their engineers already trust: real terminals, real tmux, real worktrees, nothing to migrate off of later. That is the version to bet an engineering organization on.