AQ vs Factory: one vendor's agent or every agent, one place

Published September 13, 2026 · by the AQ team

Choose AQ if you want your team running the coding agents it already subscribes to (Claude Code, Codex, Cursor Agent, and others) together, as real CLIs in persistent sessions on a VM you control, watched and steered from any browser. Choose Factory if you want to standardize on one vendor's agent: Factory's Droid is a single, polished harness that runs across terminal, desktop, web, Slack, and IDE, on models Factory serves and meters. Both keep agents working after the laptop closes, both offer machines you own as the execution surface, and both let a teammate open a running session. The fork in the road is the agent itself: Factory asks your team to adopt Droid; AQ is a harness of harnesses that runs whichever agents your engineers already use, side by side.

Factory (factory.ai) has been building agents since 2023 and announced its "Factory 2.0" software-factory positioning in June 2026. It ships quickly, so every Factory claim below is as of September 2026, verified against factory.ai and docs.factory.ai.

AQ vs Factory at a glance

What is Factory?

Factory builds Droids: autonomous software development agents intended to cover the whole development lifecycle, not just code generation. The Droid CLI is the core surface (Factory reports it at the top of the Terminal-Bench leaderboard), and the same sessions sync across a desktop app, web and mobile, IDEs, and Slack. Droid is model-agnostic in the sense that you pick the underlying model per task (Claude, GPT-5, Gemini, and others, served through Factory's managed access; API-key BYOK across roughly ten providers, including local models via Ollama, is opt-in), but the harness itself is Factory's own.

Two 2026 additions matter for this comparison. Droid Computers, opened to users in April 2026, are persistent machines for running Droids remotely: Factory provisions and manages one in its cloud (managed computers hibernate when idle and wake for delegations), or you register a Linux, macOS, or Windows machine you already manage (a VPS, workstation, or on-prem server) through Factory's relay with no inbound ports. State persists between sessions: filesystem, credentials, running services. Missions, available on Enterprise and Max plans, are multi-agent autonomous runs that decompose a large task into parallel tracks over hours or days. Around the edges sit droid exec for headless CI use with tiered autonomy, a Python SDK, and integrations with Linear, Jira, Slack, Notion, PagerDuty, and MCP tools. Sessions are shareable: a teammate can open a live session URL, watch, comment inline on the diff, or take over.

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.

On AQ the agents are the real CLIs your engineers already use (Claude Code, Codex, Cursor Agent, Kimi, Grok, or plain shells), running in persistent tmux sessions on the team's VM and streamed live to the browser. Each engineer signs in with their own Claude or OpenAI account, and AQ never marks up usage on your own subscriptions. Every workspace gets one isolated git worktree (branch ai/{id}-{slug}, automatic dependency install, one-click rebase onto main), so parallel sessions never collide. Sessions survive a closed laptop and resume from any device, including a phone. Teammates open the same workspace and watch the same live session; typing into someone else's terminal is delegated by its owner approving a control request in one click, and visibility is owner-managed: team-visible, or private and shared with specific people. Label a Linear issue ai-task and a workspace appears: status syncs both ways, ownership follows the assignee, and humans start agents (never auto-launched). There is no shared multi-tenant execution tier.

Where Factory stands out

• One agent, deeply built. Droid is a serious harness with years of iteration behind it, and standardizing on it buys consistency: one set of behaviors, one skills format, one place to tune autonomy.
• Autonomy at scale. Missions decompose day-long work into parallel tracks and verify along the way; nothing in AQ auto-decomposes a project across agents for you.
• Delegation from anywhere. A Linear or Jira issue, a Slack mention, or a tap on the mobile app becomes a running Droid, and a hibernated managed computer wakes to take it.
• Headless and SDK surfaces. droid exec with tiered autonomy and a Python SDK make Droid scriptable inside CI and internal tools.
• A free-fallback model pool. When metered standard usage runs out, the open-weight Droid Core pool keeps work moving at no extra cost, on its own rate limits.

Where AQ wins

• The agents your team already chose. Factory's answer to "which agent?" is Droid. AQ's answer is: whichever CLIs your engineers already subscribe to and trust, running unmodified, with their existing configuration. When the next great CLI ships, it joins the same workspace instead of replacing your platform. That is the harness of harnesses bet.
• Subscription economics, not metered usage. Factory's individual plans meter usage under rolling 5-hour, 7-day, and 30-day rate limits, with prepaid credits beyond that. On AQ, agents bill through each engineer's own Claude or OpenAI plan; AQ charges a flat per-user price and never marks up usage on your own subscriptions. Teams already paying for Claude Max or ChatGPT keep that investment. See running coding agents on your own accounts.
• The live terminal is the shared surface. Factory shares its own session view; AQ shares the actual running CLI. What a teammate opens is the same tmux session the agent runs in, so what you watch, steer, and hand over is the ground truth, not a rendering of it. Control passes only when the owner approves it, in one click.
• Previews that pull in non-engineers. Every AQ workspace runs a live dev-server preview with viewing links that need no account; comments pinned on the preview (screenshots supported for teammates) go to the agent as prompts, and outside reviewers comment via a review link after a quick email sign-in.
• Humans stay the starters. AQ's Linear intake creates a workspace and syncs status both ways, but a person opens it and starts the agent. Teams that want issue-to-agent automation will prefer Factory's delegation; teams that want a human deciding when an agent touches the repo will prefer AQ's default.

The deeper difference: adopt an agent, or a place for every agent

Factory and AQ agree on more than most pairs on this page: work should outlive the laptop, machines you own are a legitimate execution surface, parallel agents need worktree isolation, and a session someone else can open beats a black box. Where they part is what your team commits to. Buying Factory is adopting Droid: its behaviors, its skills format, its model menu, its meter. That is a fine trade if Droid is the agent you want, and Factory's benchmark results argue it is a good one.

AQ commits you to a place instead of an agent. Engineers keep their own CLIs, subscriptions, and muscle memory; the team gets one browser surface where every session (whatever is driving it) is visible, steerable, resumable, and reviewable. That neutrality matters because agent leadership keeps changing hands, and because delegation pipelines concentrate work on one person: a July 2026 LeadDev analysis of 25,264 agent-generated pull requests across 2,361 popular GitHub repositories found that in 79 percent of agentic PRs the same developer both reviewed and modified the agent's contribution, and only about one in eight workflows involved multiple humans. A live session the whole team can open, whichever CLI is running inside it, is the structural answer: the second engineer arrives before the diff is final, not after. See monitoring multiple agent sessions for how that works day to day.

Pricing compared

Factory, as of September 2026: Pro at $20 per month, Plus at $100 (roughly five times Pro usage, plus Droid Computers), and Max at $200 (roughly ten times Pro usage and early feature access), all governed by rolling rate limits, with the free Droid Core open-weight pool after standard usage runs out and prepaid Extra Usage credits from a $10 minimum. Enterprise pricing is a contract.

AQ has two plans. Free is a personal sandbox for one person: AQ creates a private machine in an isolated network, nothing to install, no time limit. Team is $50 per user per month early access (standard $200, billed monthly), covering VMs you connect from your own cloud or a dedicated always-on AQ-managed VM in its own isolated network, with your rate locked for your first 12 months. Inviting a teammate costs nothing until the person accepts and joins. Your own subscriptions never appear on an AQ bill: agents run on each engineer's own Claude and OpenAI accounts, and AQ never marks up usage on your own subscriptions.

Bottom line

Factory is one of the strongest single-vendor agent platforms of 2026: a benchmark-leading harness, persistent Droid Computers including your own machines, Missions for long autonomous runs, and delegation from every tool your team already uses. If you are ready to make Droid your team's agent and pay for usage through the platform, it deserves a serious evaluation. AQ answers a different question: your engineers already picked their agents, so give the team one always-on place to run them together, on your machines and your subscriptions, in the cloud, live. If the agent roster should stay yours to change, that is AQ's side of the line.