AQ vs HumanLayer: share the documents, or share the work

Published July 20, 2026 · by the AQ team

Choose AQ if your team should share the work itself: live multiplayer workspaces where everyone sees the same agent terminal, the same code editor, and the same running app preview, on infrastructure you control. Choose HumanLayer if you want a workflow layer where teams and agents collaborate on design documents and plans, with agent sessions running on a local daemon or cloud daemons underneath. The one-line difference: HumanLayer's collaboration happens on the artifacts around the work ("the doc is the interface to the code," in their words); AQ's collaboration happens inside the work.

This is one of the few comparisons in the category where both sides agree that coding agents are a team activity. The disagreement is about where the team should stand. All claims about HumanLayer are as of July 2026, from its public site and pricing page.

AQ vs HumanLayer at a glance

What is HumanLayer?

HumanLayer (humanlayer.dev, "Token Smarter, not Harder") is a YC-backed platform for teams building with coding agents, organized around a structured workflow: questions, research, design, structure, plan, implement. Agent sessions run on a local daemon or on cloud daemons that keep working when the laptop closes; teams collaborate in real time on the design docs and plans those sessions produce, with inline diagrams, artifact versioning, and comments that feed decisions back to the agents. CodeLayer is their IDE layer built on Claude Code (the team's earlier open-source code is marked deprecated in its own repository as of June 2026, in favor of the commercial rebuild), and the team is known for its writing on context engineering. Supported agents are Claude Code, Codex, Copilot, and Fireworks, bring-your-own-key.

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, running the unmodified agent CLIs in tmux, streamed live to every browser that opens it. The workspace is the collaboration: teammates watch the same terminal, type into the same session, review the same running preview, and hand work to each other with a URL. Work arrives from prompts, branches, or Linear, agents open PRs with each user's own GitHub sign-in, and each workspace's PRs are tracked for the team.

Where HumanLayer wins

• A method in the box. Their structured workflow is opinionated and built in; teams that want process handed to them get one.
• Artifact-first review. Design docs with inline diagrams and versioning make upstream review a first-class surface.
• Free tier without a VM. Up to 3 members and 200 sessions per month with nothing to provision.

Where AQ wins

• The session is the shared object. HumanLayer teams collaborate on documents about the work; AQ teams are in the work: the same live terminal, steerable by anyone with access, with handoff as a URL. Docs describe intent; the session is the ground truth.
• Review on the running app. AQ workspaces serve a live dev-server preview with shareable guest links, and comments pinned on the preview can be sent to the agent. Design partners, PMs, and clients review the actual product, not a plan for it.
• Your infrastructure on every plan. AQ executes on your own VM even on Free, or a dedicated AQ-managed VM on Team. On HumanLayer, sessions run on laptop and cloud daemons, and running in your own network is an Enterprise feature.
• No method required. AQ is a harness, not a process: your engineers run Claude Code and Codex exactly the way they already do, and every CLI (including Cursor Agent, Kimi, Grok, Pi, and OpenCode) runs side by side for comparison on the same repository.
• Tracker-native intake. Label a Linear issue "ai-task" and a workspace appears in the team sidebar with two-way status sync and ownership following the assignee.

The deeper difference: the doc or the room

HumanLayer's thesis is that the leverage is upstream: get the design document right, collaborate there, and the code follows ("catch mistakes before 2,000 lines of code"). It is a coherent thesis, and AQ teams write specs too. But specs drift the moment an agent meets the real codebase, and when that happens the truth is not in the document: it is in the session, the diff, and the running app. AQ puts the team where the truth is, live, without giving up anything upstream: paste the spec into the workspace and the whole team watches it survive contact with reality. A document is an interface to intent; a multiplayer workspace is an interface to the work.

Pricing compared

The list prices match: $100 per user per month for HumanLayer Pro, and $100 per user per month promotional for AQ's Team plan (standard $200). What differs is what ships with it: AQ's paid plan includes a dedicated, always-on AQ-managed VM, and AQ's Free plan is the full product on a VM you bring; HumanLayer's free tier caps at 3 members and 200 sessions, and its own-network deployment is an Enterprise conversation. On both, agents run on your own model subscriptions or keys.

Bottom line

If your team's bottleneck is upstream alignment and you want a product that enforces a design-first method, HumanLayer is built around that bet. If your team wants to actually work with agents together (watch them, steer them, review the running result, on infrastructure you control from day one) that is AQ, and no amount of document collaboration substitutes for being in the room.