HumanLayer alternatives now that HumanLayer is joining Factory

Published November 4, 2026 · by the AQ team

HumanLayer is joining Factory, per Factory's announcement dated November 4, 2026, and no shutdown date has been announced. The current HumanLayer platform stays up during a transition period with no end date given, and Factory says it will support users in exporting their data, migrating to Factory, or self-hosting the HumanLayer open source. Multiplayer tasks and workstreams, and in-app collaboration on agent-written plans, product specs and design docs, are promised "in the next few weeks"; as of November 4, 2026 they have not shipped. Your realistic options are to wait for Factory, self-host HumanLayer, or move to another tool, and nothing forces that choice this week.

If your team needs agents that keep running on a cloud machine with teammates able to open the same session: AQ runs real agent CLIs like Claude Code and Codex in persistent sessions on a VM in your own cloud account or a dedicated machine AQ manages, each task in its own git worktree, and any teammate with access can open the same live terminal, editor and app preview.

What changes for HumanLayer and CodeLayer users?

As of November 2026, per Factory's announcement:

• The platform keeps running. "The current HumanLayer platform will remain up and running during a transition period." Factory has not said when that period ends.
• Three supported paths. Export your data, migrate to Factory, or self-host the HumanLayer open source. The post gives no step-by-step instructions for any of them yet.
• Open source. Factory writes that it "will be open sourcing HumanLayer" and does not link a repository, license or date. Separately, humanlayer/humanlayer on GitHub has held the desktop app, daemon, CLI, web UI, API and plugins under the Apache 2.0 license since October 22, 2026; its README says "HumanLayer Cloud runs this code. Guides for running it yourself are in progress."
• Promised features. Multiplayer tasks and workstreams and in-app collaboration on plans, specs and design docs are planned for the coming weeks. Treat them as plans until they ship.
• Pricing. The post says nothing about pricing during the transition. As of November 4, 2026, humanlayer.dev still lists a free Starter plan (up to 3 members and 200 sessions a month), Pro at $100 per user per month, and custom Enterprise pricing, and shows no notice about Factory.

A note on names: as of November 2026, humanlayer.dev describes the product as HumanLayer, "a multiplayer coding agent IDE and cloud", and no longer uses the CodeLayer name. If your team knew it as CodeLayer, the transition above applies to the same product.

What should HumanLayer teams do now?

1. Keep working. The platform stays up during the transition, and no end date has been set. There is no reason to move in a hurry.
2. Export your data when the path is published. Factory says it will support exports. Watch HumanLayer's channels for the export steps and take a copy as soon as they exist, whichever path you pick.
3. Keep your workflow documents in git. The research notes, plans, specs and design docs your agents wrote are the part of a HumanLayer workflow that is hardest to recreate. Make sure each one you still rely on lives in your repository or another place you control.
4. Push every branch. Work an agent committed on a laptop daemon or a cloud daemon and never pushed exists only on that machine.
5. Write down what you depend on. List the sessions that must run while laptops are closed, the people who review plans, and the agent CLIs your team uses. That list decides which path fits.
6. Pick a path with dates in hand. Choose between Factory, self-hosting the open source, or another tool once the transition end date and the self-hosting guides are published, or earlier if your needs already point one way.

How the options compare

Vendor details are as of November 4, 2026, from each vendor's own pages linked in the sections below.

Factory: the official path

Factory (factory.com) describes itself as "an agent-native software development platform for planning, building, reviewing, testing, and shipping production software." Its agents, called Droids, work from the command line, desktop, web, mobile, Slack, Microsoft Teams, Jira, Linear and CI, per its site as of November 2026. Its pricing page lists Pro at $20 a month with cloud and local background agents, Plus at $100 a month with Factory-managed cloud computers for remote Droids, Max at $200 a month, Teams at $60 a month per team plus $40 a month for each seat, up to 10 seats, and custom Business and Enterprise plans, with on-premise deployment listed on Enterprise. Factory says the HumanLayer team will bring multiplayer tasks and plan and spec collaboration into the product in the coming weeks. Fit: teams that want continuity with the HumanLayer team and are comfortable running Factory's own agent. Our head-to-head is at AQ vs Factory.

Self-hosted HumanLayer

Factory lists self-hosting the HumanLayer open source as a supported path. As of November 4, 2026, humanlayer/humanlayer on GitHub is a public Apache 2.0 monorepo; its README says HumanLayer "runs coding agents (Claude Code, Codex, OpenCode, Pi) on your machines and lets your team steer them from one place: tasks, sessions, plans, and reviews, synced between a desktop app, a web app, and a daemon on each host." The same README says self-hosting guides are in progress and warns that the Docker Compose starter uses simple default passwords that must be changed before you deploy. Fit: teams that want to keep HumanLayer's workflow exactly and have someone to run a web app, an API, a Postgres sync layer and daemons. Wait for the published guides before you plan a production move.

AQ

AQ is a browser-based workspace for running agent CLIs as a team. Each task gets its own isolated git worktree and branch, and the real CLIs (Claude Code, Codex, Cursor, Kimi Code, Grok, Antigravity, Devin, OpenCode and others) run in persistent sessions on a machine dedicated to your team: a VM you connect from your own cloud account, so code, worktrees and CLI logins stay in your infrastructure, or an always-on VM AQ manages in its own isolated network. Closing your laptop does not stop the agent. A teammate with access opens the same workspace and sees the same live terminal, editor and running preview; to type into someone else's terminal they request control and the owner approves. You review the diff, comment on lines and merge from the Changes panel.

For multi-step work, a crew's lead agent plans the work, you approve the plan, and it delegates build, test, review and research to other agents, with review routed to a different vendor than the author where one is connected. Bots are named agents your team talks to, with routines that run on a schedule. AQ Free is a personal sandbox machine for one person; AQ Pro is $50 per user per month at the early-access rate (standard $200), per the pricing page as of November 2026.

What AQ does for the move: Import from HumanLayer brings each HumanLayer workspace into AQ as a work item on its branch, with its Claude Code sessions continuing and its plan and research files placed in the worktree, from one command pasted on your machine. What AQ does not do: it has no built-in research, plan, implement document workflow; keep your plans and specs in the repository, where any agent in AQ can read them. The full comparison is at AQ vs HumanLayer.

Other options

Warp Oz

Warp Oz (warp.dev/oz) is a cloud agent platform that runs Claude Code, Codex or Warp's own agent, gives a team "a single pane of glass to view all agents", and lets people join sessions with one click, per its site as of November 2026. It can be Warp-hosted or self-hosted, with self-hosted cloud agents listed on Enterprise. Warp's pricing lists Build at $20 a month and Business at $50 per user per month for up to 25 seats. More in our Warp Oz alternatives guide.

Cursor cloud agents

Cursor's cloud agents run in isolated VMs in Cursor's cloud, per Cursor's docs as of November 2026. You can send a run's URL to a teammate on the same Cursor team, who can view it read-only, or send follow-ups if a team admin turns that on. Self-hosted cloud agents are listed on Enterprise, and Cursor's pricing lists Teams from $30 per user per month. Fit: teams already standardized on Cursor and its agent.

OpenHands

OpenHands (openhands.dev) offers Agent Canvas, which is open source under MIT, free to self-host, and runs on your laptop or a shared VM, plus a self-hosted Teams plan at $500 a month for up to 10 users with identity and role controls, per its pricing page as of November 2026. It says you can use any agent, including Codex and Claude. Fit: teams whose deciding requirement is running everything on their own infrastructure. More in our OpenHands alternatives guide.

Orca

Orca (onorca.dev) is a free, open source desktop app for macOS, Windows and Linux that runs Claude Code, Codex, OpenCode, Grok and many other agent CLIs in parallel, each in its own worktree, and can run agents on a remote box over SSH, per its site as of November 2026. It is built for one person; it lists no shared sessions. Fit: an engineer who used HumanLayer's local daemon mostly alone. More in our Orca alternatives guide.

How to choose

Start from the list you wrote in step 5. If your team's value is the research, plan, implement workflow and the shared plan documents, the shortest paths are Factory, where the HumanLayer team is building that collaboration, or self-hosting the open source once the guides land. If what matters most is agents that keep running on a cloud machine while teammates watch and steer the same session, compare AQ and Warp Oz. If code and credentials must stay on infrastructure you control, look at AQ on VMs in your own cloud account, self-hosted HumanLayer, OpenHands, or the self-hosted Enterprise tiers of Warp and Cursor. If you work alone, Orca covers parallel local agents.