Three answers, and your agent works without you
Sign in, pick a size, pay with a card, answer three questions. The agent starts in your code. Close the tab and it keeps working. From $39 a month.
From $39 a month · Docker and sudo included · Cancel from the dashboard any time
What happens, in order
- Sign in
Google or email
Sign up with Google or with an email and password, then add a passkey in Settings and sign in with a touch from then on. Your account holds your Compute, your projects and your connected subscriptions.
- Pick
Pick a size
Pick a size and a name for your Compute; both come filled in. Every size gets the same agents, Docker and features. Sizes differ in CPU, memory, disk and terminals at once.
- Pay
One card, one monthly charge
Check out in Stripe. One subscription covers every project and terminal; a bigger size later costs only the difference for the rest of the month.
- Bring
Your code and your subscription or key
Create a project on Projects: clone a git repository, start empty or import a folder from home. Choose Claude Code, Codex or OpenCode and a subscription or API key it takes, then open its terminal and the agent starts.
- Run
The agent starts in your repo
A tab opens with the agent in /work/<project>, your CLAUDE.md and AGENTS.md in place, approvals off. Add more projects, launch more agents and run them side by side, as many at once as your size has terminals.
- Leave
Close the tab or the laptop. The agent keeps working.
Come back from any browser or your phone and see how far the agent got. The tab was only a window onto a tmux session in your terminal.
Where everything runs
Your devices only draw windows. The work happens on your Compute, and the keys that mint new tokens never leave our server.
What each part does for you
Your Compute keeps the work
Projects, each with its own home folder, live on one persistent disk. A project terminal mounts its project and that home in a fresh Linux system; a Shell sees every project. All of them share your size's CPU and memory.
The tab is only a window
Each terminal is a tmux session in its own container. The dashboard and your phone open the same session.
The server keeps the keys
Refresh tokens for your subscriptions stay encrypted on our server. Your Compute receives an access token for each CLI, renewed on schedule.
The board keeps count
Usage of every connected Codex and Claude account, the next reset, and an email when a used-up window refills.
Questions about this
How long until it is ready?
As soon as Stripe confirms the payment. You land back on the Compute page, which prepares your account disk and turns to Running; open Terminals, choose the project, agent and account, and the agent starts in your code.
Do I have to connect my subscriptions?
No. Connecting them fills the limits board and lets every agent you launch start signed in. You can also sign in inside a project's terminal the way you would anywhere else; that login stays in that project.
What if I pick the wrong agent or size?
Launch another agent or edit a saved launch at any time. Change to a bigger size on the fly, no restart; move down from the next month once your data fits the smaller disk.
Where does it run?
On servers around the world, and we connect more regions on request. The agent works next to its code, so distance only touches your keystrokes. Anyone, anywhere can sign up.
Your agent is waiting in the first tab
Pick a size now and hand the agent the long job.
Get my ComputeYour first overnight run works, or your money back · Cancel from the dashboard; it runs to the end of the paid month