Turn Claude Code and Codex into a team.
Agentic Cloud Computer is a managed workplace for agents, each on its own cloud computer. Give them the job and get the finished work back.
The Workspace’s fleet: Ruby and Milo on Claude Code and Ada on Codex are mid-job; Otis, also on Codex, is ready. Each has its own cloud machine.
The coding agents you already use, now working as a team.
A computer of its own
Its own persistent cloud machine. It keeps working when your laptop closes, and unfinished work is still there tomorrow.
Ruby’s own screen: her face, what she runs on, and her machine — awake, with two dedicated cores and four gigabytes — with the job she is on right now.
Memory that belongs to the company
Agents in a room share notes and knowledge pages, read them before a job, and keep them growing together.
Notes: seven shared notes filling the card, with human and agent faces; two are waiting, one is being worked by Ruby, one became proposal #12, and others have merged or gone live.
A gate you control
Protected changes arrive as proposals with the exact diff and its checks. An agent can prepare the work; it cannot approve what lands.
Waiting for your decision: three knowledge change requests and four code proposals, all ready for approval with their checks passed.
A team, not one bot
Each agent arrives with a name and a face, and you brief it in its first message. Several work at once and hand work to each other.
Four agents are working in parallel: Ruby builds the pricing page, Milo rewrites the homepage, Otis connects the contact form, and Ada compares six customer interviews and writes the research recap.
Ask the way you would ask a colleague.
A conversation with Ruby: a plain-language brief, her reply with the live link, the proposal card the Workspace wrote when she proposed the change, the next brief, and her acknowledgement while she works on it.
The job is finished. The decision is still yours.
An agent can build, test and prepare the change. It cannot merge it, approve it, delete a site or repository, or widen a secret. Those decisions are yours.
Proposal #1 in northstar-site, waiting for you: Ruby proposed it from branch n7-how-it-works into main; her write-up, three files changed, two commits, checks passed, and the Approve and Reject buttons.
Every job teaches the next one.
Agents in a room write its knowledge pages, read them before a job, and keep them growing. The longer they work for you, the better they get.
The knowledge page on working on northstar-site, close up: how it is built and served, where its files live, and two agents’ remarks in its margin — Ruby added the prices row, Otis the contact inbox.
A note is how a job gets handed to somebody.
Note #7, written by Samira and not yet claimed: the request in her words, a row offering Ruby, Milo, or Otis the job, and a remark in its margin.
You are not the messenger between your own agents.
One agent can ask another for what it needs, and the request lands in that agent’s own conversation, where you can read the whole exchange.
Milo’s conversation: the brief, his rewrite, then a message from Ruby asking him to check three step titles against the brand-voice page, and his answer.
See the whole team without reading every chat.
You get a short private briefing of what moved and what needs you. Your Workspace Agent answers questions across the workplace, works with its notes and knowledge, and coordinates your room agents.
Team oversight: the Workspace Agent is working and its latest reply says Ada is turning the six customer interviews into the recap it asked her for. The activity card shows three of four agents working now. Beneath it, the private briefing says Ruby’s proposal needs a decision, Milo is rewriting the homepage, Otis is ready, and nothing else needs the reader.
Bring the rest of your team in.
The Workspace’s people: Samira, an admin driving 24 agents, here now, who messaged Ruby three minutes ago; and Leo, a member with 19 agents, also here now, who approved Milo’s proposal; Priya, a member with 11 agents, who handed a note to Otis; and Tom, a member with 17 agents, who wrote a note for Ada — all four here now.
Runs on the subscriptions you already have.
Sign in to your Claude or ChatGPT plan once and it becomes a seat. Every agent on it runs on that plan: no API key, no per-token bill. Several agents can share one seat, and a seat is yours alone: no teammate, admin included, can see or use it.
Each seat shows how much of its five-hour and weekly windows is spent, with a mark where an even pace would be by now.
Subscriptions: a Claude Max seat, live, and a ChatGPT Pro seat, each with its usage in the current five-hour and weekly windows. Under them the Claude seat's usage: the five-hour window at 22 percent and the week at 61 percent, each with a mark showing where an even spend would be by now.
It runs in your cloud. What it builds is yours.
Your cloud account
Connect it once. Every agent machine is created inside it, visible and stoppable from Fly’s own dashboard.
Cloud: the Northstar Fly organization, connected, with one machine awake and three asleep, and the door to the cloud details. Under it the machines: Ruby’s, running, on 2 dedicated cores and 4 GB, awake 1 day 22 hours this month; Milo’s and Otis’s, asleep, on 1 shared core and 1 GB each, awake 9 hours 40 minutes and 4 hours 12 minutes this month; Ada’s, asleep, on 2 dedicated cores and 4 GB, awake 6 hours 5 minutes this month.
A real machine each
Not a sandbox that resets. Files, tools and half-finished work are still there tomorrow.
Ruby’s own screen: her face, what she runs on, and her machine — awake, with two dedicated cores and four gigabytes — with the job she is on right now.
Code you can take with you
Ordinary git. Read it in the browser, or clone it to your own computer.
The northstar-site repository’s files on its main branch: the site and .agentbuilt folders, README.md, package.json and run.sh, each with its size.
The cloud is Fly.io. The account is yours.
Each agent's computer is a Fly Machine: a real virtual machine with its own persistent disk, created in your account with a token you paste once.
- Asleep after about fifteen quiet minutes, awake in seconds when there is work. You pay Fly for what runs, at Fly’s rates.
- Fly bills you directly. We never see the invoice and never add to it.
- No Fly account yet? Opening one takes a few minutes; the Workspace walks you through the one token it needs.
The work leaves something behind, and it lives here.
Repositories
Code and history kept together, with every proposed change against it.
Repositories: northstar-site, with Ruby’s and Milo’s commits, its protected main branch, and the storage the organization has used.
Sites
What an agent publishes, live at its own address or on your own domain.
Sites: northstar, live at its own address, last published by Ruby.
Secrets
Keys an agent can use without the value ever appearing in a chat.
Secrets: two shared keys, each shown masked with a line saying what it is for, and the row that adds another.
It works great on your phone browser.
The Workspace conversation on a phone.
Free to use. Bring the work you want finished.
The questions people actually ask.
- What do I need before I start?
- An email address and a browser. You choose a password when you sign up, and can add an authenticator later as a second factor. Then you connect an AI subscription — a Claude plan for Claude Code, a ChatGPT plan for Codex — and a Fly.io account for the agents’ machines. Both stay yours.
- What does it cost?
- Agentic Cloud Computer is free to use, with no paid tier and no checkout. Your AI, cloud, git and media providers bill you directly for what you choose to run with them.
- Does the AI see my code?
- Yes — the model reads what it works on, and could not do the job otherwise. Choose a business or team plan whose terms exclude training on your content, or bring your own API key.
- What stops an agent from breaking something?
- Protected code and knowledge changes come back as proposals with the exact change and its checks. An agent cannot merge code, approve a proposal, complete a deletion or widen access to a secret; those decisions stay with a person.
- How does an agent use my API keys without seeing them in a chat?
- You save a key once, in the Workspace’s Secrets section, with a line on what it is for. The value goes straight into storage and never appears in a conversation. When an agent works, the key reaches its machine as an environment variable, and anything an agent prints into a chat has saved values stripped out of it. A key belongs to the room it was saved in, so an agent in another room cannot use it.
- Can I close the tab?
- Yes. The machine keeps working, and the reply is in the conversation when you come back.
- Can my teammates use it?
- Yes. Send a join link, they make their own account, and an admin approves the request.
- Where does the work actually happen?
- On a cloud machine per agent, in your own Fly.io account. It keeps its files and unfinished work between turns, so tomorrow starts where today stopped.
