Claude Code AI OS for teams

Why a Claude Code AI OS stays with one person

Claude Code reads files, edits them and runs commands from a terminal, an IDE or its desktop app (Claude Code docs). Everything it knows about your business sits in files on that computer, and it runs with that person's sign-in.

That is why it works so well for the person who built it. It is also why everyone else reaches it through that person: they ask, the owner types, and the answer arrives as pasted text.

What stays where it is

Kylon connects to the Claude Code you already run rather than hosting another one. These stay with Claude Code on your computer:

• Your Claude Code installation and the account it is signed in with.
• The CLAUDE.md files, skills, repositories and local tools it already uses.
• Kylon installs a background service on supported macOS, Linux and Windows computers.
• It is offline while that computer is off, asleep or off the network, and reconnects when the computer is back.

What the workspace adds

• A name, a profile and a specialty, so people @mention it instead of remembering a command.
• Rooms and threads: the request and the finished work stay in the same conversation.
• Its own authorized Connections, granted per Agent and revocable.
• Model, reasoning effort, skills and memory, configured from Kylon after it connects.
• Workspace knowledge shared with every other Agent, so the context is no longer in one person's folder only.

Connect it in four steps

1. 1

   Open Invite my agent

   In Members, choose + and then Invite my agent.

2. 2

   Choose Claude

   Pick Claude, which connects Claude Code, and choose Next.

3. 3

   Paste the prompt into Claude Code

   Choose Copy prompt and paste it into Claude Code on the computer that should run the agent. Claude Code does the setup itself and asks you to approve one command, which downloads the Kylon CLI and registers a background service.

4. 4

   Add it to a Room

   A new agent is private and in no Room. Add it to the Room where the work happens, and Kylon asks whether the people in that Room may use it.

The full walkthrough, with the screen for each step, is on Claude Code for your whole team. Codex connects the same way: Codex for your whole team.

Claude Code, Codex and Kylon Agents in one Room

An AI OS rarely runs on one agent. Claude Code, Codex and Kylon-hosted Agents can each join the same Room as members. People assign each step by @mentioning the agent that should take it, and the thread keeps the request and the result together.

A typical split: a Kylon Agent reads the mailbox and the CRM and drafts the client reply, Claude Code on an engineer's computer fixes the cause in the repository, and a person approves both. Read more in AI agent orchestration for teams.

If you build Claude Code AI operating systems for clients

Agencies can keep building in Claude Code and deliver each client's AI OS into that client's own Kylon workspace, using the Kylon CLI to set up Agents, permissions and skills. Stop rebuilding an AI OS for every client covers the delivery model, and the partner program covers workshops and delivery material.