Claude Code, in a Mac app built for many sessions

AI Bridge runs the real Claude Code CLI, on your own subscription, in a real terminal, and gives every session a place: the job it does, the state it is in, and a conversation you can come back to tomorrow.

AI Bridge on a Mac: Claude Code sessions grouped into workspaces, one rolling out a release through ArgoCD
The real app: Claude Code sessions in their workspaces, each with its live state in the sidebar.

Your Claude Code, unchanged

Each agent in AI Bridge is the claude command you already run, in a real terminal: the same models, the same commands and slash commands, the same subscription or API key. Nothing sits between you and Anthropic.

AI Bridge never edits your ~/.claude/settings.json. To know whether a session is working or waiting on you, it adds its settings to that one launch only, so the configuration you rely on in your own terminal stays yours.

See which session needs you

Every session shows its live state: working, waiting on you, done or asleep. When Claude Code stops at a permission prompt, the session says so in the sidebar and in the Inbox, the one list of everything that is waiting for you, so you answer it now rather than find it an hour later. That is the heart of running several agents at once.

Pick it up tomorrow where it stopped

Put a session to sleep and wake it later with its conversation resumed. Quit the app and every terminal comes back with what was on it, the full depth of it, not just the last screen. A long-running piece of work stays a session you can return to, not a tab you are afraid to close.

Organised by the job, not the repo

Sessions live in workspaces: an area of responsibility, such as the product, infrastructure or support, with its own instructions, context files, repos and tools. Start a Claude Code session in a workspace and it already knows the rules for that work. A project above them carries the rules for the whole business.

Bring in the sessions you already have

Claude Code sessions you started in a terminal before AI Bridge are not lost. Bring them in, and they take their place beside the rest, with their history.

Beside Codex, Grok and OpenCode

Claude Code is one of the four CLIs AI Bridge runs. Put Codex beside it for a second opinion, or give each CLI the job it does best, and watch them all on one Grid.

Where your Claude Code sessions can reach

Questions

Does AI Bridge replace Claude Code?

No. It runs it. Every agent is the claude CLI you already use, in a real terminal; AI Bridge is where you start sessions, see their state, answer them and pick them up again.

Is AI Bridge a GUI for Claude Code?

In a way: it is a Mac app around the real claude CLI, not a chat window that imitates it. You still type to Claude Code in its own terminal; the app adds what a terminal cannot show, such as which session is waiting on you, every session grouped by the job it does, and a wall of them on one screen.

How do I resume a Claude Code session?

In a terminal, claude --continue picks up the most recent conversation in that folder, and claude --resume lets you choose one. In AI Bridge you wake a sleeping agent and its conversation comes back, with what was on its terminal.

Do I need another Anthropic subscription or API key?

No. Claude Code uses your own Claude plan or API key, exactly as it does in your terminal. AI Bridge’s own plans are separate, and its Free plan has no limit on agents.

Does it change my Claude Code settings?

No. AI Bridge never writes to your ~/.claude/settings.json. What it needs to know a session’s state is passed to each launch, so your own configuration stays exactly as you left it.

Does my code go anywhere?

No. Everything runs on your Mac, and Claude Code talks to Anthropic exactly as it would without AI Bridge. The app collects no usage data.

What do I need?

A Mac with macOS 14 or later, Apple silicon or Intel, and Claude Code. If Claude Code is not installed yet, Settings › Setup can install it for you.

More of what AI Bridge does

Almost ready.

AI Bridge is in its last round of testing. Leave your email and we’ll send you the build the day it ships.