Start a Task on Your Laptop, Finish It From Your Phone — Claude Code Remote Control

You give your AI agent a big task at the office. Then you leave. With Claude Code Remote Control, you pull out your phone at the coffee shop, scan a QR code, and pick up exactly where you left off. Your files never leave your machine. The session never stops.

The Problem: Your AI Is Chained to Your Desk

Here is a scene that plays out constantly if you use AI agents for real work.

You sit down at your laptop. You start Claude Code. You give it a task: "Rebuild the landing page, add the new testimonials section, update the pricing table, and deploy it." The agent starts working. Files are being created. Code is being written. Progress is happening.

Then you have to leave. A meeting. A school pickup. A client lunch. Whatever it is, you close your laptop lid and walk away.

Now what? Is the task done? Did it hit an error? Does it need your input on something? You have no idea. You are completely disconnected from the work that is supposedly being done on your behalf. The whole point of an AI agent is that it works while you do other things — but if you can only interact with it from one physical machine, you are still tethered to your desk.

This is the problem Claude Code Remote Control solves. And it solves it in the simplest way imaginable.

How Remote Control Actually Works

The concept is dead simple, even if the engineering behind it is not.

When you start a Claude Code session on your laptop, that session runs locally. Your files are on your machine. The commands execute on your machine. The AI reads and writes to your local filesystem. Nothing changes about that.

What Remote Control adds is a secure tunnel. Your running session becomes accessible from any other device — your phone, a tablet, another computer, any web browser. You see the same session. You can type into it, give it new instructions, approve actions, or just watch it work. The session is not duplicated or moved. It is still running on your laptop. You are just looking at it through a window from somewhere else.

Think of it like a security camera for your AI agent's work, except you can also talk to the agent through the camera.

The key technical detail that matters for business owners: your files never leave your computer. The session streams the interface — the text, the conversation, the status updates — but your actual code, documents, and data stay exactly where they are. On your machine. Behind your firewall. Under your control.

This is not screen sharing. This is not remote desktop. This is a purpose-built connection between your local Claude Code session and whatever device you happen to have in your hand.

The QR Code Experience

Here is what it actually looks like in practice.

The entire connect-and-go process takes about five seconds. Scan, tap, connected. That is it.

What makes this genuinely useful rather than just clever is the auto-reconnect. Real life is messy. You walk through a dead zone. Your phone locks. You switch apps to answer a text. With Remote Control, none of that matters. The session is resilient. It reconnects automatically, every time, without you having to do anything.

Server Mode: 32 Sessions at Once

Remote Control gets more powerful when you combine it with server mode.

By default, Claude Code runs a single session — one conversation, one task at a time. Server mode, activated with the --spawn worktree flag, lets you run multiple sessions simultaneously. Up to 32 concurrent sessions, each working on a different task or a different part of the same project.

For a business owner, think of it this way: instead of one AI employee sitting at one desk doing one thing, you now have up to 32 AI employees, each at their own desk, each working on something different, and you can check in on any of them from your phone.

Each session gets its own isolated workspace (a "worktree" in git terminology), so the agents do not step on each other's work. Agent 1 can be rebuilding your checkout flow while Agent 2 updates your blog content while Agent 3 writes email sequences — all at the same time, all on the same machine, all accessible from your phone.

This is where Remote Control stops being a convenience feature and becomes a genuine force multiplier. The limiting factor is no longer "how fast can one AI work on one thing" — it is "how many things can I direct simultaneously." And the answer, with server mode and remote access, is: a lot more than you could before.

5 Business Scenarios Where This Changes Everything

Remote Control sounds like a developer feature. It is not. Here is how business owners are actually using it.

Security: Why "Files Stay Local" Matters

If you are a business owner handling client data, proprietary processes, financial information, or anything remotely sensitive, this section matters more than all the others combined.

Most remote work tools solve the "access from anywhere" problem by putting your files in the cloud. Google Drive. Dropbox. Cloud-based IDEs. The convenience is real, but so is the trade-off: your data now lives on someone else's servers, subject to their security practices, their data policies, and their breach exposure.

Claude Code Remote Control takes a fundamentally different approach. Your files never move. They sit on your local machine — your laptop, your office server, whatever you use. The remote connection streams the conversation interface, not your data. When the agent reads a file, it reads it locally. When it writes code, it writes it locally. When it runs a command, it runs it locally.

What travels over the connection is the text of your conversation with the agent — your instructions, its responses, status updates. Not your source code. Not your client database. Not your financial records.

For businesses in regulated industries — healthcare, finance, legal — this is not a nice-to-have. It is a requirement. Many compliance frameworks prohibit sensitive data from being stored on third-party cloud services without specific security certifications. With Remote Control, the data never leaves your environment, so those restrictions do not apply to the remote access mechanism itself.

How to Set It Up

Getting started with Remote Control requires no special configuration. If you already have Claude Code installed and running, you are 90% of the way there.

Prerequisites

• Claude Code installed on your laptop or workstation
• An active Anthropic account (any plan that includes Claude Code access)
• A second device — phone, tablet, or any device with a web browser
• The Claude app (iOS or Android) or access to claude.ai/code in a browser

Basic Remote Access

Start Claude Code on your main machine. When you want to connect from another device, the session provides a QR code or URL. Scan or navigate to it. You are connected. The session auto-reconnects if interrupted.

Server Mode for Multiple Sessions

To run multiple concurrent sessions, start Claude Code in server mode with the --spawn worktree flag. Each new session gets its own isolated workspace. You can run up to 32 sessions simultaneously, each accessible independently from any device.

# Start a single session with remote access
claude

# Start server mode for multiple concurrent sessions
claude --spawn worktree

# Each session gets its own workspace and remote URL

The first time you connect from a new device, you authenticate with your Anthropic account. After that, reconnection is automatic. Close your phone, open it again later, and the session is still there.

Frequently Asked Questions

Q: What is Claude Code Remote Control?

Claude Code Remote Control lets you start an AI coding session on your laptop and continue it from any other device — your phone, a tablet, or any web browser. The session keeps running on your local machine, so your files never leave your computer. You connect remotely via a QR code or URL, see the full conversation history, and can give new instructions, approve actions, or just monitor progress. It works with claude.ai/code in any browser, plus the dedicated iOS and Android apps.

Q: Do my files leave my computer when using Remote Control?

No. Your files stay on your local machine at all times. Remote Control streams the session interface — the conversation, status updates, and controls — to your other device. All code, files, and commands execute locally on your machine. This is a critical distinction for businesses handling sensitive data, client information, or proprietary code. Nothing gets uploaded to a separate cloud service for remote access purposes.

Q: How many concurrent Claude Code sessions can I run?

Up to 32 concurrent sessions using server mode with the --spawn worktree flag. Each session gets its own isolated workspace, so multiple agents can work on different tasks simultaneously without interfering with each other. You can monitor and direct each session independently from any device. This turns one AI agent into a team of up to 32, all running on your local machine.

Q: What devices work with Claude Code Remote Control?

Any device with a web browser or the Claude app. This includes iPhones and iPads (via the iOS app or Safari), Android phones and tablets (via the Android app or Chrome), laptops and desktops (via claude.ai/code in any browser), and even smart TVs or any other device with a modern browser. The session auto-reconnects if your connection drops, so you can switch between devices seamlessly without losing your place.