Updated June 9, 2026

How to Claude Code from your phone

Remote Control gives you the hands. Drafty gives you the eyes — and the memory.

Claude Code can run from your phone now. Remote Control lets you start a session, answer its questions, and keep work moving while you're away from your desk — the live, back-and-forth part works well on a small screen.

The wall you hit is everything that isn't back-and-forth. A chat window can't show you a plan, hold your feedback in one place, or remember an idea you have mid-stream. That's the gap Drafty fills.

Here's the honest frame: Remote Control is the hands — the live session, the typing, the building, running on your Mac. Drafty is the eyes and the memory — what you need to see, steer, and keep.

You can already Claude Code from your phone

Remote Control drives a live Claude Code session from wherever you are. You kick off work, it asks a question, you answer, it keeps going. For the conversational part of coding — decisions, nudges, "try X instead" — your phone is genuinely enough.

It's the moment Claude produces something to look at that the chat starts to strain.

What a chat on your phone can't do

You can't see the plan. Claude writes a plan, a spec, or a design, and it lands as a wall of text in the chat. There's nothing rendered to look at, so on a phone you end up approving things you can't really see.

Giving feedback means screenshotting. When something's off, you screenshot it, paste it back into the chat, and type out which part you mean. It's lossy and slow, and Claude still has to guess the spot.

Threads get lost. Everything is one long conversation. What's been handled, what's still open, what referred to what — it all scrolls away.

Ideas have nowhere to go. Halfway through one chat you think of something for a different project. There's no place to put it, so it either derails the thread you're in or you forget it.

Drafty: the layer on top

Each of those has a fix, and underneath they're the same fix wearing four hats — put the work on a canvas, a link that renders anywhere, and let you point at it.

See it, rendered

Claude publishes the plan, design, or page as a Drafty canvas. You open the link full-screen on your phone and look at the actual thing, instead of reading its source.

Without Drafty
<!doctype html>
<section class="hero">
  <h1>Ship faster</h1>
  <p>Lorem…</p>
</section><style>…
raw .html · won't render
With Drafty
Same file, same phone. Without Drafty you get raw HTML that won't render; with Drafty it's a real page you can read.

Point, don't describe

Tap the exact element — a heading, a button, a row — and leave a comment pinned to it. No screenshot, no "the third card down." Describing a piece of an interface in words is the slow, error-prone part; pointing removes it.

Describe it
"the first tab from the left… the icons one?"
Point at it
this one
"The first from the left" makes Claude hunt and count. Tap the tab instead — "this one" pins to the exact element.

Threads that hold

Every comment is its own thread, anchored to its element and resolvable. You can see what's open and what's handled across the whole canvas, instead of digging back through a chat.

A memory layer

This is the part that's easy to miss. Chats are throwaway; your projects are long-lived. The memory layer is where a stray idea becomes a durable note on the project it actually belongs to. Mid-conversation, drop the idea onto another initiative's canvas — linked back to where it came from — instead of losing it to scrollback. Over time, your canvases become the map of every initiative, and the chat is just the live wire.

Drafty — phone article
live chat · what you're on now
idea: reuse this layout on Koala
+ filed
Koala — homepage
another initiative's canvas
Drafty — phone article
↑ backlink: where the idea came from
An idea mid-chat files itself onto the right initiative's canvas — linked back to where it came from. Your canvases are the memory; the chat is just the live wire.

The loop

Day to day, it's a loop. You comment on the canvas. Claude, in the session, reads the comments, makes the change, and ships a new version at the same link. You glance, and either approve or comment again.

Your comments are the sign-off — there's no separate approval step. A two-word "ship it" works because you can see exactly what shipped.

before
Welcome to our website
JKmake this heading punchier
✓ v18
Ship in minutes, not weeks.
Comment on the weak heading → Claude rewrites it in place → ships the better version, marked done.

The honest split

So the division of labour is clear: Remote Control is the hands, Drafty is the eyes and the memory.

Remote Control — the hands
The live session: the typing, the building, the terminal — running on your Mac.
Drafty — the eyes + memory
What you see, track, and keep: rendered canvases, anchored comments, ideas filed by initiative.
The Mac does the work; your phone is how you see it, steer it, and remember it.

One honest caveat: Claude acts on your comments when a session is live — it isn't a bot answering on its own around the clock. But nothing is lost if you comment when no one's home. The next time Claude is on the project, it reconciles the whole inbox and catches up. Leave the note whenever the thought lands. Chat for the conversation; canvases for everything that has to persist.

Get started

Add Drafty to your home screen so it's one tap. The next time Claude publishes a canvas, open it on your phone, comment on one thing, and watch it ship.

Drafty icon on the home screenDrafty Your-canvases list, full-screen on iPhone
One tap from your home screen into all your canvases — see open it like an app.

Related help