Your OpenAI dot, typing into your Claude Code. Here is the setup from the reel.
In the reel, a dot reads a Claude Code session that is waiting on an answer, then types the reply it was told to send. This guide sets up that link. Claude Code stays on your computer, and your dot opens the session through claude.ai in its own cloud browser. You sign in yourself, have the dot read a session, and then approve one test message. There is no API to build, no MCP bridge, and nothing to install on the dot's computer.
The same six steps with the exact wording to send your dot, a read-only check, one approved test message, and a table of fixes.
02Sign in to claude.ai in your dot's cloud browser yourself
03Have the dot read a session, then send one message you approved
Steps
6
The guide and the PDF
Free
Claude plan needed: Pro, Max, Team or Enterprise
Paid
Installs needed on the dot's computer
0
What you will have when you finish
Your dot, able to read a Claude Code session and send the one message you tell it to.
A Claude Code session on your computer that shows up under Remote Control at claude.ai/code
Your Claude account signed in, by you, inside your dot's cloud browser
A read-only check: your dot names the session and tells you what it is waiting for
One test message you worded yourself, sent once and confirmed in the same conversation
Instructions, typed or spoken, that name the session and the action, so nothing gets approved by accident
Six steps
Claude Code stays on your computer. Your dot works through claude.ai in its own browser. This is the setup we used on 7 October 2026, checked against Anthropic's documentation the same day. The test message in step 6 is one we suggest and have not run.
Claude Code v2.1.51 or later on your own computer, signed in with your Claude account rather than an API key, on a Pro, Max, Team or Enterprise plan. On Team and Enterprise, an Owner has to turn Remote Control on first. Keep the computer awake and online with Claude Code running. You also need a dot with a cloud computer and browser: dots launched on 29 September 2026 and access can vary, so check yours first. Voice is optional.
01
Put your Claude Code session on Remote Control. If the session you want already shows as online in Remote Control, leave it running and go to step 2. If it does not, type the command below inside that session. The name after the command is the one your dot will look for later, so pick one you will recognize.
Type in your Claude Code session
/remote-control Website Work
Your part: swap Website Work for your own session name, and accept the one-time Remote Control confirmation if Claude Code shows it. To start a fresh session instead, run claude --remote-control "Website Work" in your project's terminal. Do not start a replacement session because the browser needs to sign in.
02
Ask your dot to open its own browser and let you take over to sign in. Send the request below in your usual conversation. When the dot hands you the browser, check that you are looking at the dot's cloud computer. Signing in to Claude in your laptop's browser does not sign in the dot's.
Send to your dot
Open claude.ai/code in your cloud browser and let me take over to sign in. Don't open or send anything in my Claude Code sessions yet.
Your part: use the browser handoff your dot gives you. In our setup this was the visible Chrome window on the dot's cloud desktop. The separate command workspace was a different environment, and signing in there did nothing for the browser.
03
Sign in to the right Claude account yourself, in the dot's browser. Go to claude.ai/code and use the account, and the organization if you have one, that owns your local sessions. Then open one real Remote Control session and check that its conversation loads. A list of sessions on its own does not prove access.
Your part: type passwords and verification codes only into the real sign-in page, never into a chat message. If Claude asks for another sign-in or device verification, finish it yourself. Trusted Devices, if your account uses it, asks you to enroll the device and sign in recently before a remote session opens. If the browser cannot finish a verification, stop there and do not turn off a security setting to get past it.
04
Give the browser back to your dot. Finish the handoff, then tell the dot that sign-in is done and which session to look at. Stop typing and clicking in the shared desktop while the dot is working.
Your part: one controller at a time. In our setup, overlapping input sent text to the wrong window. If you want control back, ask your dot to pause first.
05
Check a session without changing it. Name the session and ask for a read and nothing else. Have your dot report the session name and the latest status it can see, then compare that with your local Claude Code session. Fix a wrong title, an old conversation or missing text before you let it send anything.
Send to your dot, with your own session name
Open the Website Work session. Read its latest status and tell me whether it needs input from me. Don't send a message, approve anything, or change the session.
Your part: compare what the dot reports with the session on your own screen. If they disagree, stop and sort that out first.
06
Send one test message that you approved word for word. Use a session with no important work underway, or skip this step and stay with reads for now. Name the session and the exact message. Your dot should confirm the session, type the text, send it once, then check that your message appears and Claude starts to reply.
Send to your dot, with your own test session
In the Connection Test session, send exactly: Reply with CONNECTION OK only. Do not run tools or change files.
Your part: if the dot is unsure whether the message went, have it look at the conversation before it tries again. We have not run this exact test. In the reel, the dot read a video editing session and, when told to, sent "I didn't upload that reel." The message appeared and Claude began replying.
A status check is not permission to answer questions, approve tool requests, edit files, start work or stop an agent. Give your dot a separate, specific instruction for each. To use voice, call your dot and name the session and the action the same way you wrote them here. Voice changes how you instruct the dot. The connection still runs through the signed-in cloud browser.
The specifics
What it costs
The guide and PDF are free. Remote Control needs a Claude Pro, Max, Team or Enterprise plan and does not work with an API key.
Tested on
Our own session and dot on 7 October 2026. The dot read a session and sent a reply we dictated. Step 6's test message was not run.
What your dot can reach
Whatever the signed-in browser can do in that session, including replies and tool approvals. Name one action per instruction.
Where your data goes
Claude Code keeps running on your computer. Your dot reads the session through claude.ai in its cloud browser, so treat what it shows as visible to the dot.
Left out on purpose
No custom API and no MCP bridge. Claude Code and Claude Desktop on the dot's computer were tried and dropped, and neither is needed.
What you need from OpenAI
A dot with a cloud computer and browser. Dots launched on 29 September 2026 and access can vary, so check yours first.
Questions people ask
Will my dot send or approve things without me asking?
It acts on what you tell it, so word each request narrowly. A status check is not permission to answer questions, approve tool requests, edit files or stop an agent. Keep Claude's own safeguards on.
The session list loads, but the session will not open. Why?
Open the session yourself in the dot's cloud browser and finish any extra sign-in or device verification. That happened in our setup. If the browser cannot finish the check, stop there and leave your security settings alone.
Do I have to phone my dot like in the reel?
No. Typed instructions work the same way. Voice is optional and needs voice access. It changes how you instruct the dot, and the connection still runs through the signed-in cloud browser.
What if my computer goes to sleep?
Keep it awake and online with Claude Code running. Anthropic's docs say a sleeping laptop reconnects when it comes back online. If the session goes missing, check power, network and the Claude process, then run claude doctor.
That was the free part
The file is yours. Nothing above asked for an email.
If it earned a minute of your time, the paid part is one scroll down.
Everything above this line is free and stays free. Below it is what we sell. The prices are the prices, and the two subscriptions carry 30 days money back on the first month.
Mica Social Media Analyzer
Your Instagram, answered by the ChatGPT or Claude you already use. Connect the account once, then ask in plain words what worked and what to post next.
Who it is for
You post on Instagram and you are done guessing from the Insights tab.
Price
$29/ month
What you get
Which Reels held attention, ranked, without opening Insights
Readings at 1, 24 and 72 hours after every post, so you see what happened while it was moving
Next captions written in the style of your most saved posts
Read only. It never posts or changes anything
The facts
7 days free with a card on file. About a dollar a day after that.
Our guarantee
30 days money back on the first month. Email guru@graniteai.co and we refund it.
Send us access to your repo. We run the nine-lane review, change nothing, and hand back the report with a 30 minute call on the three gaps to close first.
Who it is for
You have a product you charge for, or want to, and you would rather read the verdict than run the agents.
Price
$349One week
What you get
The report at docs/commercial-readiness.md: a verdict, a scorecard, every gap with the file to open and how to verify it
Legal, security, billing, support, marketing, operations and the rest, checked at once
A 30 minute call on the three gaps to close first
Nothing in your repo changed. The workflow prompt is yours to run afterward
The facts
The review is the one in the video. On our own product it found 87 gaps. The overnight run that followed was 148 commits and one pull request.
Our guarantee
If the review finds no gap worth your time, you pay nothing.
The first run, with Jason
We run the review on your repo together, on a call, and you leave with the report in your repo and the three gaps to close first.
Who it is for
You have the kit and a repo, and you would rather do the first run with someone who has done it.
Price
$9960 minutes
What you get
The review running on your machine, in your Claude Code
The verdict read together, and the three gaps picked
The workflow prompt set up so you can run it that night
The call recording
The facts
New. You would be one of the first. The run itself is the one in the video: 190 agents, 148 commits, one pull request.
Our guarantee
If the review will not run on your setup during the call, there is no charge.
Mica Social
An AI that writes, designs, checks and publishes a post to your Facebook and Instagram every day. You send a photo when you have one. It does the rest.
Who it is for
You know the feed should be alive every day and it is not.
Price
$49/ month
What you get
A finished post every day, with a photo made for it. Nothing to shoot
Timed to your local weather, season and county numbers
Five checks before anything publishes. No approvals, no calendar, no logins
Every photo you send becomes its own extra post
The facts
An agency running a daily feed bills $500 to $1,500 a month. 3 days free with a card on file.
Our guarantee
30 days money back on the first month. Email guru@graniteai.co and we refund it.