Your schedule runs with the laptop shut
Claude working on a schedule from Anthropic's computers instead of yours, and the free credit to try it.
First, Claim The Credit
Here from the video? This is the link. Anthropic are giving people already on a Pro or Max plan a one off credit to spend on cloud sessions, up to $250 depending on your plan. Claim it before 11:59pm Pacific time on 7 October, which is about 8am on the 8th in the UK. The claim page asks you to connect a GitHub account, which is free and takes a couple of minutes, because a cloud session works from a copy of your files kept there.
Claim the creditWhatever you have not used by 4 November goes, and after that cloud sessions come out of your normal plan like everything else.
What you'll walk away with
- The link to claim the free credit, and what it asks of you
- What a cloud session and a routine are, in plain English
- A prompt that sets up your first scheduled job with the rules it must keep, and runs it once so you see it working today
Claiming takes a couple of minutes. Setting up the first routine takes about fifteen.
Why It Works
A job scheduled on your own laptop is an assistant who can only work while you are in the office, so the moment the lid shuts they go home and the 7am job waits until you are back. That is why people buy a Mac Mini just to leave switched on, and why I ended up with a script that wakes my own Mac at twenty to nine every night. A cloud session is the same assistant working from their own desk at Anthropic, and a routine is that session on a timer, so the job gets done whether your laptop is open or not.
Step One, Your First Routine
Paste this into Claude Code, in the desktop app or the terminal. It writes the rules in first, drafts and never sends, then interviews you for the one job worth running without you. A routine can only see a private GitHub repository and the apps you have connected to Claude, never the files on your laptop, so it checks what the job will be able to reach and sets up anything missing with you before it builds. Then it hands over to /schedule, which saves the routine, and runs it once so the first result is in front of you today. If you ever set one up by hand in the desktop app, choose Cloud rather than Local, because Local is a schedule on your own computer again.
Step Two, Check It Ran
Every run is a session of its own, listed at claude.ai/code beside your other sessions, so you can open any of them and read exactly what it did. Read the first few properly, because that is where you find out what it gets wrong. The morning after the first scheduled run, open that session and paste this into it. Anything you want changed, tell it there and have it update the routine.
One Honest Bit
A cloud session is not your laptop. It cannot see the files sitting on your computer, only what is in the copy on GitHub and whatever it can reach through the apps you have connected to Claude, so a job that needs a spreadsheet on your desktop has to have that spreadsheet moved somewhere it can read first.
Routines are still a preview, and while one is running nobody is there to say yes or no, so it can act in the apps you have connected without asking each time. That is why the drafts only rule goes in before anything else, why it gets only the connectors the job needs, and why the first week is for reading what it did rather than trusting it. Once the credit is spent, every run comes out of your plan like any other chat, the shortest gap between runs is an hour, and there is a daily cap on how many can start.
Where This Goes Next
If you want a job that is already proven to run on a schedule, your inbox already has the replies written is the evening one, and the meeting your AI books with you every morning is the one I read first thing. If it could not reach your mail or calendar, tools are how your AI stops being useless is which accounts to connect first.
A routine is an agent with working hours. If the idea of an agent is still fuzzy, an AI agent is a chat with a job is the plain English version and the start of the series.
Want to get better at this?
Let's set up your agent ecosystem.
This page is one job. The whole thing is a set of them running off the same foundation, the folder your AI works out of, the brief on how you like things done, the accounts it can reach and the skills that run off all of it, so your admin gets done the way you would do it. It is what I teach one to one, on your own machine and against your own work, and the quickest way to find out what yours would look like is a proper chat about it.
Not ready for a call? Start with the free agent series, what an AI agent actually is, and build up from there.