Free guide
Build Your Own Morning Briefing
The exact prompt, the one-page context template, and the setup for all three rungs, from the video. Start with the ten-minute version. Move up only if you want to.
Just want the prompt? → Rung 1No email sign-up. Everything below is free on this page, the same as the ChatGPT setup guide.
The three rungs
A chief of staff is the person whose whole job is to get ahead of your day. Almost nobody outside a CEO's office ever had one. These three rungs build your own, each one trading a little manual effort for a little more automatic.
Rung 1
Automatic, generic
A scheduled prompt in ChatGPT. No code, ten minutes, briefs you on your calendar. Start here.
Rung 2
Personal, manual
A Claude Project holding one page about your actual work. You trigger it, but the answer is about your week, not the news.
Rung 3
Personal, automatic
A machine that wakes itself, reads your files, and emails you the result. The version in the video.
The setup
Do rung 1 today. Move up only when rung 1 stops being enough.
Rung 1: the ten-minute version
In ChatGPT, open the sidebar and find Scheduled tasks (sometimes under a "Tasks" or template menu). Create a new task, set it for a weekday morning, and paste this as the prompt:
Every weekday at 7am, give me a five-line brief. Most important thing first. My calendar for today, anything I flagged yesterday, and the one task I should protect time for. No preamble.
Save it, then leave it alone for a week. Don't tune it yet. Just see what a week of it feels like before you touch anything.
If scheduled tasks aren't on your plan or account yet: paste your calendar for the day into a regular chat with the same prompt below it. Same brief, you just have to ask for it yourself each morning.
Rung 2: make it about your work
A generic brief goes stale fast, not because the AI is bad, but because it's never read anything of yours. Rung 2 fixes that with one page of context in a Claude Project.
Write one page, three headings:
What I'm working on right now: What matters this quarter: Who's waiting on me, for what:
Create a Claude Project, add that page as a project document, then open it each morning and type "morning brief." The honest trade: Claude doesn't run on a schedule yet, so you're the trigger. What comes back is worth it, specific to your actual week instead of generic advice.
Rung 3: the version that runs itself
This is the one in the video. It's not one tool, it's three parts wired together:
- Something that wakes the machine on a schedule (a built-in OS setting, not an app).
- An AI agent that's allowed to read your files, pointed at instructions to act as your chief of staff and write the brief.
- A step that emails you the result, so it's waiting when you wake up.
This rung is a real build, not a ten-minute setup, and it's genuinely optional. Most people get most of the value from rung 1 or rung 2. Move up only if you want the triggering itself to disappear.
Build the alarm before you build the feature. Whatever your version is, the first thing you add is: email me when this breaks. A quiet failure looks exactly like everything being fine, and the only way to catch one is to have something that shouts.
The order
If you only do one thing today
- Rung 1 (the scheduled prompt, ten minutes)
- Live on it for a week before you touch anything
- Write your one-pager whenever rung 1 starts feeling generic
- Rung 3 only if you still want it after that
And to be clear: rungs 1 and 2 need nothing beyond a ChatGPT or Claude account you may already have. No extra subscription, no automation platform, no server.