Issue #5: The Map, and Your First Automatic Briefing

New here? If this landed in spam or promotions, drag it to your inbox and mark it "not junk" so you don't miss next week. Takes five seconds and it genuinely helps.

I need to own something.

For about a year, I used this stuff without really knowing what I was doing. Not in a dangerous way. In an embarrassing way. I opened a chat window for everything, because a chat window was the only room I knew about.

It's like owning a house and only ever using the kitchen. The kitchen's great. But there's a garage out back full of tools, and I spent a year at the counter making sandwiches.

Nobody handed me a map. I found the other rooms by tripping into them.

Last week I promised you that map. No homework, just clarity. Here it is, and it fits in one breath.

There are three ways to work with Claude.

1. CHAT. Talk it through. This is for thinking, learning, and understanding. You've got something in your head and you want to work it out with somebody. Reach for chat when: what you need isn't a task, it's a conversation. "Help me understand this." "What am I missing?" "Explain this like I'm new."

2. CO-WORK. Hand off a job. This is for work you want done while you go do something else. Research, digging, gathering, summarizing, on repeat. Reach for co-work when: you catch yourself thinking "I wish somebody would just go find this out for me, every week, without me asking."

3. CODE. Build things. This is for making something that didn't exist before. An app, a game, a little tool that solves your one specific problem. Reach for code when: you don't want an answer, you want a thing.

That's the map. That's the whole thing.

And here's the part that makes it stick, because you've already been living on it. Issue 2 and Issue 3, going deep on a book, me working through Job and asking about one verse or the meaning of a single word? That was chat. Last week, the podcast summary and the daily digest, handing off the legwork? That was co-work. Next week, the apps I've started building for myself? That's code.

One thing before we go, because it confused me for months. Co-work is both a move and a room. The move is handing off a job instead of asking a question, and you can do that today, in a plain browser chat, for free. You did it last week. The room is a feature literally called Cowork, the upgraded version of the same move with a timer bolted on. Same idea. Nicer garage. More on that in the Power Move.

Now, I'm not the guy who hands you a map and walks off. So let's use it.

Last week you built a briefing and ran it by hand. I said the automatic version was a later rung, and that we'd climb it together. It's later. Let's climb.

Watch this. Every weekday morning at 7am, before I'm even fully awake, a briefing gets written for me. I didn't ask. I didn't open anything. It just runs, and it's sitting there waiting when I come get my coffee.

I follow the markets. Not as a job, as a thing I find genuinely fascinating, the way some guys are into vintage cars. So mine is a Daily Market Brief: a read on what's happening, a walk through the areas I care about, and a short ranked list of what's most worth my attention. Written in the voice of somebody sharp who commits to a view instead of hedging every sentence into mush. I've got others, too. A weekly digest of my favorite podcast. A general morning briefing. They all just run.

Now, real talk before anybody gets excited about the markets part. That's just my hobby. This is not investment advice, I'm not telling you to point AI at your money, and AI gets things wrong. Verify anything that matters. The topic is not the lesson here.

The recipe is the lesson, and it works for anything you care about. Your industry. Your team. Your league. Your side hustle. Here are the four moves that make a briefing prompt actually good.

Use case: A recurring briefing on any topic you're tired of being half-informed about

This is my real recipe with the markets stripped out and handed to you. All four moves are built in: the role, the research, the shape, the point of view. Fill in the brackets before you paste. Everything else works as written.

What to put in the brackets. Here's a filled-in one, start to finish, so you can see it working:

Swap in yours. A youth sports parent tracking a league. A nurse tracking their specialty. Somebody who just wants to actually understand their own industry for once.

Run it once by hand. Then steer it like a new hire. "Too long." "More on that second section." "You're hedging, commit." Two or three rounds and it's yours. Save the version that finally sounds right, because that's the thing you're about to put on a timer.

Here's the feature, and this is the week you step out of the browser.

It's called Claude Cowork, and it's where you hand Claude a job on a schedule instead of only when you ask. It's the difference between a helper who waits for you to call, and one who already has it sitting on your desk when you walk in.

Three honest things before you go looking for it.

One: you need the Claude desktop app, not the browser version. Go to claude.ai/download and put the Claude app on your actual computer. Same login, same conversations, same everything. It just has rooms the browser doesn't. Two minutes. By the way, Claude can sometimes change where things are. If the desktop app isn’t there, just ask Claude where to go, it knows.

Two: this lives on the paid plans. Pro, Max, Team, or Enterprise. But if you're on free, you're not shut out of anything that matters this week. The recipe above is the actual skill, and running it by hand takes thirty seconds. I ran mine by hand for a long time before I ever put it on a timer. The prompt is the hard part. The timer is the easy part.

Three, and this one trips people up: there are two flavors. A local task runs on your own machine, so it only fires while the desktop app is open and your computer is awake. A remote routine runs up on Anthropic's cloud, so it goes even when your computer is off. Think of a load of laundry: local is the machine in your garage, running only when you're home with the power on; remote is dropping it at the cleaners, done whether you're there or not. For a 7am morning briefing, you want remote.

One catch. A remote routine lives on a server, so it can work with the web, but it can't reach the files on your own desktop. If your job needs your own documents, that's a local task. A news-and-web briefing like ours is perfect for remote.

The schedule menu is short: manual, hourly, daily, weekdays, or weekly. Mine is weekdays, because the market doesn't open on Sunday and I don't need a briefing about nothing.

One last thing, because people build the wrong picture here. When it runs, the result is sitting there waiting for you inside Claude. You go open it and read it. Nothing gets emailed to you. (I have a fussier setup that pipes mine somewhere else. Project for another day.)

Here's the concept underneath all of it, and it's bigger than the feature: you just hired for a shift. Not a task. Something that happens on a rhythm whether or not you remember to ask. That's a genuinely different relationship with a tool, and it's the part that changed my mornings.

No desktop app. No paid plan. No download, no settings, no timer. Plain browser chat, five minutes, and it works on the free plan.

Open Claude the same way you always have. Pick the one topic you'd genuinely like a briefing on. Your industry. Your kid's sport. Your neighborhood. Something you're always half-informed about.

Fill in the brackets in the Prompt of the Week. Paste it in. Read what comes back.

Then do the one thing almost everybody skips. Steer it once. Tell it what was wrong.

Watch it come back sharper. That's the whole skill right there. A schedule is just a timer bolted onto a prompt you already made good. Get the prompt good first and you're ahead of most people who have the fancy version and never taught it anything.

I'll go first, like always.

For years my mornings started with a scroll. Phone in hand before my feet hit the floor, thumbing through posts and headlines, twenty minutes gone, and I'd walk into the kitchen knowing a hundred small things and understanding none of them. Busy, not informed. There's a real difference, and I was on the wrong side of it.

Now the brief is written before I'm awake. I open it with coffee. I know what's going on, I know what somebody with a real point of view thinks about it, and I know what would have to happen for that view to change.

But the actual win isn't the time I saved. I stopped spending my mornings gathering and started spending them thinking. The gathering was grunt work I was never good at and never enjoyed. The thinking is the part I'm actually here for, and for years I was too worn out from the hunt to do much of it.

That's the trade. Not "AI does my work." More like: the part that was eating me alive got handed off, and I got to keep the part I love.

And listen, I'm not smarter than you. I got here by fumbling into it and being willing to look dumb for a while. If you're sitting there thinking everybody else already has a system, they don't. Most people are still scrolling in bed.

So here's your hand-off, and it's one thing, not five.

Pick the one topic you're tired of being half-informed about. Fill in the brackets in the Prompt of the Week. Run it. Then steer it twice until it's genuinely good, and save that chat.

That's the whole assignment. If you've got the desktop app and a paid plan, spend ten more seconds and set it as a remote routine on weekdays. If you don't, run it by hand on Monday mornings like I did for months. Same win, one extra click.

You started this newsletter asking Claude questions. Today you put it on a shift.

Next week we open the third door. Code. Not the scary version. Here's what I actually mean. I've started building my own little apps instead of downloading other people's. A notes app that finally works the way my brain does. A tracker for my workouts. A small math app to help my kid practice. I'm not an engineer, and I never wrote a line of it. I described what I wanted, in plain English, and it got built. (My 8-year-old has picked up enough to build his own stuff too. That's how not-hard this turned out to be.) Next week I show you how, start to finish.

After that, I'm going to slow us down. We've moved fast, chat, co-work, now code, and each of those rooms has good stuff hiding in it that I've blown right past. Skills. Artifacts. Connectors. A handful of things that made me go "wait, it can do that?" We'll go back through, slower this time, and I'll put them in your hands one at a time.

And somewhere in there, I owe you the real conversation. Not what any of this does. Why it matters. Why I'm up at night writing this to you, and what I think it could mean for your work, your kids, your actual life. That's the one I'm building toward. It's the whole point.

Until then.

Stop overthinking it.

Just get started.

JGS

P.S. Housekeeping, because it matters: none of this is investment advice and I'm not a financial advisor. Markets just happen to be my hobby, the same way yours might be woodworking. Point this recipe at whatever you love instead, and double-check anything that counts. AI gets things wrong the same way people do, confidently.

P.P.S. Hit reply and tell me what your daily briefing would be about. I read every email, and your topic might be the example in a future issue.