Cowork setup beginners always get wrong

Are you looking to get in front of 150,000+ AI readers? Sponsor CoolDeep AI Newsletter.

Was this email forwarded to you? Subscribe here!

Today’s Sponsor

No bots. No awkward intros. Just great notes.

Ever had a meeting where a random bot joins the call and suddenly everyone’s distracted?

Granola works differently. No meeting bots. Nothing joins your call.

It transcribes directly from your device’s audio; on your computer or phone. Works with Zoom, Google Meet, Microsoft Teams, and even in-person conversations.

You stay focused. Jot down notes like you normally would. Granola quietly handles the rest in the background.

Want to be extra thoughtful? You can auto-send a quick consent note beforehand, too.

Try Granola free

The whole system runs on two text files

That is it. Two files.

The first one is called claude.md. Think of it as the instruction manual for your AI. It tells Cowork how to behave, what rules to follow, and where to look when it needs more information. You write it once. Cowork reads it every session automatically.

The second one is memory.md. This is the notepad. Whenever you say "remember this," Cowork writes it down here. The next time you start a session, it reads this file before it responds to anything. That is how it knows what you are working on. That is how it remembers your preferences. That is how it sounds like it actually knows you.

Open either of these files in any basic notes app and you will see plain English. No code. No technical language. Just rules written the way you would explain something to another person.

That is the entire system. Two files. Plain text. Done.

Why it feels intimidating (and why that goes away fast)

The reason most beginners freeze is because they see the folder structure and the .md file extensions and assume it requires technical knowledge to operate.

It does not.

Here is a helpful way to think about it. The US has a constitution that applies to every person in every state. Then each state has its own laws that stack on top. Claude Cowork works the same way.

Your root claude.md file is the constitution. Rules that apply in every session, no matter what you are doing. Below that, you have workstations, which are just folders for different areas of your life. Email. Personal finances. Your newsletter. Each one has its own rules that only kick in when you are working on that specific thing.

You do not need to build all of this on day one. You need to start with just the root level and let everything else grow naturally.

The Free AI Guide That Can Save You Time (And Money)

Want an A+ when it comes to AI? Subscribe to The Current written by Kim Komando and get a free AI Prompt Hack Pack filled with powerful prompts you can enter into ChatGPT to save you hours (and money) every week.

Get The Free AI Guide Here

*Partnered

The exact beginner setup in four steps

Step one: create your main folder.

Name it anything. "Cowork OS" works. This is where everything will live.

Step two: download the starter files.

You need three files to begin: claude.md, memory.md, and a voice principles file. There are free templates available that give you the exact structure without starting from zero. Drop the first two files directly into your main folder. Create a subfolder called "resources" and put the voice principles file in there.

Step three: mount the folder inside Cowork.

Open Claude Cowork on your desktop. Select your parent folder. Click "always allow." Star it so Cowork defaults to this folder every time you start a session.

Step four: let Cowork build your voice profile.

This part surprises people. You do not have to write your voice principles yourself. Connect Cowork to Gmail and it will read your last 30 sent emails, extract your natural writing patterns, and fill in the voice principles file for you. If you do not want to connect Gmail, paste in five writing samples you like and it does the same thing from those.

After this, Cowork will know how you write before you type a single word.

That is the entire beginner setup. Most people finish it in under an hour.

What you build next (and when)

Once the root level is working, you add workstations only when you actually need them.

Do you write emails regularly? Build an Email HQ workstation. Working on a specific project? Create a project subfolder with its own memory file. Managing personal finances? There is a workstation for that too, where Cowork reads your credit card statements and builds a spending tracker automatically.

The critical thing for beginners is this: do not try to build everything at once. Start with two or three workstations. Learn how the files interact. Then add more when a real need shows up.

The system compounds. Every rule you add, every memory Cowork writes, every preference it learns makes the next session sharper than the last. Someone who starts today will always have an advantage over someone who starts a week from now, because the rules and memory build up over time.

At the end of every session, type /session audit. Cowork will scan the entire conversation, find anything worth saving, and write it to memory automatically. You do not have to manually update anything. You just close the session knowing that next time, everything will still be there.

Start with the two files. Mount the folder. Let it learn your voice. Then work inside it for a week and watch what happens.

Talk soon, Stay curious,

CoolDeep AI
Helping you stay ahead with AI & productivity