80% of Claude Code explained: the five things I wish someone had told me

This is what I wish someone had told me back when I was first starting with Claude Code.

The name put me off. "Code" sounds like it's for developers, and every guide I found opened with a black screen full of commands and a list of words it assumed I already knew. I'm not a developer. I'm an operator. I wanted Claude to help me run my business, not build me an app.

What I found out is that Claude Code is the reason my AI knows who I am. The reason I use Claude Code over Claude chat is because it has rules, I give it a personality, and it connects to my computer, so it knows who I am all the time. That's the whole difference, and it's what my AI COO and my team of AI employees run on.

It sounds technical. It really isn't. Here's the 80% you need.

The five things, in order

1. Claude Code vs terminal. Why you need to know both, and how to open the terminal without leaving Claude.
2. How to connect Claude to your computer. One folder, picked before your first message.
3. The Customize tab. Skills, connectors and plugins.
4. Routines. Claude doing jobs on repeat.
5. Artifacts and Dispatch. The things Claude makes for you, and sending it work from your phone.

They're in that order because each one builds on the last. Claude can't follow your skills until it can read your files, and a routine is just a skill on a timer.

What you need before you start

• The Claude desktop app on your Mac or Windows computer. Claude Code lives in the Code tab, next to Chat and Cowork.
• A paid Claude plan. Claude Code needs Pro or above. Dispatch needs Pro or Max.
• One folder with your business in it. Your notes, your offers, your client details, your processes. If you haven't got one yet, the Co-Founder Setup walks you through it.

You don't need a coding app, a developer or any commands memorised.

1. Claude Code vs Terminal

Claude Code is basically Claude with access to your computer. So instead of just chatting to you, it can actually read your files, create things, change things and run commands for you.

Claude chat is someone you ring for advice. Claude Code is someone sitting at your desk with your computer open in front of them. Anthropic's own description is that Claude "reads your project files, makes changes, and runs commands".

The terminal is the word that scares people off, so let's take the fear out of it. It's just an app that lets you type instructions to your computer instead of clicking around. That's all it is. Every Mac and Windows computer has one built in.

You need to know what it is for one reason: when Claude Code does things on your computer, some of those things are terminal commands. Installing something, moving a batch of files, checking what's in a folder. You don't need to type them. Claude runs them for you, and asks your permission first, depending on the setting you pick at the bottom of the chat.

And you don't need to sit in a scary black window to use any of it. The terminal is now built into the Code tab itself.

Explain what you're about to do in plain English before you run any
terminal command, & ask me before you change or delete anything.

That's a good first prompt for any new chat. Claude will tell you what it's doing and why, in words you understand, before it touches anything.

If you ever want to see the terminal for yourself:

1. Open a chat in the Code tab.
2. Click Terminal in the bar at the top of the chat. On a Mac you can also press Ctrl and the ` key.
3. A terminal opens beside your chat, already pointing at the folder you're working in.

You'll see a plain window with your folder name in it. You can close it again and never open it, and Claude will still do everything it needs to.

2. How to connect Claude to your computer

This is the step that makes it yours. Claude reads whatever folder you point it at, so if you point it at the folder where your business lives, it answers about your business, not in generic terms.

Read this folder & tell me, in five bullet points, what you now know
about my business. Then tell me what's missing that you'd need to help
me run it.

Before you send that first message, set the chat up:

1. Click Code at the top of the Claude app, then start a new chat.
2. At the bottom of the chat, in the box where you type, choose where Claude runs. Local means on your computer, with your files. Cloud means on Anthropic's computers, so it keeps going after you close the app, but it can't see the files on your computer. SSH is for a remote machine you manage, and you can ignore it.
3. Pick Local.
4. Select the folder you want it to work from.
5. Type your prompt and press Enter.

You'll see Claude open your files, read them, and come back with its five bullet points. That's it. That's how you connect Claude to your computer, and then you can chat away.

The part that made the biggest difference for me is a rules file. Claude Code reads a plain-text file called CLAUDE.md at the start of every chat. Mine tells it my name (Samantha, never Sam), my businesses, where my vault lives and how I like to work. Think of it as the induction pack you'd give a new hire, except Claude reads it every morning. You don't have to write it yourself.

Write me a CLAUDE.md for this folder. Ask me five questions about who I
am, what my business does & how I like to work, then write it from my
answers.

3. The Customize tab: skills, connectors and plugins

Click Customize in the sidebar and you'll find the three things that turn Claude from a clever chat into something that works the way you work. You can also reach all three from the + button next to the box where you type.

I want to set up Claude for my business. Look at what's in this folder &
tell me which connectors, skills and plugins would save me the most time.
Start with one.

Connectors give Claude access to the other tools you use. Google Calendar, Gmail, Slack, Notion, your task manager. Once a connector is on, Claude can do the job inside that app instead of telling you how. If you've switched connectors on in Claude chat before, it's the same idea. Mine connects Claude to ClickUp, which is where all my work lives.

1. Click the + button next to the box where you type.
2. Click Connectors.
3. Pick the app and sign in when it asks.

Skills are basically SOPs. You teach Claude how you want a job done once, and it can do it that way again. The difference between a prompt and a skill is that a prompt is something you have to remember to paste, and a skill is something Claude reaches for on its own when the job comes up. My Scriptwriter runs on skills, and so does the employee that writes the first draft of every one of these blog posts. If you want to build your first one, here's how I build a Claude skill.

Plugins are packs that bundle skills, connectors and other tools together, so you can install a whole set up in one go instead of building it piece by piece. Click +, then Plugins, then Add plugin to browse them. Anthropic has its own marketplace of them. You don't need any plugins to start. They're useful once you know what you're missing.

4. Routines: Claude doing jobs on repeat

This is how you get Claude doing jobs on repeat. Every hour, every night, every Monday morning, whatever the job needs.

Every weekday at 6am, read my calendar & my task list & write me a short
brief of today. Don't change anything, just tell me what's on.

Type that into any chat and Claude will set the routine up for you. If you'd rather click:

1. In the Code tab, click Routines in the sidebar.
2. Click New routine.
3. Choose Local (runs on your computer, with your files) or a cloud routine (runs on Anthropic's computers).
4. Give it a name, write the instructions the way you'd write any message, pick the folder and choose how often it runs.
5. Click Run now once, so you can see it work and approve anything it asks for.

The choice in step 3 matters. A local routine can read your files, but it only runs while the Claude app is open and your computer is awake. A cloud routine runs even when your computer is off, but it can't see the files on your computer. Most of mine are local, because they need my vault.

Mine run all day. A morning brief at 06:06 reads my calendar, ClickUp and my client status pages, and tells me what today is. My inbox gets sorted every 30 minutes from 7am to 9pm. At 22:37 a routine reads what happened that day and updates my knowledge base, so the morning brief is reading the truth and not my memory of it. If you want the full breakdown, here's how I build a Claude routine, and here's my morning brief.

5. Artifacts and Dispatch

Artifacts

Artifacts are basically the actual things Claude makes for you: documents, dashboards, tools, designs, whatever you're building. In Claude Code, you ask for one and Claude builds it as a live web page and gives you a link. When the work changes, the page updates at the same link, and it keeps its old versions so you can go back.

Make me an artifact that shows every client I'm working with, what stage
they're at & what I'm waiting on from each of them.

You'll get a link back in the chat. Open it in the Claude app or your browser. On Pro and Max plans, artifacts are private to you.

Dispatch

And fifth, Dispatch. If you have access to it, this is how you can send Claude work from your phone and have it do the job on your computer.

Dispatch is one ongoing conversation with Claude that lives in the Cowork tab and on the Claude app on your phone. You message it a job, it does the job on your computer with your files and connectors, and your phone gets a notification when it's finished or needs your approval. If the job is something like fixing a build, it starts a Claude Code session for it, and that session shows up in your Code tab with a Dispatch label.

Find the proposal I sent last week in my Clients folder & draft a
follow-up email for me to check.

To set it up:

1. Update the Claude desktop app and the Claude app on your phone.
2. Open the Cowork tab and click Dispatch in the side panel.
3. Click Get started.
4. Turn on file access, and keeping your computer awake if you want it.
5. Click Finish setup, then message it from either device.

It needs a Pro or Max plan, and Anthropic says it isn't open to every new user yet. If you don't see Dispatch in the Cowork side panel, you don't have it yet.

The catch and the fix

Everything above runs on two things: your files, and your computer being on.

Claude Code is only as good as the folder you give it. Point it at an empty folder and it's back to being a clever chat that doesn't know you. The fix is your AI brain, your business written down in one place where Claude can read it. Here's how I set mine up in Obsidian.

Local routines and Dispatch need your computer awake with the Claude app open. If the laptop's shut, nothing runs. There's a setting for it: turn on Keep computer awake in the app's settings, and keep the lid open. For jobs that don't need your files, use a cloud routine, which runs even when your computer is off.

What this looks like as a business

I teach AI from the business outcome back, not from the tool. Claude Code is the tool. The outcome is that I've built my own AI COO and a team of AI employees that run the operations side of my business, and every one of them runs on the five things in this post: a folder, a rules file, skills, connectors and routines.

You don't need to be techy to use any of it. Tech isn't coding, it's logic. If you can write down how you want a job done, you can teach Claude to do it.

Start with one folder

Open the Claude app, click Code, start a new chat on Local, point it at the folder where your business lives and run the first prompt in step 2. Read what it says it knows about you. What's missing from that list is what goes into your AI brain next.

If you're ready to build your first AI employee on top of this, start with an AI COO.

If you haven't got your AI brain set up, that's the foundation for all of this: start here.