Claude Code Tutorial: Build Your First Business Automation in 30 Minutes

If you’ve been searching for a Claude Code tutorial that focuses on running a business instead of building software, you’re in the right place. Most guides assume you’re a developer looking to write code. This one assumes you’re a founder who wants fewer repetitive tasks, fewer disconnected tools, and fewer mornings lost to administration.

If you run a service business, you’ve probably experimented with ChatGPT, Zapier, Make, or n8n. Each solved a small problem, but none became the system that tied everything together. Information stayed scattered, every conversation needed fresh context, and AI never really became part of how your business operated.

In this guide, you’ll install Claude Code, build your first workspace, and create a practical automation in around 30 minutes. It won’t transform your business overnight, but it will show you the foundation for building an AI system that remembers your business, understands your workflows, and starts taking real work off your plate.

No programming experience required—just a willingness to spend half an hour building something that compounds over time.

Key Takeaways

• Claude Code becomes significantly more powerful when it has persistent business context instead of starting every conversation from scratch.
• A well-written CLAUDE.md file acts as the foundation, teaching Claude about your business, clients, priorities, tools, and communication style.
• Setting up Claude Code requires only a few basic tools, including an Anthropic account, Node.js, a terminal, and a dedicated workspace folder.
• Storing structured client information allows Claude Code to generate personalised emails, reports, proposals, and other business documents with far greater accuracy.
• Reusable slash commands eliminate repetitive prompting by turning common workflows into one-command operations.
• The best first automations are repetitive, structured tasks such as follow-up emails, meeting summaries, proposals, and reporting—not complex strategic decisions.
• Claude Code is most effective when used as the foundation of a broader AI operating system built around context, data, intelligence, and automation.
• Small improvements compound over time. Every context file, reusable command, and automation increases the value of the entire system.

What Claude Code Actually Is (and Why It Matters for Business Owners)

Claude Code is Anthropic’s agentic coding environment. Forget the word “coding” for a second. What it really is: a command-line AI assistant that can read files, write files, run scripts, connect to your tools, and take actions on your behalf. It is Claude with hands.

Most people use Claude (or ChatGPT) like a vending machine. Type a question. Get an answer. Copy the answer somewhere. Move on. Every conversation is isolated. The AI has no memory of your business, your team, your clients, or your priorities.

Claude Code flips that. You build a workspace, a folder full of context files, data, and reusable commands. Claude Code reads the whole workspace every time it runs. So when you say “draft a follow-up email to this client,” it already knows who the client is, what you sold them, what your voice sounds like, and what stage the relationship is at.

That is the difference between a chatbot and an operating system for your business. One answers questions. The other does work.

If you want the fuller context on where this fits, I have written about the underlying system in the AI operating system for business and what Anthropic Claude Code actually is. For this post, we are building the first small piece of it.

Before You Start: What You Need

Five minutes of setup before we build anything.

1. An Anthropic account. Sign up at anthropic.com and add credit. You will use the API, which is pay-as-you-go. For this tutorial, expect to spend under $1.

2. Node.js installed. Grab it from nodejs.org. Install the LTS version. Click through the installer.

3. A terminal. On Mac, use Terminal. On Windows, use PowerShell. Both come pre-installed. You are not writing code. You are typing one or two commands.

4. A folder for your workspace. Somewhere on your computer, create a folder called business-workspace. This will be the brain of your business. Everything Claude Code needs lives here.

That is it. Total setup time: five minutes if your internet cooperates.

Step 1: Install Claude Code (2 minutes)

Open your terminal. Type this:

npm install -g @anthropic-ai/claude-code

Press Enter. Wait 30 seconds. Done.

Now navigate to your workspace folder:

cd path/to/business-workspace

Then start Claude Code:

claude

The first time you run it, it will ask you to authenticate. Paste in your API key from Anthropic. Claude Code is now running inside your workspace folder.

That is the whole installation. The complicated-sounding “agentic AI development environment” is a command-line tool that took two minutes to set up.

Step 2: Build Your Context File (10 minutes)

This is the most important step. Skip it, and Claude Code is just another chatbot.

Inside your workspace folder, create a file called CLAUDE.md. This is the file Claude Code reads first, every single time. It is the briefing document you would give a new executive hire on their first day, except Claude never forgets it.

Open CLAUDE.md in any text editor. Write a version of this, filled in with your actual business details:

# [Your Business Name] - Context File

## Who I Am
[Your name, role, location. One paragraph about your background.]

## The Business
- Name: [Business name]
- What we do: [One sentence. What you sell. Who you sell to.]
- Revenue: [Rough figure. Helps with financial questions.]
- Team: [Who works there. Roles. Contractors vs employees.]

## Our Clients
- Target client: [Who you serve best. Size, industry, situation.]
- Average deal size: [Rough figure.]
- Typical project length: [Weeks, months, ongoing.]

## Current Priorities
1. [Top priority this quarter]
2. [Second priority]
3. [Third priority]

## Tools We Use
- CRM: [What you use]
- Accounting: [What you use]
- Communication: [Slack, email, whatever]
- Other: [List the ones that matter]

## How I Write
- [Describe your voice. Direct? Warm? Technical? Casual?]
- [Any specific rules. NZ spelling. No exclamation marks. Whatever.]
- [Words you never use.]

## What I Don't Want to Do
- [Things you want Claude to flag or avoid]

Spend ten minutes on this. Not two. Not an hour. Ten minutes, writing like you are explaining your business to a smart friend.

When you are done, save the file. Close the editor.

Step 3: Your First Real Test

Go back to your terminal. Claude Code should still be running. If not, type claude again.

Now ask it something only someone who knows your business could answer:

“What should my top three priorities be this week, given where the business is at?”

Watch what happens. Claude Code reads your CLAUDE.md File understands the context and gives you a real answer. Not generic. Not “it depends on your specific situation.” An actual recommendation based on the business you just described.

This is the moment most people feel the shift. The AI is no longer a vending machine. It is a colleague who has done the reading.

If the answer is off, your context file needs more detail. Add it. Ask again. Iterate. Ten minutes of context writing equals every future conversation being 10x more useful.

Step 4: Build Your First Automation (15 minutes)

Now for the part that actually saves you time. We are going to build a simple automation: a client follow-up email generator that knows your voice, your business, and the client’s history.

In your workspace, create a new folder called clients. Inside it, create a file called sample-client.md. Fill it in:

# Sample Client Name

## Contact
- Company: [Company name]
- Contact person: [Name, role]
- First contact: [Date]

## What They Bought
- Service: [What you delivered]
- Price: [What they paid]
- Delivered: [Date]

## History
- [Key moments. What went well. What was tricky.]
- [Current status. Active. Paused. Renewal due.]

## Notes
- [Anything important about this relationship]

Now, back in your terminal, with Claude Code running, try this:

“Read the sample-client.md file. Draft a quarterly check-in email to this client in my voice. Include one specific observation from our work together and one soft question about their current priorities.”

Claude Code will read the client file, read your CLAUDE.md for voice and context, and produce a draft that sounds like you wrote it. Because it knows the relationship.

You can now build this pattern out:
– Add every client to the clients folder
– Ask Claude Code to draft follow-ups for any client by name
– Ask it to flag clients you have not contacted in 60 days
– Ask it to draft monthly reports, proposals, or check-ins

Step 5: Create Your First Reusable Command

Typing the same instructions every time gets old. Claude Code lets you create slash commands, reusable prompts you trigger with a single word.

In your workspace, create a folder called .claude. Inside it, create another folder called commands. Then create a file called followup.md. In that file, write:

Read the client file specified in the user's request.
Read CLAUDE.md for my voice and business context.
Draft a quarterly check-in email to this client.

Include:
- One specific observation from our work together
- One soft question about their current priorities
- A clear but non-pushy offer to catch up

Match my voice exactly. No corporate fluff.
Save the draft to outputs/drafts/ with the filename [client-name]-followup-[today's date].md

Save it. Now in Claude Code, you can type:

/followup sample-client

And it runs the whole sequence. You can build commands for anything you do repeatedly: drafting proposals, writing scopes, processing meeting notes, generating invoices. Each command is a script Claude Code runs on demand.

This is the foundation of everything. Context, plus data, plus reusable commands. That is Claude Code’s actual value for a business owner.

What You Just Built (and What Comes Next)

Stop and look at what you have done. In 30 minutes:

1. Installed Claude Code
2. Created a context file that teaches the AI your business
3. Stored structured client data that the AI can read
4. Built a reusable command that drafts client communications in your voice
5. Started separating work that used to require you from work that the system can handle

This is Layer 1 of an AI operating system: Context. It is the foundation. Before you automate anything meaningful, the AI needs to know who you are, what you sell, how you operate, and how you communicate.

Next steps from here are about building out the other layers:

• Data layer. Connect your numbers. Your CRM, accounting, and analytics are all feeding into the workspace so Claude Code can answer “how did we do last week?” with real data.
• Intelligence layer. Overnight sweeps that generate a morning brief. What happened yesterday? What needs attention today? Delivered to your phone before you are out of bed.
• Automation layer. List every recurring task in your business. Score each one. Start crossing them off, one at a time, permanently.

I have written about the full approach in AI operating system for business and the broader AI workflow automation context. The pattern is always the same: context first, then data, then intelligence, then automation. Skip any layer, and the rest collapses.

Common Mistakes to Avoid

A few things that trip people up in their first week.

Treating it like ChatGPT. Do not just ask questions. Build context files. Build commands. The whole point is reusable, compounding work, not one-off prompts.

Over-engineering the first build. You do not need 30 context files on day one. You need one CLAUDE.md, one client file, and one command. Get the loop working. Then expand.

Trying to automate judgment calls first. Claude Code is excellent at repetitive, structured work: drafting emails, processing data, generating reports. It is not yet excellent at high-stakes strategic decisions that require nuance. Start with the boring stuff. That is where the hours are anyway.

Not writing the context file in your voice. If your CLAUDE.md sounds corporate, everything Claude Code produces will sound corporate. Write it the way you actually talk. The AI mirrors what it reads.

Skipping version control. Once you have built something useful, put your workspace under Git. A simple git init in your workspace folder means you never lose work and can roll back if something breaks. Anthropic’s documentation on Claude Code’s best practices covers this in more detail.

Why This Matters Beyond One Tutorial

Most businesses don’t have a technology problem—they have a dependency problem. Too much knowledge lives inside the founder’s head, too many routine decisions require manual effort, and too much time disappears into work that could be handled by a well-designed system. More software rarely fixes that. It usually just creates more places to check.

This Claude Code tutorial is the starting point for building something different. With one context file, one reusable command, and one practical automation, you’ve taken the first step toward creating an AI system that remembers your business, understands how you work, and helps execute repetitive tasks without needing to be re-taught every day.

The next step is simply to keep building. Each improvement strengthens the foundation, each automation recovers more time, and each new capability makes the next one easier to create. Over time, those small changes compound into a business that relies less on constant founder involvement and more on systems that keep delivering value every day.

Where to Go From Here

If you want the big-picture map of how this scales into a full operating system, ai-operating-system-for-business walks through the five layers in detail. If you want the specific context around Claude Code itself, anthropic-claude-code has the fuller explanation of why Anthropic’s environment is different from everything else.

If you’d like to map this out for your specific business, book a 30-minute Discovery Call. I’ll walk you through what AI could realistically take off your plate, how to roll it out properly at your size, and whether there’s a fit. No pitch, no obligation.

For now, go build your first context file. Ten minutes. See what happens when the AI finally knows your business.

Frequently Asked Questions

About Octavius

Titus Mulquiney is the founder of Octavius AI, where he builds AI brains and AI workforces for founder-led businesses stuck running everything out of their own head. Twenty years in marketing, ex-Sony product manager, ex-GM Zeal NZ. Based in Auckland, working with operators across NZ, Australia, and the US. Connect on LinkedIn.