Build an AI system that actually knows you. A personalized assistant that understands your work, your clients, and your values.
For coaches, consultants, entrepreneurs and leaders ready to build AI that actually understands their work.
This guide is different. We believe AI should serve your growth as a whole person, not turn you into a more efficient machine. Before diving into Claude Code, let's be clear about what we stand for.
Service Designer turned transformational coach. I'm the founder of The Four Languages and PurposeDriven.AI.
I founded The Four Languages in 2023 to help people grow through all dimensions of life: mind, heart, body, and soul. Since 2018, I've been helping knowledge workers make the most out of their life and work through a mix of tech, real-world solutions, and human coaching.
Through PurposeDriven.AI, I help coaches, consultants, and service-based entrepreneurs build AI systems that actually understand their work. Not generic chatbots, but personalized systems that know your clients, your process, and your values.
Why I created this guide: Non-technical people shouldn't be left behind in the AI revolution. My goal is to help intentional people harness the full power of personalized AI without having to become a developer. If you can describe what you want, Claude Code can help you build it.
Connect on LinkedInMost AI guides focus on "getting more done." We focus on growing across all four languages of life: mind, heart, body, and soul. Efficiency is a byproduct, not the goal. If your AI system makes you more productive but less alive, you've lost.
We don't believe in generic AI setups. Your CLAUDE.md file, your skills, your MCPs should reflect YOUR life, YOUR values, YOUR patterns. The AI learns about you, not the other way around.
Not everything should be automated. Ask yourself: Does this automation serve my growth? Or does it bypass something I need to feel, process, or do myself? Some friction is healthy. Some tasks are medicine.
The best system isn't the most complex one. It's the one you actually use. Start with one skill, one workflow. Let your system grow as YOU grow. Your results will change. That's not a bug, it's the point.
Tech should help you close the laptop faster, not keep you glued to it. Every tool, every automation, every system should serve one purpose: giving you more presence, more energy, and more time for what actually matters in the real world.
This guide is for you if:
This guide is NOT for you if you just want to hack together the fastest automation possible without thinking about whether it actually serves you.
You probably already have a system. Maybe it's Notion. Maybe it's Google Drive. Maybe it's scattered across five different apps. Let's start there.
They exist. But they're scattered across different tools, different logins, different interfaces.
┌─────────────────────────────────────────────────────────────────┐ │ THE TYPICAL COACH/CONSULTANT TECH STACK │ └─────────────────────────────────────────────────────────────────┘ CONTENT SCHEDULING EMAIL ───────── ────────── ───── Canva Google Calendar Gmail Notion Calendly ConvertKit Later / Buffer Acuity Mailchimp FILES CRM WEBSITE ───── ─── ─────── Google Drive Dubsado Squarespace Dropbox HoneyBook WordPress iCloud Spreadsheets Kajabi NOTES PAYMENTS COMMUNICATION ───── ──────── ───────────── Apple Notes Stripe Slack Evernote PayPal Zoom Notion (again) Thrivecart Voxer ⚠️ 12+ apps • 12+ logins • 12+ interfaces • Nothing talks to anything
Every time you need something, you have to remember which app it's in, log in, navigate to it, then manually copy information between them. Sound familiar?
Yes, to a degree. Notion has AI. Canva has AI. Most tools are adding AI features.
But here's the catch: You're always tied to that platform's decisions. If Canva changes their AI logic, your workflows break. If Notion redesigns their interface, you have to re-learn everything. Any pricing change, UI overhaul, or feature removal leaves you at the mercy of someone else's product roadmap.
With Claude Code and terminal, you learn one way of working and it stays consistent. The interface doesn't suddenly change because a company decided to "improve" it. You swap out the underlying tools as technology advances, but your workflows, your muscle memory, your way of operating stays the same. You own your system instead of renting someone else's.
Imagine you had one single interface where you could talk to ALL of it. Not clicking through menus. Not remembering which app has what. Just... telling it what you need in plain language:
It searches your files, finds it, and shows you.
It creates the event. Done.
It reads your brand docs, creates the content, exports the images.
It queries your database and tells you who needs attention.
That's what an AI Life System is. One interface that can access, search, edit, and create across everything you use:
┌─────────────────────────────────────────────────────────────────┐ │ YOUR AI LIFE SYSTEM │ └─────────────────────────────────────────────────────────────────┘ ┌─────────────┐ YOU ─────────▶ │ TERMINAL │ (text, voice, images) └──────┬──────┘ │ ┌──────▼──────┐ │ CLAUDE CODE │ (the translator) └──────┬──────┘ │ ┌──────▼──────┐ │ CLAUDE.md │ (your map) └──────┬──────┘ │ ────────────────────────┼──────────────────────── │ │ │ │ │ │ ▼ ▼ ▼ ▼ ▼ ▼ 📁 Files 📅 Calendar ✉️ Email 🎨 Content 👥 CRM 🌐 Web Your Google Gmail Create Supabase Scrape folders Calendar Newsletters carousels Notion research Notes Scheduling Outreach PDFs Airtable APIs Docs Website ✓ One interface • Natural language • Everything connected
The CLAUDE.md file is your map. It tells Claude Code where everything lives, how you work, and what you care about. Claude reads this file every time you start a session, so it always knows your context.
ChatGPT is a conversation in a box. You copy-paste things in, it responds, you copy-paste things out. An AI Life System is connected to your actual life. It sees your files. It can edit them. It can talk to your other tools. It's not separate from your work. It's woven into it.
Terminal is the window where you communicate with your system. Think of it like a phone line. It's not the person you're talking to. It's the medium that carries your words. You type (or speak, using tools like Wispr Flow). You can even drag in images or screenshots. Terminal just delivers your message.
Claude Code is the intelligence on the other end. It takes your human language and translates it into action. Need a file edited? It translates that into the right commands. Need a PDF created? It figures out how. Need to query your database? It writes the code. You speak human. It handles everything else.
The foundation of your AI system is a folder on your computer that holds YOUR life, YOUR work, YOUR thinking. This is private. It lives on your machine, not someone else's server.
Here's what this actually looks like when you map your WHOLE life into it. Hover or tap each card to see what apps it replaces:
✓ MCP = Has a ready-made connector. API = Needs custom setup. manual = Copy/paste or file export.
This isn't about replacing every app. It's about having ONE place that knows about all of them, can pull from them, and helps you think across your whole life instead of in scattered pieces.
You need a few things before you can use Claude Code. Don't worry, this is a one-time setup and most of it is free.
Claude Code requires a paid Anthropic subscription. Claude Pro is $20/month and works great for most people. Claude Max ($100/month) is for heavy daily use. You can't use the free tier.
You'll want an app to see and organize your folder system. Here's how to choose:
Our recommendation: Start with VS Code. It's free, well-supported, and works for everything.
Node.js is a tool that lets your computer run JavaScript, which is what Claude Code is built with. You don't need to understand it or use it directly. You just need it installed so the install command works.
Mac: You already have it. Press Cmd+Space, type "Terminal", and hit Enter.
Windows: Use PowerShell (search for it in Start menu) or install "Windows Terminal" from the Microsoft Store for a nicer experience.
Once Node.js is installed, open Terminal and paste this command:
npm install -g @anthropic-ai/claude-code
This downloads and installs Claude Code. You only need to do this once. The "npm" command comes with Node.js.
CLAUDE.md is the most important file in your Claude Code setup. It's your map. Every time Claude Code starts, it reads this file first to understand who you are and where everything lives.
Your CLAUDE.md should point to where things are, not contain every detail. It's like a table of contents for your system. The names of your folders, the locations of your files, what each section is for. Claude uses this map to navigate your system and find what it needs.
From a real session helping people set up Claude Code:
"CLAUDE.md is what covers the basics. It's something that the AI will look at every single time it does anything. It will see this. So this should be like the map that it has. It doesn't need to have the details on what everything is, but it needs to have the names of everything that you need to find."
In other words: point to the destinations, don't write the full directions. Your brand guidelines can be in a separate file. Your client list can be in a separate file. But CLAUDE.md knows they exist and where they are.
How do you edit it? CLAUDE.md is just a text file. You can edit it directly in VS Code, Obsidian, or any text editor. Just navigate to your project folder and open it. Or you can ask Claude Code to edit it for you through the Terminal. Both methods work equally well. Remember from earlier: Terminal is the channel, your text editor is another channel. They both connect to the same file.
Here's an example of a CLAUDE.md for someone managing their personal growth, work, and wellbeing:
# My Life System - Claude Instructions ## Who I Am I'm a coach building my AI Life System. I care about whole-person growth: mind, heart, body, and soul. I don't want to just be productive, I want to be alive. ## How I Work - Energy aware: I track my energy levels. Don't suggest heavy work when I'm low. - Voice-first: I often speak instead of type. Keep responses conversational. - No overwhelm: Give me 2-3 options, not 10. I can ask for more. ## Morning Check-In Protocol When I say "good morning" or start a new day, ask me: 1. How's your energy today? (1-10) 2. What's on your mind? 3. Any intentions for today? ## Current Projects - Main focus: Building my coaching practice - Side project: Writing a book - Learning: React Native app development ## Files & Folders - /projects/ - Active work projects - /journal/ - Daily reflections - /learning/ - Notes and courses - /planning/ - Weekly and monthly plans ## Important Notes - Never schedule calls before 10am, I'm not a morning person - Ask before making any commits to git - When I'm stressed, remind me to step away from the screen
A brief description of yourself and what matters to you. This helps Claude understand your values and priorities.
Your preferences, constraints, and working style. Do you like detailed explanations or brief answers? Morning person or night owl?
What are you working on right now? What phase of a project are you in? This prevents Claude from suggesting irrelevant things.
Help Claude understand where things live in your system so it can find and reference the right files.
Any specific instructions or constraints. "Never schedule calls before 10am" or "Always ask before making commits to git."
MCP stands for "Model Context Protocol." In plain English: MCPs let Claude Code talk to other apps and services, like your calendar, databases, or web services.
Without MCPs, Claude Code can only read your local files and run terminal commands. With MCPs, it can also check your calendar, add events, query databases, fetch web pages, and more.
| MCP | What It Does |
|---|---|
| Google Calendar | Read and create calendar events |
| Supabase | Query and modify your database |
| GitHub | Work with repos, issues, and PRs |
| Firecrawl | Scrape and read web pages |
| Slack | Send messages and read channels |
Remember our values: automate the right things. Before adding an MCP, ask yourself: Does this serve my growth? Or am I just automating for the sake of it? Start without MCPs and add them when you have a real need.
# List available MCPs claude mcp list # Add an MCP (example: filesystem) claude mcp add filesystem # Check your installed MCPs claude mcp list --installed
MCPs can be tricky to set up, especially on Windows. If you run into issues: check that Node.js is updated, restart your terminal after installing, and check the official MCP documentation for your specific tool.
See how to set up an MCP (Firecrawl) from start to finish:
Claude Code is surprisingly good at understanding what you want. But there are patterns that work better than others.
Less effective: "Help me with my project"
More effective: "Read my project folder and summarize what the main files do"
Less effective: "Fix the bug"
More effective: "There's an error in checkout.js when users try to submit. The error message says 'undefined is not a function'. Can you find and fix it?"
Less effective: "What should I do?"
More effective: "Give me 2-3 options for how I could approach this, with pros and cons"
Paste screenshots directly into your prompt. Claude can read and understand images - error messages, UI problems, design mockups. One screenshot often explains what 10 sentences couldn't.
It's easy to get distracted by advanced features before mastering the basics. Agent teams, custom MCPs, multi-file refactoring - these are powerful but not where you start. Get comfortable with simple file reading, writing, and conversations first.
When Claude edits a file, it shows you exactly what's changing using color coding:
Editing: /Users/you/project/config.js - const API_KEY = "old-api-key-here"; + const API_KEY = "your-new-api-key"; - const DEBUG = true; + const DEBUG = false;
Red lines with minus (−) show what's being removed.
Green lines with plus (+) show what's being added.
For bigger changes, Claude will ask before proceeding:
I'm about to make changes to 3 files: • src/components/Header.js • src/components/Footer.js • src/styles/main.css Allow this action? 1. Yes 2. Yes, and don't ask again for this tool 3. No
You stay in control:
1 = Approve this one action
2 = Trust this action type for the session (saves time)
3 = Stop and discuss before proceeding
Let's put everything together with a real example. You'll install an MCP, scrape your website, and create your first CLAUDE.md. All in 10 minutes.
Open your terminal and type:
claude
If you see Claude respond, you're ready. If not, go back to the setup page.
If you followed along on the MCPs page (Step 5) and already have Firecrawl installed, skip to Step 3!
Firecrawl is a website scraping tool. It lets Claude read any website. We'll use it to scrape YOUR website and create your CLAUDE.md from real information about you.
# In your Claude Code session, just say:
"Install the Firecrawl MCP for me"
Claude will walk you through the installation. You'll need a free Firecrawl API key (it takes 30 seconds to get one).
This gives Claude real information about YOUR business to work with. Instead of you typing everything, Claude reads your website and understands who you are, what you do, and how you work.
Now the magic. Ask Claude to create your system using your real website content:
"Use Firecrawl to scrape [yourwebsite.com] and create a CLAUDE.md file for me as the founder. I want to start a new project to organize my work."
Claude will:
Now your system exists with real context. Try these:
"Read my CLAUDE.md and tell me what you understand about my work"
"What services do I offer? Where do you see that information?"
"What else should I add to my CLAUDE.md to help you understand my workflow better?"
"Draft a short bio for me based on what you know from my website"
"What's my name and what do I do?" (It should know from your CLAUDE.md)
In 10 minutes, you:
Everything else builds on this. Skills, more MCPs, complex workflows. All of it starts with Claude Code running and a CLAUDE.md that knows you. You now have both.
You don't need to memorize many commands. Here are the ones you'll actually use.
# Navigate to your project folder first cd /path/to/your/folder # Then start Claude Code claude
Shows all available commands. Use this when you're stuck.
Clears the current conversation and starts fresh. Useful when context gets messy.
Summarizes the conversation so far to save context space. Use when you've been chatting for a while.
Shows how much of your API usage you've used in this session.
Switch between Claude models. Use opus for complex reasoning, sonnet for balanced tasks (default), or haiku for quick simple tasks. Example: /model opus
Claude Code can think more deeply about complex problems. Just type these keywords anywhere in your message:
Forces Claude to think deeply before responding. Use for complex architecture decisions, debugging tricky problems, or when you need thorough analysis. Just include "ULTRATHINK" anywhere in your prompt.
Lighter versions that encourage more thoughtful responses without the full deep thinking mode.
Once you create custom skills, you run them with a slash:
# Run your custom "morning" skill /morning # Run your "content-creation" skill /content-creation
Press Tab after typing the first few letters of a command to auto-complete. This works for file paths too.
Your AI Life System is designed to evolve. Once you've been using it for a while, here's how to expand its capabilities.
Your entire system lives in a folder. That folder can be version-controlled with Git and shared via GitHub. This means:
Every change to your system gets tracked. You can see how your CLAUDE.md evolved, roll back mistakes, and never lose your work.
Push your system to GitHub and others can clone it. Teams can collaborate on shared systems.
Your system becomes portable. Pull it down on a new machine and you're immediately productive.
Just ask Claude: "Help me set up Git and GitHub for my system folder." It will walk you through every step, from installation to your first commit.
"Help me set up Git for this folder. I want to back up my system to GitHub so I never lose it. Walk me through the whole process."
Claude will guide you through creating a GitHub account (if needed), initializing Git, and pushing your first backup. No commands to memorize.
You can run Claude Code in multiple Terminal windows at once. Each window is a separate conversation with separate context.
Keep one window for content creation, another for planning, another for a specific project. Each stays focused on its task without mixing contexts.
Working on something complex? Have one window research while another implements. Let them work in parallel and combine insights.
As you grow, you'll hear these terms. Here's what they mean and when to use each:
| Skill | Sub-Agent | Agent Team | |
|---|---|---|---|
| What it is | Saved instructions | Separate Claude instance | Multiple Claude instances |
| Context | Shares your conversation | Own isolated context | Each has own context |
| Visibility | See every step | Only see final result | See coordination + results |
| Best for | Repeatable tasks you want to watch | Heavy or parallel tasks | Complex multi-perspective work |
| Invoke with | /skill-name |
Task tool (auto) | "Create an agent team..." |
| Complexity | Beginner | Intermediate | Advanced |
Watch them work, understand the flow, refine your workflows.
When tasks get heavy or you need parallel execution.
When you have complex projects needing multiple perspectives.
Claude can only "remember" so much at once (the context window). Skills share your conversation's context. Sub-agents get their OWN fresh context, so they can go deep on one thing without cluttering your main conversation. Agent teams coordinate between separate contexts.
Once you're comfortable with the basics, Claude Code can coordinate multiple AI agents working together as a team. This is powerful for complex projects.
One session acts as team lead, coordinating work and assigning tasks. Teammates work independently in their own context windows and can communicate directly with each other.
Agent teams are an advanced feature. Master the basics first, then explore this when you have complex projects that would benefit from parallel AI work.
Using AI in your business isn't just technical. There are legal and ethical considerations.
Our team includes a lawyer (Josue) who can help you navigate compliance. This is part of making AI work for your whole business, not just your productivity.
You now know the fundamentals. The rest is practice and evolution. Remember: your AI Life System should grow with you. Start simple, add what you need, let go of what you don't.
Join our live session to see how an AI Life System actually works in practice:
Thursday, February 19, 2026
5:00 PM CET / 11:00 AM EST
See the folder organization, CLAUDE.md setup, AI-powered daily planning, content creation, project tracking, and backend integrations that make this system work. Watch it live, ask questions, and understand what's possible.
Register for the Live SessionYou don't have to figure it all out alone. Here's how to get support:
Or book a free discovery call to discuss what's right for you.