Start here: what an AI employee is

This library gives you AI employees. One per post. This page explains what that means, what you need, and in which order to read the other guides.

What an AI employee is

An AI employee does one job for your business, the same way every time. It writes your hooks. It checks your posts. It drafts a caption in your voice.

Under the hood it is a set of plain text files that Claude Code reads. Claude Code is Anthropic’s tool that works on the files in a folder on your computer: it reads them, writes them and follows the instructions in them. You don’t need to code. You need to read and type.

Every skill in the library is one of those employees. You download it, put it in your folder, and it works.

The four layers

I build everything in four layers. Think of a small company.

Layer	In a company	In the library	What it is
1	The company	Founder AI OS	Everything together, plus one shared memory of your business
2	A department	An OS folder, like Marketing OS	A folder for one area of work
3	An employee	An agent	A role with its own instructions that runs one or more tasks
4	A task	A skill	One job, written down as a recipe Claude follows

A skill is the smallest piece. It is one file called SKILL.md with instructions for one job, for example “write five hooks for this post”. An agent is a role that uses several skills, for example a social lead that plans, writes and reviews a week of posts.

Most of the library is skills. That is on purpose. You start with one task, see it work, then add the next.

Why one prompt is not an employee

You have tried this. You paste a prompt into ChatGPT or Claude. The answer is fine. The next day you paste it again and get something different. It doesn’t know your offer or how you talk. And nobody checks the result.

An employee has three things a prompt doesn’t.

1. Knowledge of your business. What you sell, to whom, how you talk, what proof you have. Written down once, read every time.
2. Rules. What a good result looks like. For hooks: first the pain, then “but”, then the promise. For a caption: one call to action, no prices.
3. A review step. A second skill, or you, checks the work against the rules before it goes out. Nothing goes live without a human.

Take away any of the three and you are back to a prompt.

Common mistake: starting with ten skills at once. Each one then guesses your voice in its own way. Start with your business knowledge and one skill.

What you need

What	Why	Where
A paid Claude plan	Claude Code works with Pro, Max, Team or Enterprise. The free plan does not include it.	claude.com/pricing
Claude Code	The tool that reads and runs the employees	The next guide shows you how to install it
One folder	Your company lives here: your business knowledge and your employees	Anywhere on your computer

You don’t need a server, a developer or any code.

Check: you can also use a Claude Console account (pay per use through the API). For a first setup I use a normal subscription. It is simpler.

How to read these guides

Read them in this order. Each one ends with a link to the next.

1. Start here. You are reading it.
2. Install Claude Code. From zero to your first command, on Mac or Windows.
3. Write your business down once. The Founder Brain: one folder that every employee reads.
4. Install your first employee. The hook writer, run on your own text.

After guide 4 you have a working setup: one memory, one employee, one result you can judge.

The next four go deeper. The anatomy of a skill shows you how to write your own employee. The delta hook explains the technique behind the hook writer. Human in the loop shows how to make sure nothing goes live without you. From three skills to a department shows how employees work together.

Next

Install the tool that runs every employee: Install Claude Code.