FIELD GUIDE

How to Set Up Claude

A Death Doula's 30-Minute Setup

By River Braun  ·  The Timebender Collective  ·  skool.com/the-timebender-collective

Why this guide exists

Most people who try AI come away frustrated. They prompt. They get bland output. They prompt again. They quietly give up on the whole thing and tell themselves it's not for them.

It is for you. The problem isn't the tool — the problem is that nobody told you the real secret: setup matters more than prompting.

This guide walks you through the same setup I use every day in my own death doula work and the work I do for Christy. It's adapted from Ruben Hassid's framework (one of the most-read AI writers on Substack — link at the end), and tuned specifically for the kind of work we do — work that lives or dies on tone.

Thirty minutes. Three files. One global instruction. After that, every conversation you have with Claude already knows who you are, who you serve, and how you sound. You stop re-explaining. You start working.

The bottom line

• You'll create one Project in Claude — your dedicated workspace.
• Inside it, three short files: about-me, my-business, voice-and-style.
• You'll generate each file by letting Claude interview you (the prompts are below).
• You'll set one global instruction so Claude reads those files automatically.
• You'll run one test to confirm it worked.
• Then you'll never explain yourself to Claude again.

What you need before you start

• A Claude account — free works for testing, but Pro ($20/month) is what unlocks the model and the project space we'll use. Sign up at claude.com. Free is fine (for now).
• 90-120 uninterrupted minutes.
• A cup of tea or coffee.
• Permission to be honest with yourself in the interview prompts. The whole thing only works if you tell Claude the truth about how you actually work.

Claude Courses ➡ https://anthropic.skilljar.com/

Timebender Collective ➡️ https://www.skool.com/the-timebender-collective/about?ref=c5c387076c0d46b7aba33fe7dc468e24

Step 1 · Pick the right model

Open Claude. Look at the model dropdown at the top of the chat window.

• Select Claude Opus 4.8 (the most capable model).
• Turn on “High”. [Extended is not needed for this and will only run through your tokens faster and take up more time]
• Don't change these. Wrong model means generic output.

Why this matters: Claude has multiple models. The default isn't always the smartest. Opus is the one that handles nuance well — and nuance is the entire job in this work.

Step 2 · Create your Project

A Project in Claude is a workspace that holds context — files, instructions, and chats — all in one place. Think of it as a binder you hand a thoughtful assistant on day one.

• In the left sidebar, click "+ New project."
• Name it: Your Practice (or whatever you'd call your work).
• Add a short description:
• That's the workspace. We'll fill it in the next step.

Step 3 · Extract Yourself

These three files are what makes Claude sound like you instead of a wellness chatbot. You don't have to write them from scratch. Claude will interview you to create each one.

Copy the prompt into a fresh chat inside your Project. Answer Claude's questions honestly. When the interview is done, Claude will write your 3 files. Please save them and upload them back into the Project so every future chat can read them.

Prompt 1: Copy this prompt into a new Claude chat

<aside> 🔥

After Claude generates the .md files, download them and upload them into the Project's Files field, then run the follow-up prompt (ideally in a fresh chat in the same Project).

</aside>