A free guide by Mariah Brunner ↗ INSTAGRAM ↗ TIKTOK
Setup

Set Up Claude
In 5 Steps

The 10-minute setup that turns Claude from a chatbot into a thinking partner that remembers you, writes in your voice, and skips the AI fluff.

You're not bad at Claude. You just never set it up right.

Most people open Claude, type a question, get a generic response, and walk away thinking AI is overhyped. The problem is not Claude. The problem is you are starting every single conversation from scratch. Claude has no idea who you are, what you do, how you write, or how you want it to respond. So it gives you the average answer for the average person.

These 5 steps fix that in about 10 minutes. After this, every conversation you have inside Claude starts with it already knowing you. Your writing has your voice in it. The annoying AI fluff is gone. And the answers actually fit the way you work.

If you only have time for one, do Step 1. That alone unlocks 80 percent of the value.

The 5 Steps 10 Minutes Total

Step 1

Don't Start A Chat. Start A Project.

A Project is a workspace inside Claude where it remembers everything across every conversation. Set it up once, and every session from then on already knows the context for what you are working on. Without a Project, every new chat is a stranger meeting you for the first time.

How to do it: Open Claude, click "Projects" in the left sidebar, click "Create Project." Name it something simple like "Work," "Personal," or your business name. Everything that follows in this guide goes inside this Project.

Pro tip

Create separate Projects for the different parts of your life. One for work, one for personal, one for creative. Each Project gets its own context, so Claude is never mixing up your business emails with your meal plans.

Step 2

Tell Claude Who You Are

This is the step everyone skips. Then they wonder why Claude's answers feel slightly off. The fix is to give Claude a real briefing on you. Don't half-ass this. The more thorough you are here, the better every response becomes for the rest of your life.

How to do it: Open a new chat inside your Project. Click the microphone icon and just talk. Voice mode is dramatically faster than typing here and you'll naturally be more thorough. Walk through who you are, what you do, your goals, your background, how you like to communicate, and the things you don't want from Claude.

Or paste this prompt and fill it in

I'm setting you up to help me long-term, so I'm going to give you a real briefing on who I am. Save this and use it as context for every conversation we have in this Project. WHO I AM Name: [your name] Where I live and timezone: [city, timezone, when you actually work] Age and life stage: [optional but useful, e.g., 31, no kids, full-time founder] WHAT I DO My role or business: [job title, company, or the business you run] My main responsibilities or revenue streams: [the 3 to 5 things you actually spend time on each week] Industry or niche: [your industry in plain English] Who my customers or clients or audience are: [describe them like you'd describe them to a friend] WHAT I'M WORKING ON RIGHT NOW My biggest 1 to 3 goals over the next 6 months: [be specific, with numbers when possible] The single hardest problem I'm trying to solve: [the one that keeps you up at night] What I'm avoiding or procrastinating on: [be honest, this helps Claude push you] HOW I COMMUNICATE My communication style: [direct or warm, casual or professional, short or detailed] Words and phrases I use a lot: [list 5 to 10] Words and phrases I would never use: [the AI cliches that make you cringe] HOW I WANT YOU TO WORK WITH ME The default response length I want: [3 sentences, 1 paragraph, 5 bullets, etc.] The format I want for complex answers: [step-by-step, bullet points, prose, etc.] When I'm wrong, I want you to: [tell me directly, ask me questions, etc.] When you don't know the answer, I want you to: [say so, give your best guess, ask me, etc.] WHAT NOT TO DO Never do these things: [list everything that annoys you, e.g., start with "Great question," repeat my question back, add disclaimers, use corporate language, end with a summary, etc.]

Pro tip

If you don't know what to put in some of these sections, ask Claude to interview you. Say "Ask me 15 questions to fill out this About Me." It will pull more out of you than you'd write on your own.

Step 3

Have Claude Write Your Custom Instructions

Pasting your background once is good. Saving it as permanent Custom Instructions is the move. Custom Instructions tell Claude not just who you are, but how to behave with you by default in every single conversation inside this Project. The trick is to let Claude write them for you.

How to do it: In the same chat where you filled out Step 2, paste the prompt below. Take the output, and paste it into your Project's "Custom Instructions" tab (it's in the Project settings on the right side of the screen). That's now Claude's permanent operating mode for everything inside this Project.

Copy this prompt

Based on everything I just told you about myself, write me a clean set of Custom Instructions for this Claude Project. The instructions should: 1. Describe who I am and what I do in 2 to 3 sentences 2. Set my default communication style, tone, and format 3. List the things Claude should NEVER do when working with me 4. Define how to handle uncertainty or disagreement with me 5. Include any default behaviors I would want in every single session Write the instructions in second person, as if Claude is reading rules about how to help me. Be specific. No generic advice like "be helpful." Under 400 words total. When you're done, give me the instructions in a clean code block I can copy-paste into the Custom Instructions tab.

Pro tip

Custom Instructions are the single highest-leverage setup move in Claude. Once they're in place, every new chat in your Project starts with them loaded. You'll never have to re-explain yourself again.

Step 4

Drop In 5 Examples Of Your Writing

When Claude writes for you without examples, it writes in its own voice. That voice is grammatically correct and sounds nothing like you. The fix is to show Claude what you actually sound like. Three to five samples is enough for it to pattern-match your voice.

How to do it: Pull together 3 to 5 examples of writing you've actually done. Long emails, Slack messages, presentations, captions, documents. A mix is better than all one type. Paste them into a new chat inside your Project with the prompt below.

Copy this prompt

I'm going to give you 5 examples of my actual writing. Analyze them in detail. Then save my voice as a permanent rule for this Project. Here are my samples: [PASTE SAMPLE 1] [PASTE SAMPLE 2] [PASTE SAMPLE 3] [PASTE SAMPLE 4] [PASTE SAMPLE 5] Analyze my writing style in detail. Look at: 1. Sentence length and rhythm. Do I write short, punchy sentences or long flowing ones? 2. Vocabulary. What kind of words do I gravitate toward? What kind do I avoid? 3. How I open and close. What do I tend to lead with? What do I tend to land on? 4. Formality. Am I casual, formal, or somewhere in between? 5. Patterns. What sentence structures, transitions, or moves do I repeat across samples? 6. Voice tells. The things that make my writing sound like me and not generic AI. When you're done, give me a clean "Voice Rules" summary in about 200 words. Then save this voice as a permanent rule for this Project: whenever I ask you to write anything for me, match this exact style. Do not default to your own voice patterns.

Pro tip

After the analysis, ask Claude to write one test paragraph in your voice. If it doesn't sound like you, tell it what's off and give one more sample. Two rounds of feedback locks the voice in.

Step 5

Add Behavior Rules

This is the step that makes Claude actually pleasant to use. It is just as important to tell Claude what NOT to do as it is to tell it what to do. No disclaimers. No restating your question. No filler summaries at the end. This one step cuts a lot of the annoying parts of every AI response.

How to do it: Paste the prompt below into Claude. Take the output and add it to the bottom of your Custom Instructions from Step 3. You'll feel the difference in your very next response.

Copy this prompt

Write me a "Behavior Rules" block I can add to my Custom Instructions. The rules should tell Claude exactly what NOT to do when working with me. Include rules like: - Never start responses with "Great question," "Absolutely," "I'd be happy to," or any opener that restates what I just asked. - Never repeat my question back to me before answering. - Never end responses with a summary of what you just said unless I specifically ask for one. - Never add unnecessary disclaimers, hedging, or "I'm just an AI" notes unless legally or ethically required. - Never use corporate-speak words like "leverage," "synergy," "robust," "elevate," or "unlock" unless I use them first. - Never agree with me when I'm wrong just to be nice. If I'm wrong, say so directly. - Never pretend to be confident when you're not sure. If you don't know, say "I don't know" or ask me. Format as a clean numbered list with 8 to 12 rules total. Add anything else you think would make your responses feel less like AI and more like a sharp colleague.

Pro tip

The single biggest behavior change comes from killing the preamble. "Great question!" "Absolutely!" "I'd be happy to help with that!" Adding one rule to never start a response that way makes every conversation feel 10x more direct.

Once You're Set Up How To Actually Use It

Start a fresh chat every time you switch topics. Inside your Project, but a new chat. Your About Me, Custom Instructions, and Voice Rules all stay loaded. The conversation does not carry random baggage from earlier topics. Claude stays sharp and your answers stay clean.

Use voice mode for the messy first draft of anything. Brainstorming, processing a hard situation, talking through a decision. Typing is too slow for the way you actually think. Voice mode is the fastest way to get your real thoughts in front of Claude so it can help.

Stop using Claude like Google. Don't type "what is X." Type "I am dealing with X. Here's the context. Help me think through it." Claude is a thinking partner, not a search engine. The second you change how you prompt it, the answers change with you.

Run a 5-minute weekly reset. Once a week, open a new chat in your Project and paste: "Based on everything you know about me and what I've been working on, what should I do more of next week, less of, and what's the one thing that would unlock the rest?" Claude will surface patterns you won't see on your own.

If Claude still feels generic

Go back to Step 2 and add more detail. The depth of your setup directly determines the quality of every answer you'll ever get. Most people spend 5 minutes on the About Me and wonder why Claude feels mid. The people getting the most out of Claude spent 30 minutes on it.

The people getting the most out of Claude aren't smarter than you. They just took 10 minutes to set it up properly. Now go do it.

The Only AI Masterclass You Need

Build AI Systems That Run Your Work, Business, And Life

If this guide helped, but you’re looking to go deeper, I got you!! My 30-Day Challenge takes you from saving AI tips you never use to actually building with AI, step-by-step.

I show you exactly how I automated two e-commerce brands, my social media, and most of my personal life, then hand you the agents, workflows & systems to do the same. I’m teaching you every single thing I know with one lesson and one build a day.

Join the AI Masterclass →

© 2026 Mariah Brunner. All rights reserved.