Train Claude on Your Brand
What You Get
A complete brand context file (brain file) that teaches Claude your voice, values, products, and constraints. Every piece of AI-generated content after this will sound like you, not a chatbot.
The Prompt
Below is my brand information. Create a comprehensive "brain file"
that teaches Claude exactly how to sound like me. Include:
BRAND INFORMATION:
[Paste your website copy, about page, social bios, mission statement,
product descriptions, email templates, social media posts—anything
that shows your voice and values]
BRAND VOICE SAMPLES:
[Paste 3-5 real examples of YOUR writing. These are reference points
for tone, vocabulary, and style.]
WHO YOU SELL TO:
[Describe your ideal customer in behavioral terms: what they do,
what they worry about, what they want]
WHAT YOU ACTUALLY SELL:
[Product/service name, what it does, why people buy it]
CONSTRAINTS & NON-NEGOTIABLES:
[Things you never do: avoid certain words, don't use passive voice,
always mention something, etc.]
OUTPUT: Create a structured "brand brain file" with these sections:
- WHO YOU ARE (mission, values, founder story)
- HOW YOU TALK (voice rules, words to use/avoid, tone by context)
- WHAT YOU SELL (products, pricing, positioning)
- WHO YOU'RE TALKING TO (audience, behaviors, what drives them)
- SYSTEM RULES (formatting, structure, don't-do's)
Make it specific and actionable. If I paste this file into a Claude
project as "System Prompt" or "Project Instructions," Claude should
immediately understand how to write like me.
How to Use It
Step 1: Gather your source material. Find your website copy, social media posts, email templates, about page, and sample writing. The more real examples you include, the better Claude learns your actual voice.
Step 2: Paste everything into Claude along with the prompt above. Don’t overthink it—just dump it all in. Claude will extract the patterns.
Step 3: Claude generates your brain file. Copy the output into a new document. Save it locally or in your Claude project settings.
Step 4: Use this file as project instructions for all future AI work. Every prompt you give Claude after that will reference your brand context. No more generic AI voice.
Step 5: Refine over time. After a few outputs, see where Claude misses. Add specific examples (“Don’t write like this, write like this”) to the file. Update it quarterly.
Example Output
# [Your Brand] — Brain File
## Who You Are
**Mission:** [One sentence describing why you exist]
**Values:** [3-5 core values with brief explanations]
**Founder Story:** [Your story in 2-3 sentences. Why you started this.]
**What Makes You Different:** [Your unfair advantage, perspective, or approach]
---
## How You Talk
**Voice:** Direct, specific, no corporate filler. You use contractions. You're friendly but not saccharine. You get straight to the point.
**Words to use:** Action-oriented language, concrete examples, plain English
**Words to avoid:** "delve," "elevate," "unlock," "reimagine," passive voice, buzzwords
**Tone by Context:**
- **Social/Threads:** Conversational, warm, slightly irreverent humor. Shows personality.
- **Email:** Direct, permission-granting. Lead with the benefit. Short paragraphs.
- **Long-form:** Clear and confident. Build credibility through specificity, not credentials.
- **Teaching:** Encouraging. Show immediate wins. Build confidence step-by-step.
**Writing Rules:**
- Short, punchy sentences mixed with longer ones for rhythm
- Second person for teaching ("You'll," "You can")
- Active voice. Always.
- Contractions always.
- Test: Does it sound like a smart friend texting at 10pm? Ship it.
---
## What You Sell
**Product 1:** [Name, what it does, who buys it, price]
**Product 2:** [Name, what it does, who buys it, price]
**Pricing Model:** [How you make money]
---
## Who You're Talking To
**Primary Audience:** [Who they are behaviorally—what they do, what they worry about, what they want]
**Secondary Audience:** [If applicable]
**What Drives Them:** [The emotional hook or outcome they're actually after]
---
## System Rules
**Formatting:** [Any specific preferences—no semicolons, always numbered lists, etc.]
**Structure:** [How you prefer content organized]
**Content Boundaries:** [Things you never cover, never recommend, never endorse]
**Audience Sensitivity:** [Any topics to handle carefully]
Tips
One: Your brain file isn’t static. Update it when your brand evolves. Add examples when Claude misses your voice. Review it twice a year.
Two: Be specific about what NOT to do. “Don’t use corporate jargon” is vague. “Never write ‘at the end of the day’ or ‘in a world where’” is actionable.
Three: Include real examples from your own writing—real tweets, real email subject lines, real website copy. Claude learns from patterns in actual work, not abstractions.
Four: Share your brain file with anyone who writes for your brand (team members, contractors, future collaborators). Everyone should sound like you.
Five: The more you refine this, the better. After your first 10 outputs with Claude, note where it went off-brand. Add those corrections to the file and watch Claude immediately adjust.
Want more methods? Browse the full library at hazelq.com/methods.
Built with Claude. Every prompt in this library has been tested in the latest Claude model.