Make Claude Code Write EXACTLY Like You (Free Templates)

Key Concepts

• AI Slop: Generic, bland, or robotic content generated by AI that lacks a unique human voice.
• Claude Projects/Skills: Features within Claude that allow for the creation of custom, reusable workflows and knowledge bases.
• Data Extraction: The process of scraping personal communication (emails, transcripts, social media) to train an AI model on a specific individual's writing style.
• Prompt Engineering: Using specific, structured instructions to automate the population of knowledge files.
• Skill/Slash Commands: On-demand workflows that trigger specific AI behaviors (e.g., /linkedin) without needing to re-input context every time.

---

1. The Problem: Generic AI Content

The video identifies a major issue with standard AI tools: they are trained on a massive, average dataset, resulting in "AI slop"—content that is technically correct but lacks personality, specific anecdotes, and a distinct human voice. The goal is to move from generic output to content that sounds exactly like the user by providing the AI with a deep, personalized knowledge base.

2. Methodology: Building a Personal Knowledge Base

To solve the "generic" problem, the creator outlines a framework for creating seven specific files that Claude uses as a reference for all future writing.

The Seven Core Files:

• Vocabulary: Words the user frequently uses (e.g., "sweet," "awesome") vs. words they never use (e.g., "henceforth," "moreover").
• Tone: The personality of the writing (e.g., warm, direct, conversational).
• Stories: Real-life anecdotes (e.g., running a wedding DJ business) that the AI can reference to build credibility.
• Humor: Specific styles of comedy (e.g., "dad joke" energy, self-deprecation).
• Business Context: Specific details about the user's professional life and offerings.
• Beliefs: The user's core opinions and professional stances.
• Analogies: Unique ways the user explains complex concepts.

3. Step-by-Step Implementation Process

1. Connect Data Sources: Within Claude’s settings, connect Gmail and Fireflies.ai (for meeting transcripts) to allow the AI to ingest real-world communication patterns.
2. Automated Population: Use a master prompt (provided in the video description) to instruct Claude to scan the last 100 emails and 20 meeting transcripts to populate the seven files automatically.
3. Manual Augmentation: Supplement the automated data by pasting in high-performing LinkedIn posts and specific passages that represent the user’s desired humor or style.
4. Filtering: Instruct the AI to filter out sensitive personal information or irrelevant data (e.g., empty emails).
5. Creating "Skills": Instead of manually referencing these files every time, create a "Skill" (an on-demand workflow). By defining a command like /linkedin, the user can trigger a prompt that automatically pulls from the seven files and follows a specific structural template.

4. Key Arguments and Evidence

• Authenticity through Data: The creator argues that AI can sound human only if it is fed "real" data. By using actual stats and stories from the user's life, the AI stops "making up BS" and starts producing content that builds trust.
• Efficiency: The creator notes that writing from scratch feels like "pulling teeth." Using this framework allows the AI to generate a high-quality first draft that requires minimal editing.
• Structural Consistency: The creator demonstrates that even with a personalized voice, the AI needs a structural template (e.g., "Hook -> Story -> Lesson -> Call to Action") to ensure the output matches the user's typical content format.

5. Notable Quotes

• "Claude is trained on millions of distinct voices and you get the sum of the average and it's just one bland generic voice that sounds like it's AI."
• "It's honestly, it feels like pulling teeth whenever I have to write anything... without Claude writing at least the first draft for me."
• "I feel like Claude has classified my humor and my personality better than anyone else could possibly tell me."

6. Synthesis and Conclusion

The transition from "generic AI slop" to a personalized AI assistant is achieved by shifting from broad, unguided prompts to a context-heavy, modular system. By treating Claude as a repository of one's own history, vocabulary, and beliefs, users can automate high-quality content creation across various platforms (LinkedIn, emails, proposals) while maintaining a consistent, authentic brand voice. The ultimate takeaway is that the quality of AI output is directly proportional to the quality and specificity of the personal data provided to it.