Workflow zur Humanisierung von KI-Texten mit Anti-AI-Stilrichtlinie

Step 1. Build your ban list Open a new Google doc. Name it anti-ai-writing-style .md. Put every single word you consider as AI. This 1 file replaces every "sound human" prompt. Step 2. Fill the file with the right rules Add banned words (delve, unlock, leverage). Add banned phrases (let's dive in...). Add banned patterns ("It's not X. It's Y" & more). Add pacing rules (no em dashes, vary line length). Step 3. Drop it into Cowork Open the Claude app, go to Cowork. Open your master folder on your computer. Inside it, find the ABOUT ME subfolder: ABOUT ME - who you are + how you write; TEMPLATES - your best work as structures; OUTPUTS - where Claude saves finished files. Upload anti-ai-style .md inside ABOUT ME. Step 4. Lock it into Global Instructions Go to Settings > Cowork > Global Instructions. Add these prompts: "Read every file in ABOUT ME before every task." "Audit drafts against anti-ai-writing-style .md." "Save deliverables in OUTPUTS, never delete without approval." Step 5. Write with one prompt (stick to it) Use: "I want to [TASK] for [SUCCESS CRITERIA]." After the draft: "Audit it against anti-ai .md." Repeat the audit until the AI patterns are gone.