The AI-polished message that got pattern-matched

The AI-polished message that got pattern-matched
Popular advice: "Let AI rewrite your outreach so every line is sharp."
Myth. A team ran 2,000 DMs: half raw founder-written, half AI-polished for "clarity." Raw: 10.8% replies. Polished: 7.2%. Several creators replied to the polished batch with "ChatGPT-ass message, no thanks." The AI smoothed out the friction that signals a real person typed it — the typo, the abrupt ask, the specific aside. It's not eloquence that earns a reply — it's evidence of a human on the other end. Frictionless reads as automated now. Leave the rough edge in.