Table of contents3
Prospects can smell an LLM opener from three inboxes away. The fix isn't to write less with AI — it's to ground every generation in signals the model couldn't have invented on its own.
Why AI openers get flagged
Default LLM output leans on adjectives ("impressive growth", "exciting journey") and vague compliments. Humans don't write like that at 8am on a Tuesday.
Six grounding signals to feed the model
- Recent job change or promotion
- A specific podcast, panel, or LinkedIn post from the last 60 days
- Funding round, layoff, or leadership change at the company
- Tech stack signal (new install / new integration)
- Hiring signal for a role adjacent to your buyer
- Product launch, pricing change, or press mention
Prompting for restraint
Cap outputs at 25 words. Ban adjectives. Require a specific noun (person, product, or date) in the first sentence. This alone lifts reply rates by 30–40% in our tests.
The goal isn't to hide AI. It's to give AI enough real context that its output is indistinguishable from a well-researched human note.
Tags:PersonalizationCopy
Found this useful? Share it.