Skip to main content

How to Use AI Message Writing

Per-step AI personalization

You can have AI write a unique message for each individual lead, based on their profile and company research.

  1. Open the step you want in your campaign sequence.

  2. Turn on the AI Personalization toggle.

  3. Write an AI Prompt describing how the message should be personalized.

  4. Save.

The prompt field's guidance is "Describe how AI should personalize the message" — focus on the angle and the ask, not the raw data.

An example prompt:

"Open by referencing something specific from their role or recent activity. Explain in one sentence that we help B2B sales teams cut prospecting time in half. Ask if they're open to a quick chat. Keep it under 60 words and conversational."

AI inserts real names, not variables

This is important and catches people out: the AI writes the actual name and company directly into the message. You don't add {{firstName}} to an AI-personalized step — the generated text already contains "Hi Sarah" rather than a placeholder.

Mixing AI personalization with manual variables in the same step is unnecessary and can produce odd results.

The fallback template

Each AI step can have a Fallback Template — a complete alternative message used if AI generation fails for a lead.

Always set one. Without it, a lead whose generation fails may be skipped entirely. The fallback should be a solid generic version of the same message, using standard variables.

Generate a whole sequence with AI

To draft an entire sequence rather than one message, use the AI Sequence Generator. It asks for:

  • Company Name

  • Your Services

  • Target Audience

  • Case Studies

  • No. of Steps

Generation runs in the background and the sequence appears when ready.

Review everything before launching

AI drafts are a starting point:

  • Read every generated message and rewrite anything that doesn't sound like you.

  • Check any claims about your product.

  • Make sure the tone suits LinkedIn — conversational, not formal sales copy.

  • Confirm the length is right. LinkedIn messages should be short.

Getting better output

  • Be specific about the outcome you deliver. "We cut reporting time by 80% for B2B agencies" produces far better copy than "we do sales software."

  • Specify length in your prompt. Without it, AI writes too long for LinkedIn.

  • Include real results in Case Studies so the AI has something credible to work with.

  • Test on a small batch first and read what it produced before scaling up.

Did this answer your question?