§ Playbook
How to use AI to draft LinkedIn comments that actually get replies
Generic AI automation is why LinkedIn comments read like a bot convention. Here's the human-in-the-loop workflow we use inside Spark to draft comments in 15 minutes a day that actually earn replies.
1. Start with a signal, not a prompt
The best comments react to something specific — a launch, a hot take, a hiring post. Feed the AI the post text, the author, and one line of context about your relationship. Skip the "write me a comment about sales" prompts; you'll get filler back.
2. Draft in your voice, not a template
Give the model 5–10 of your past comments as a style sample. This is the single biggest lever between "AI slop" and "sounds like a real person." Spark's style library does this automatically; if you're rolling your own, keep the samples fresh.
3. Keep a human in the loop
Auto-send is where trust dies. Every draft should go through a 5-second review — approve, tweak, or skip. This is the difference between showing up as a familiar name in someone's feed and getting quietly muted.
4. Measure replies, not volume
The point of commenting isn't to hit a daily quota — it's to become recognizable to a small list of people who matter. Track replies, profile views, and connection-accept rates from the accounts you're targeting. If those numbers don't move, your drafts are still too generic.
5. 15 honest minutes beats 500 auto-sends
The best-performing Spark accounts leave 6–10 thoughtful comments a day, mostly on Sales Navigator prospects. That's the whole habit. No spray, no auto-DM, no volume game.
Want the same workflow, without wiring it up?
Spark watches your prospects, drafts comments in your voice, and hands you a review queue. You stay human. The AI just does the typing you'd skip.