Last spring I replied to a snarky LinkedIn comment about our pricing with a line I genuinely, in the moment, thought was witty. It picked up 14 reactions — every single one from people who disagreed with us — and then the original commenter screenshotted my reply and dropped it into a Slack community I happened to already be a member of. Three people in that channel were prospective clients I'd spent months trying to get in front of. I sat there, in a room I'd worked hard to be invited into, watching my own brand get dunked on in real time by my own keyboard. I learned two permanent lessons that afternoon: everything is screenshottable, and I should not be allowed within reach of a reply box when I'm annoyed.
The reply itself was deleted within an hour. The screenshot was not. That is the asymmetry nobody warns you about — you control the post for sixty minutes and the internet controls the screenshot forever.
The prompt as an enforced pause
I now route every spiky comment through the Comment Triage — Three On-Brand Replies for Any Tricky Comment prompt before I type a single character on the actual platform. It is not that I can't write a reply — I write for a living. It's that the prompt makes me fill in a 'desired outcome' field, and the simple act of typing 'address the concern publicly and signal confidence without sounding defensive' physically drags my brain out of the gutter where the witty-revenge replies live. By the time I've described what I actually want the reply to accomplish, the cheap-shot version has completely lost its appeal. The pause is the product.
GPT-4o is the right model here for a specific, unglamorous reason: it's fast and cheap. On a day when a post is doing well and I've got forty comments stacking up, I am not going to sit through reasoning-model latency forty separate times. I need three solid, on-brand options, now, for almost no cost. This is a high-volume, voice-consistency task, not a deep-reasoning one, and matching the model to that reality is half of why it actually gets used instead of admired and abandoned.
Why three options instead of one perfect reply
Early on I wondered why the prompt insists on three replies every time instead of just handing me the 'best' one. The answer became obvious the first busy week: there is no single best reply, because the right length depends entirely on the room. A short acknowledgment that's perfect on a chaotic Tuesday would look dismissive on a thread where a thoughtful customer raised a real concern in front of a hundred quiet onlookers who never comment but absolutely judge.
- Option A (short) goes out when the comment is simple and I'm buried — and a genuine eight-word acknowledgment still beats a copy-paste 'thanks!' every single time.
- Option B handles roughly 80% of real comments — substantive, on-voice, but not a TED talk delivered inside a reply box.
- Option C is for the threads where everyone watching is quietly judging how you handle pressure. Those are the comments that build or quietly burn trust with a hundred people at once.
- The 'Send first' note is right often enough that I read it even when I've already chosen, because it occasionally catches a tone risk my own annoyance was hiding from me.
The benefit I didn't expect: speed on good days
I assumed this prompt would slow me down — one more step to protect me from my own worst instincts. In practice it sped up my best days. When a post is genuinely performing and the comments are pouring in, the blank reply box used to cost me a full minute of thought each, and forty comments became an afternoon I resented. Now I run the trickier ones in batches, the prompt hands me on-brand options, and I lightly edit and ship. Good comments get real, specific replies instead of forty identical 'love this!' responses that fool absolutely nobody and read like a bot left a tip jar out on the counter.
That hostile comment from last spring? If I'd had this prompt then, the 'de-escalate and offer DM' default would have spared me a year of that screenshot existing in a channel full of buyers. Grab the Comment Triage — Three On-Brand Replies for Any Tricky Comment prompt on Prompt Dock and run it on the next comment that makes your jaw tighten — your future self, scrolling past that screenshot a year from now, will thank you for the two-second pause. I pair it with my One Article, Seven Native Posts repurposing prompt, because a week of seven posts is a week of seven comment sections, and I refuse to walk into those unarmed ever again.