Founder of Erudience. Head of AI at Absolute Intelligence UK. Ships production n8n and voice AI systems for UK and international teams.
LinkedIn engagement automation avoids feeling automated by never pitching in the first comment, only reacting genuinely to the post itself, and introducing a soft, favour-framed mention of a product only if the person replies to that comment. A reply cap per person prevents repeated automated messages, and self-reply filtering prevents the system from replying to its own comments in a loop.
Building visibility on LinkedIn through genuine engagement, commenting on relevant posts, replying to responses, does not scale manually. Most attempts to automate it are either obviously robotic (a generic 'Great post!' on everything) or have no way to detect and respond to replies at all, which defeats the point of engaging in the first place.
The system below solves both problems with a two-stage design and a hard cap on how often it will follow up with any one person.
Stage one: comment like a person, not a salesperson
The system searches for recent LinkedIn posts, under 24 hours old, matching target keywords, then posts a genuine, specific reaction to the post plus a reaction or like, never mentioning any product at this stage. This is the entire point: the first touch has to look and read exactly like a real person engaging with content they found interesting, because it is functioning as exactly that.
A comment that references something specific in the post, rather than a generic compliment, is what separates this from the obviously automated version everyone has learned to ignore.
Detecting a reply without a webhook
LinkedIn offers no webhook for comment replies, so the system polls each tracked comment on a schedule, every 10 to 15 minutes, checking for new replies. This is a constraint of the platform, not a design choice, and it shapes the whole system: everything downstream assumes a polling delay rather than instant notification.
Stage two: only pitch after someone engages back
If someone replies to the original comment, a second stage AI reply engages further and, only at this stage, introduces a soft, favour-framed mention of the relevant product, never a hard pitch. The person has already shown interest by replying, which is what makes this the right moment, not the first comment where no signal exists yet.
This ordering, engage genuinely first, mention something only after reciprocal interest, is the actual mechanism that makes the automation feel like participation rather than prospecting.
Reply caps and self-reply filtering
Each person is capped at a maximum number of automated replies per thread, tracked via a unique constraint on person plus comment, so the automation never spams the same person regardless of how long a thread runs. This cap is a hard technical constraint, not a guideline the AI is asked to respect, because an AI asked nicely to stop eventually does not.
The system always filters out its own account's replies to prevent infinite reply loops, an easy failure mode to miss until the first time a comment thread between the bot and itself runs unsupervised overnight.
Turning engagement into pipeline
People who engage repeatedly are scored as warmer leads and flagged, including Slack alerts, for human follow up. The automation's job ends at surfacing a genuinely warm signal; closing is still a human conversation, which is exactly where the line should sit for anything past initial engagement.
- ·Never mention a product in the first comment; the first touch has to read as genuine engagement with no ulterior signal.
- ·A soft, favour-framed product mention only after the person replies is what keeps the automation feeling like participation, not prospecting.
- ·Polling on a 10 to 15 minute schedule is the workaround for LinkedIn having no webhook for comment replies.
- ·Per-person reply caps and self-reply filtering are hard technical constraints, not soft guidelines, and both are necessary to avoid spam behaviour and infinite loops.
- ·The automation's job is to surface a warm lead; closing still goes to a human.
Frequently asked
How do you avoid the comments looking obviously automated?+
By making the first-stage comment genuinely specific to the post's content and never including any product mention at that stage. The soft pitch only appears in a reply to a reply, at a point where a real signal of interest already exists.
Why poll instead of using a webhook for replies?+
LinkedIn does not expose a webhook for comment replies, so polling each tracked comment on a schedule is the only available mechanism to detect a reply at all.
What stops the automation from replying to itself in an endless loop?+
Self-reply filtering checks the author of each new reply against the automation's own account and discards matches before any reply logic runs.
How is a 'warm lead' identified from engagement alone?+
By repeated engagement from the same person across the two stages, tracked in the system and scored, then flagged to a human via Slack rather than auto-escalated into a sales pitch.
Does the reply cap apply per thread or per person overall?+
Per person per comment thread, tracked with a unique constraint, so a person cannot receive unlimited automated replies within a single conversation even if they keep responding.
Further reading and references
Related work on this site, and the tools and profiles referenced above.
Get new guides like this one
Whatever I ship next, straight to your inbox. No noise, unsubscribe any time.
