LinkedIn Comment Agent
Turn keyword searches into consistent, high-quality LinkedIn engagement — AI-drafted likes and comments, human-paced, running locally in your own browser.
Key takeaways
- Three targeting modes: keyword search, your own LinkedIn feed (by position), or a single post URL you paste in.
- Likes each match, then drafts an AI reply shaped around your product, audience, and writing style — not generic filler.
- Two run modes: Once (one like + comment, then stop) or All day (keeps engaging with safe spacing until the daily target is met).
- Editable daily ranges — 8–14 comments and 16–20 likes by default — with likes always kept above comments by a randomized gap.
- Four pacing profiles plus automatic smart breaks: 12–25 min gaps at the safest, 5–18 min steady, or clustered bursts, with a 15–30 min pause every few actions.
- Runs in its own Chrome profile on your machine. Live friendly log, session counters, and a persistent History tab record every action.
Showing up in the right LinkedIn conversations, every day, is what actually moves the needle — and it's the first thing that falls off your to-do list when the week gets busy. The LinkedIn Comment Agent handles the repetitive parts: finding posts worth replying to, liking them, and drafting a comment that reads like you wrote it between meetings.
It's not a scheduler that dumps the same line on fifty posts. Each run reads the post it's about to reply to, writes something specific to it, spaces itself out like a person browsing on and off through the day, and stops when it hits the daily number you set.
How it works
- 1
Pick how it should run
Once performs a single like + comment and stops — good for testing tone. All day keeps engaging through the day with randomized spacing until it reaches the daily target it rolled for today.
- 2
Point it at the right posts
By keyword searches LinkedIn using the keyword list from your Setup tab (one phrase per line; built-in defaults apply if you leave it empty). My feed engages a post by position — 0 is the top post, 1 the next one down. Specific URL takes any linkedin.com/posts/… link.
- 3
Like, then draft the reply
The agent likes the post first, then sends the post text to the AI along with your product details, comment guidance, and writing style. On a single run you can skip AI entirely and type the exact comment yourself.
- 4
Post at a human pace
In All-day mode each action waits out a randomized gap, and every few actions the agent takes a 15–30 minute smart break. Nothing fires on a fixed interval.
- 5
Watch it in the log
The activity panel narrates each step in plain language — which post it opened, what it liked, how long it's waiting next. Session counters tally comments and likes, and the History tab keeps a record across restarts.
The three targeting modes
| Mode | How it finds posts | AI replies | Best for |
|---|---|---|---|
| By keyword | Searches LinkedIn for recent posts matching your keyword list from Setup | Yes | Daily automation — the mode All-day runs use |
| My feed | Takes the post at a position you choose in your own feed (0 = top) | No — you type the comment | Engaging your existing network, one post at a time |
| Specific URL | The exact linkedin.com/posts/… link you paste | Yes | A post you already know matters — a prospect, a partner, a big thread |
Keyword targeting is where the quality lives. Broad terms put you in threads your buyers never read; a tight list of five to ten phrases that mirror how your customers describe their problems will fill the queue with posts you can actually say something useful on.
How the AI writes the comment
Three inputs shape every reply: your product details (name, tagline, description, demo link) from Settings, your shared writing style — which you can generate from a sample of your own writing — and optional comment-specific guidance for this agent only.
The default prompt enforces a structure that keeps comments from reading like sales bait:
- 1React to a specific detail from their post first — not "Great post!" — so it's clear the post was actually read.
- 2One natural sentence on what you're building and how it relates to that point. A builder sharing context, not a pitch.
- 3An optional soft connect ask, only when it fits. Never a demo, walkthrough, call, or booking link.
- 4Two to three sentences total, no lists, no hashtags, no URLs, casual peer-to-peer tone.
The prompt also guards against the classic AI-comment failure: it explicitly separates the post author from you, the commenter, so the agent never replies as if it wrote the post or drags in your job title where it doesn't belong.
If you want a different shape entirely, the Advanced tab exposes the full system prompt with variables for your soft connect ask, product name, and extra guidance — edit it, or reset to the default at any time.
Daily limits you control
Limits are ranges, not fixed numbers. Each morning the agent rolls a target inside each range, so its footprint differs day to day instead of hitting exactly the same count every time.
| Setting | Default range | What it does |
|---|---|---|
| Comments per day | 8–14 | How many AI replies the agent may post before it stops for the day |
| Likes per day | 16–20 | Likes are the cheap, low-risk signal — kept deliberately higher than comments |
| Likes over comments gap | 4–8 | Minimum extra likes above the comment count, randomized daily |
These defaults sit comfortably under the hard cap on your license plan — 15 comments and 30 likes a day on Starter, 30 and 60 on Pro, 60 and 120 on Agency. You can raise the ranges toward your plan cap, but the app recommends keeping the tested defaults unless you have a specific reason.
Pacing: why the gaps are irregular
Predictable timing is what makes automation obvious. The Speed setting picks how the agent spaces actions, and the default lets it vary automatically based on time of day.
| Speed | Gap between actions | Notes |
|---|---|---|
| Recommended (natural pace) | Varies per action | Picks a profile per action, weighted by time of day — steadier off-hours, more clustered during active hours |
| Slowest — safest | 12–25 minutes | Lowest footprint per hour |
| Steady — even spacing | 5–18 minutes | Predictable throughput, still randomized |
| Fastest — bursts | 2–5 min, then 15–25 min | Clusters a few actions, then goes quiet — how people actually browse |
On top of whichever profile is active, a smart break kicks in every few actions: a 15–30 minute pause that can't be turned off. The first action of a day also waits a short random delay, so a run never starts on the exact second you press the button.
Review before you trust it
- Preview only — don't post: on a single run, the agent types the AI comment into the box and waits for you to submit it yourself. The fastest way to sanity-check tone.
- Type your own comment: turn AI off on a single run and post an exact line you wrote, in the agent's pacing and targeting flow.
- Run once, repeatedly: fire a handful of single runs on keyword targets before enabling All day.
- Read the friendly log: every step is narrated in plain language, including which post was opened and what got posted.
Running alongside the other agents
The Comment Agent has its own Chrome profile, separate from Campaigns and the Post Generator. That means public-feed engagement and per-person outreach can run at the same time without fighting over the same browser window. Run All starts every connected agent's daily mode together in one click; each keeps its own limits and pacing. Campaigns are started from the Campaigns screen.
One constraint worth knowing: don't start a second LinkedIn automation inside the same profile while a run is active — stop the current one first. The app warns you when that would happen.
Who it's for
- Founders and solo marketers who need to stay visible daily without hiring a VA to do it.
- B2B sales teams engaging in the threads where their buyers already ask questions.
- Agencies running the same motion across client accounts, with Agency-plan caps and up to five device licenses.
- Anyone who's tried commenting manually, seen it work, and then watched it slide off the calendar after two weeks.
Setup checklist
- 1Connect LinkedIn once through the desktop app — you log in yourself, and the session stays in a local Chrome profile.
- 2Fill in product name, website, tagline, and description in Settings; these feed every AI reply.
- 3Paste a few paragraphs of your own writing into Settings → Writing style and let the app build a style card.
- 4Add 5–10 keyword phrases in Setup → Targeting, one per line.
- 5Run once in preview mode and read the comment it drafts. Adjust guidance until the tone is right.
- 6Check Setup → Limits, then start All day.
Frequently asked questions
It runs locally in your own Chrome profile with your own login, spaces actions with randomized pacing (12–25 minutes apart on the safest profile) and forced 15–30 minute smart breaks, and stops at the daily comment and like ranges you set under your plan cap. Optional preview mode lets you check a comment without posting. No tool can guarantee zero risk, but local execution and conservative pacing are designed to reduce it substantially compared with cloud agents on shared IPs.
Yes. On a single run, enable "Preview only — don't post" in Advanced options and the agent types the AI comment into the comment box without submitting, so you can read it and decide. Preview is a single-run option; All-day mode posts under your daily caps once you're comfortable with the output.
Three ways: keyword search using the phrase list in Setup → Targeting, your own feed by post position (0 = the top post), or a specific post URL you paste. Keyword search is what All-day mode uses. AI replies work on keyword and URL targets; feed mode posts a comment you type yourself.
You set a range in Setup → Limits — the default is 8–14 comments and 16–20 likes per day, and the agent picks a random target inside those ranges each day. Your license plan sets the hard ceiling: 15 comments + 30 likes on Starter, 30 + 60 on Pro, 60 + 120 on Agency.
Likes are a low-risk signal and comments are the expensive one, so the agent deliberately keeps likes above comments by a randomized gap (4–8 extra by default). An account that comments on everything it touches looks nothing like a real person browsing.
Three layers. Settings → Writing style sets your shared voice across every agent (and can be generated from a sample of your own writing). Setup → Style adds comment-only guidance. Setup → Advanced exposes the full system prompt with variables for your product name, soft connect ask, and guidance — editable, with a reset-to-default button.
By default, yes — one natural sentence after reacting to their post, framed as a builder sharing context rather than a pitch. The prompt explicitly forbids demo asks, walkthrough offers, booking links, and URLs. If you'd rather it never mentions your product, say so in comment guidance or edit the prompt.
Platform AI, running on the monthly AI credits included in your plan (3,000 on Starter, 8,000 on Pro, 25,000 on Agency). There's no separate API key to buy or configure.
Yes — "Run without showing the browser" moves Chrome off-screen rather than using a detectable headless mode. It's handy once you trust the output, but harder to debug when something breaks, so leave it off while you're dialing things in.
Yes. The Comment Agent, Campaigns, and Post Generator each use their own Chrome profile, so they don't collide. Start campaigns from the Campaigns screen; Run All starts the other connected agents' daily modes. What you shouldn't do is start two automations inside the same profile at once — stop the running one first.
Nothing already posted is affected, and the day's progress is saved — restarting won't double-count actions against today's target. The History tab keeps the record across restarts.