An alert that only fires when it should.
It classifies every mention, then branches — positive and neutral are logged quietly, only the negative reach you.
6 steps · runs on demand · added to your workspace as a draft
The branch
Six mentions. One alert.
An alert that fires on everything is one you learn to ignore. The branch decides who hears about what.
01 · Six mentions of you turned up today.
02 · Each is read in full and classified, not skimmed for keywords.
03 · Positive and neutral are logged. Nobody is pinged.
04 · One is negative. It takes the other branch.
05 · Six mentions, one alert — and it says what the complaint is.
Web search · your brand name6 mentions
forum.examplepositive
newsletter.exampleneutral
blog.examplepositive
community.exampleneutral
review-site.examplenegative
podcast.examplepositive
Logged quietlyno alert
- forum.examplepositive
- newsletter.exampleneutral
- blog.examplepositive
- community.exampleneutral
- podcast.examplepositive
Alert sent1 of 6
review-site.example
“Support took four days to answer us.”
The complaint: slow first response from support. Not pricing, not the product.
What it does
Pre-built, then yours to customize.
Add it to your workspace in one click. Sensible defaults and example inputs are already filled in.
Reads from · writes to
- Web search
The template, step for step
- Trigger
- Web search
- LLM call
- Branch (if/else)
- Notify
- Notify
Added to your workspace as a Draft, with example inputs filled in. Every step is yours to change.
- 01
Only fires on negative — no alert fatigue
- 02
Sentiment read from the full page, not a snippet
- 03
Tells you what the complaint is, not just that one exists
- 04
Positive and neutral still logged, quietly
In, and out
You give it your brand name. This is what comes back.
- Shape
- 6 steps
- On demand
- run mode
- 1
- click to add
Illustrative output.
The steps
What happens on every run.
- find_recent_mentions
- fetch_mention_pages
- classify_sentiment
- branch_on_negative
- summarize_the_issue
- notify_and_log
6 steps · Runs on demand0 / 6
Run history keeps each step's output and runtime, and the input of every step that completed.
Before & after
The time this gives back.
The templates that publish put a Human review step before the CMS step, and nothing ever posts to Reddit for you. Review is a step, not a lock: a workflow you build or edit can publish or message a channel without one, so add it wherever you want the sign-off.
Doing it by hand~8.75 hrs
Negative-mention alert~35 min
Bars to scale: the time a person spends, each way.
Doing it by handyour team
- Running mention searches~30m / day
- Reading each one~30m / day
- Deciding what needs a response~15m / day
Per week, by hand~8.75 hrs
Negative-mention alertLive
- Agent searches and readsauto
- Agent classifies and branchesauto
- You act on real issues only~5m / day
One agent~35 min
FAQ
Questions, answered.
Because an alert that fires on every mention gets muted within a week, and then it protects nothing. The Branch step exists so the notification stays meaningful.
No. Everything is logged to your workspace. The branch controls what interrupts you, not what is recorded.
It fetches and reads the full page before classifying. Judging tone from a search snippet is where most false positives come from.
Free to start
Add this agentto your workspace.
Start free. Customize the steps, run it on demand or on a schedule, and approve anything that ships.