Mentionova

Agent · Monitoring

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.

Add to workspace

6 stepsRuns on demandWeb search · HTTP request · AI call · Filter / Branch · Workspace data
6chained stepsOn demandrun mode5step types1click to add
Trusted by 70+ brands & agencies
Client logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logoClient logo

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.

  • Only fires on negative — no alert fatigue
  • Sentiment read from the full page, not a snippet
  • Tells you what the complaint is, not just that one exists
  • Positive and neutral still logged, quietly
Web search
Negative-mention alert
find_recent_mentions2.1s
fetch_mention_pages1.7s
classify_sentiment1.5s
branch_on_negative0.2s
summarize_the_issue1.4s
notify_and_log0.4s
6 steps · Runs on demand
Every run logs each step’s input, output and runtime.

The steps

What happens on every run.

1

Find the recent mentions

Fresh mentions of the brand across the web.

2

Fetch the pages

Sentiment from a snippet is guesswork; the agent reads the source.

3

Classify the sentiment

Positive, neutral or negative, per mention.

4

Branch on negative

The Branch step is the point: positive and neutral stop here.

5

Summarise the issue

For the ones that pass, what the complaint actually is.

6

Notify and log

You hear about it, and it is on the record.

Before & after

The time this gives back.

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

Steps that touch the outside world — a CMS, a channel, a live page — gate on human approval. You keep final approval.

FAQ

Questions, answered.

Why not alert on everything?
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.
Are positive mentions thrown away?
No. Everything is logged to your workspace. The branch controls what interrupts you, not what is recorded.
How does it judge sentiment?
It fetches and reads the full page before classifying. Judging tone from a search snippet is where most false positives come from.

Other agents in Monitoring

See the whole agent library →

Trusted by 70+ brands & agencies

Add this agent to your workspace.

Start free. Customize the steps, run it on demand or on a schedule, and approve anything that ships.

Add to workspace

First Growth Signal in ~2 minutes