Competitor monitoring that reports what changed
Agents watch competitor sites, changelogs, pricing pages and news — and brief you on what actually changed, with links to the evidence.
Opens in AGNT and takes about a minute.
Didn’t open? AGNT may not be installed here yet. Download AGNT free · Download the .json
The short version
| Tools it needs | Firecrawl, Slack, Notion, Google Sheets, X (Twitter) |
|---|---|
| Setup | 4 steps, about ten minutes |
| Approval | Yours, per category — nothing is sent on your behalf unless you say so |
| Where it runs | Your own machine. Credentials stay in a local vault. |
What the agent actually does
You hear about the change that matters
Pricing pages, changelogs, docs, careers pages, funding news. You choose the pages; the agent checks them on a schedule and keeps a history so trends are visible rather than just today’s snapshot.
Diff, not dump
A page-changed alert is close to useless because pages change constantly for trivial reasons. The agent reads the diff and reports whether anything meaningful happened — a price moved, a tier appeared, a claim changed — and stays quiet when a footer was reworded.
Briefing where you work
Findings arrive in Slack or your inbox as a short written brief with links to the evidence, not as a raw feed you would need to triage yourself.
The tools this job needs
Firecrawl
Where the work arrives. Clean page extraction for research-grade scraping. The agent watches it and reads what turns up in full, rather than matching a rule against a subject line.
Slack
Context the agent pulls in before deciding. Team chat where operations actually surface. It is read, not just referenced — which is what lets the decision account for it.
Notion
Context the agent pulls in before deciding. The team wiki agents can actually keep current. It is read, not just referenced — which is what lets the decision account for it.
Google Sheets
Where the result lands. The database every team already has open. Nothing is written here until the agent has formed a view and, where you asked for it, you have approved it.
Set it up
- List the pages worth watching per competitor and how often each should be checked.
- Tell the agent what counts as material for you — pricing, packaging, positioning, hiring, or all of it.
- Choose the delivery channel and cadence, including permission to say nothing happened.
- Review the first fortnight and tune the materiality threshold until the briefings are worth opening.
You can see exactly what it did
The brief it works from
This is the actual instruction set the template installs — what the agent is told to do, and what it is told never to do. Every line of it is yours to edit in AGNT after install.
Read the full brief
You watch competitors and report what actually changed.
Check what the user listed
Pricing pages, changelogs, documentation, careers pages, news. Keep a history so you can describe trends rather than snapshots. The history is what makes you useful; without it you are a page-changed alert, and those get muted.
Report materiality, not diffs
A page changes constantly for trivial reasons. Read the diff and decide whether a human would care. A price moved, a tier appeared or vanished, a claim changed, a feature shipped, a role was posted that implies direction — those matter. A reworded footer, a new stock photo, a legal boilerplate update do not.
SAYING NOTHING HAPPENED IS A VALID AND VALUABLE REPORT. Inventing significance to justify the run is how a monitoring channel loses its audience, permanently and quickly.
The signal is usually the pattern
Any single change is mildly interesting. The sequence is what matters: three enterprise-flavoured features, two enterprise sales roles, an enterprise tier added to pricing. No individual alert says "they are moving upmarket" — the accumulation does, and you are the only one holding all of it. Name the pattern when you see it, and say how confident you are.
Always link the evidence
Every claim points at the page you read. If a page was unreachable, say so — a monitor that silently stops working is worse than no monitor, because it reads as "nothing is happening".
Boundaries
Read public pages only, on a polite schedule. Nothing behind a login, nothing requiring an account, nothing you would not access by visiting the page yourself. Never impersonate a customer or sign up for a competitor's product.
Be factual, not editorial. The user forms the strategy; you supply what changed and what it suggests.
Shown exactly as it ships. The agent inherits whichever model you already use.
The signal is in the pattern, not the event
Any single competitor change is mildly interesting. What is genuinely useful is the sequence: three enterprise-flavoured features shipped, two enterprise sales roles posted, an enterprise tier added to pricing. No individual alert says "they are moving upmarket"; the accumulation does. Because the agent keeps the history and reads across it, it can name the pattern rather than reporting the third event in isolation.
Why most monitoring gets switched off
Change-detection tools alert on noise, teams learn to ignore them, and the whole thing quietly dies within a month. The fix is not better filtering rules but a step that can judge materiality — something that reads the change and decides whether a human would care. Silence on an unimportant week is a feature, and it is the one that keeps the channel trusted.
Positioning is a public document
Competitors publish their strategy continuously and mostly by accident: in what the docs now explain, what the careers page is hiring for, what the pricing page stopped mentioning. Reading that consistently is a research habit almost nobody sustains manually, which is precisely why automating it produces an edge that compounds.
Common questions
Is this scraping?
It reads public pages the way any browser does, on a polite schedule. Nothing here requires access you would not have by visiting the page yourself.
What if a competitor blocks automated access?
The run reports the failure rather than pretending nothing changed — a monitor that silently stops working is worse than no monitor.
How often should it run?
Weekly suits positioning and pricing. Changelogs and news are worth daily if you act on them; anything more frequent is usually noise.