The work is manual and it disappears
Copying a competitor's price into a spreadsheet takes hours and is outdated in two days. Whoever did it knows; whoever didn't starts from zero.
All services
The appetite · Collection
The internet works through the night. Your team opens the finished table in the morning.
Extract a page, crawl an entire site or watch a price table that changes without notice. The engine is ours and runs inside your own infrastructure: the address you monitor never passes through any third party server.
DELIVERED ASClean MarkdownJSON with schemaEntire siteChange alert
ACTIVE MONITOR
concorrente.com.br/precos
Mon 06:00no changestayed silent
Tue 06:00no changestayed silent
Wed 06:00no changestayed silent
Thu 06:00R$ 199 R$ 179ALERTED
WEEKLY COST4credits: one read a day, alert or not
Example of a daily monitor. An unchanged page doesn't generate an alert, doesn't generate a webhook and doesn't generate an email: an alert that always fires is an alert nobody reads.
The value isn't in reading the page. It's in knowing it changed.
Extracting once solves a one off problem. Knowing the competitor dropped their price yesterday, that the supplier changed the deadline or that the bid opened. That changes what the company does that day.
What happens today
Someone on the team opens 30 tabs every Monday.
The change only shows up when it hurts
You find out the competitor dropped the price when the client mentions it during negotiation. Two weeks after it happened.
The outside tool takes your data with it
Cloud scraper sends the address (and sometimes the content) to their server. For anyone collecting contracts, bids, or confidential pricing, that's already grounds for a veto.
What you start to see
Yesterday's page and today's, side by side.
Each check saves a snapshot of the page. When something changes, the system shows what it was and what it became, in the exact field. Not just "something changed on this URL".
EXAMPLE OF A SNAPSHOT
0
Pages per month0
Active monitors0h
Check interval0
Alerts worth havingPages per month
How many pages have come in, with clean markdown or JSON in the format you asked for, ready for another module to use.
The free plan already includes 2,000 pages per month
Active monitors
Up to fifty pages watched per brand, each with its own interval and its own monthly spending cap.
A per monitor cap keeps the bill from running away
Check interval
Hourly, every six hours, twice a day, daily, or weekly. You choose per page, not for all of them.
Price changes by the hour, delivery time doesn't
Alerts worth having
Only what changed becomes an alert. Three identical checks pass in silence, and that's what makes someone open the fourth.
A rare alert is a read alert
Under the hood
Four things that set this apart from a script.
Anyone can write a scraper in an afternoon. What costs you is everything that comes after: a page that changes layout, an address that dies, an alert that turns into noise, and a bill that creeps up unnoticed.
The engine is ours, and it runs on your side
This isn't a wrapped cloud service. The address you monitor and the content that comes back never pass through a third party server, which is the only way to collect contracts and bids without asking legal for permission.
the data leaves the site and enters your infrastructure
Only alerts if it changed
Each reading is compared with the previous one in the field that matters. An unchanged page doesn't generate an alert, a webhook, or an email.
the monitor's email gets read again
Spending cap per monitor
An hourly monitor uses 24 credits a day. Forgetting one turned on would silently drain the account, so each one has its own monthly limit.
nobody wakes up to a zeroed out balance
Pauses itself when the target dies
Five failures in a row and the monitor stops, saying why. Better stopped with a warning than burning credits against an address that no longer exists.
failure becomes a warning, not a bill
To settle any doubt
Four comparisons that settle it.
An in house script works, a cloud scraper works, and an intern with a spreadsheet works. The point is what each one costs you as volume and confidentiality grow.
A script from the teamCollection
When the site changesBreaks silently and nobody notices Fails five times, pauses, and alerts
MaintenanceFrom someone with other priorities None on your side
HistoryWhatever fits in a file Stored and comparable from the first reading
From zero to the first snapshot
Three steps. Your first table in minutes.
011 MINUTE
Paste the address
One page, a list of pages, or an entire site to discover and track.
023 MINUTES
Say what you want back
Clean Markdown for reading, or the exact fields you need, name, price, deadline, in JSON.
03SAME DAY
Turn on the monitor
Choose the interval and spending cap. From then on, you are only notified when something changes.
Inside the company
The same collection, three uses.
Enter the negotiation knowing the price the competitor published yesterday, not last quarter.
Track supplier catalogs and lead times without depending on someone opening the site every week.
See market movement as a time series, instead of a screenshot in the WhatsApp group.
What comes out of here, concretely
01
The page in clean Markdown
No menu, no banner, ready for an AI to read.
02
The table with the fields you asked for
JSON in your schema, straight to the spreadsheet or database.
03
The change alert
What it was, what it became, when, and where you want it sent.
Where the alert arrives
A detected change does not wait for someone to open the dashboard: it arrives where the team already talks.
SlackEmailWhatsApp
View Connect How much it costs
You pay per page read.
There is no scraper subscription here: it is the same balance that runs the other modules, spent per page actually retrieved.
One page read3 credits. Extract, track, or map: the unit is the page retrieved.
1 credits
One monitor check3 credits, the same calculation: one read of the monitored page, whether it changed or not.
1 credits
Page that requires a browser8 credits, when content only appears after JavaScript runs.
1 credits
The free plan includes 2,000 pages per month and two monitors. A daily monitor uses 90 credits per month, so you can watch the two pages that matter most without a card.
See all plansBefore you ask
What people usually want to know.
Does this respect robots.txt?
Yes, by default. It tells you when a page was rejected because of it instead of pretending it was empty. For pages you control yourself, you can deliberately disable this safeguard: the decision is recorded.
What if the site layout changes?
Extraction is based on meaning, not a fixed HTML position, so layout changes usually go unnoticed. When they do not, the monitor fails, retries, and if it fails five times in a row, pauses and tells you why, instead of continuing to spend on nothing.
Can I monitor a page that requires login?
Yes, with the credential stored in your account vault. Because the engine runs within your infrastructure, that credential does not go anywhere: this is the kind of thing you do not do with a cloud scraper.
How many monitors can I have?
Up to fifty per brand, each with its own interval and monthly credit cap. The cap prevents a forgotten hourly monitor from silently consuming the account.
What happens to what was collected?
It stays in your database, encrypted with your key, available to the other modules and fully exportable whenever you want. It does not become anyone's model training data.
THE OTHER ORGANS
What it reads, the others use.
Collection without a destination is a stagnant file. Here, what comes in becomes context for the other modules in the same minute.
THE OTHER ORGANS
No module works alone.
Each one solves a front, and all write to the same memory. See the others and how they fit with this one.