AI THAT WATCHES WEBPAGES FOR YOU

Know the moment a webpage changes.

Paste a link. Say what to watch for in plain English — a price drop, a new job, a restock, a new release. PageDrones checks it for you and emails you only when it actually happens.

Try one:

Free to start. No credit card. Just an email for the alerts.

northwind.dev/pricing
Plans for every team
Starter
$0
Pro
$20$15/mo
CHANGE DETECTED
Team
$40/mo
MISSION MATCH · northwind.dev
Northwind dropped Pro from $20 to $15/month.
Your drone is watching for Pro getting cheaper.
SCANNED 09:14 · VIEW EVIDENCE →
PROMPT · EVERY HOUR
Tell me if the Pro plan gets cheaper.
STATUS
ON PATROL
  • Anthropic Pricing · hourly
    Prompt: new plans or price changes
    On schedule
  • "frontier model release" · daily
    Whole-web prompt: a new frontier model
    On schedule
  • Tahoe Cabin · every 15 min
    Prompt: availability Feb 12–16
    On schedule

01 · THE PROMPT

Write the prompt. Skip the pipeline.

Paste a URL (or a search topic), describe what matters in plain English, pick a schedule. That's the whole setup.

No cron jobs. No scrapers to patch. No CSS selectors. No diff parsing.

“Tell me if they lower the price.
“Alert me when applications open.
“Watch for new product launches.
“Tell me if this language changes.
“Let me know when the cabin is available Feb 12–16.
“Tell me when the final rule is published.

02 · THE PIPELINE

The whole monitoring pipeline, managed.

Every stage of a hand-rolled cron + scraper + LLM stack, already built and running: scheduling, fetching, prompt execution, change detection, evidence and delivery.

Scheduler

Every 15 minutes to once a week, or a custom cron expression. Time zones handled; no server to keep alive.

Prompt execution

Your prompt runs against the fresh content on every scan. A moved button isn't news; a new price is. The model decides what changed and whether it matters.

Fetch pages or search the web

Point a monitor at a URL and it's scraped, JavaScript and all. Give it a topic instead and it runs a web search on every scan.

Evidence, stored

Every run is logged; a screenshot is stored every time something changes, so you can see exactly what was fetched.

Change detection

Content hashes catch identical fetches; the model catches semantic changes. Alert on matches only, or get a digest every run.

Delivery

Email, Slack or Microsoft Teams, with a delivery log per run. Written like a colleague's note, not a diff dump.

03 · SETUP

From zero to a running pipeline in under a minute.

  1. 01

    Give it a source

    Paste a URL — or describe a topic and let it search the whole web.

    https://competitor.com/pricing
  2. 02

    Write the prompt

    Plain English. The way you'd brief a colleague.

    Tell me when they change plan prices or limits.
  3. 03

    Pick a schedule and deploy

    Test it once and see the exact alert you'll get. Then let it run.

    DEPLOYED · NEXT RUN 10:00

04 · THE ALERT

It reports. It doesn't chatter.

Every alert leads with what happened, why you're seeing it, and the evidence. Read it in five seconds. Act or ignore.

  • → What changed, in one sentence.
  • → Why it matched your mission.
  • → A snapshot of the page as proof.
PAGEDRONES ALERT SCANNED 2:34 PM

Northwind changed its Billing pricing.

What happened
Northwind added a 0.7% fee for usage-based billing. Previously 0.5%.
Why you're seeing this
Your drone is watching for changes to Northwind Billing pricing.
Changed
0.5% → 0.7%
VIEW EVIDENCE →ADJUST MISSIONPAUSE DRONE

06 · THE DIFFERENCE

You could build this. You shouldn't have to maintain it.

A cron job, a scraper, a snapshot store, a diff, an LLM call and a Slack webhook is a weekend project — and a permanent on-call rotation once the pages change under it.

HAND-ROLLED STACK
  1. WRITE SCRAPER
  2. SET UP CRON + HOSTING
  3. STORE SNAPSHOTS
  4. DIFF + PROMPT + PARSE
  5. WIRE UP ALERTS
  6. FIX IT WHEN THE PAGE CHANGES ×47
PAGEDRONES
  1. URL OR TOPIC + PROMPT
  2. PICK A SCHEDULE
  3. RUNNING

Then nothing — until it matters.

07 · QUESTIONS

Questions.

What is a scheduled web data monitoring pipeline?

A job that fetches web data on a schedule, runs a prompt or rule against it, compares the result with the last run, and delivers an alert when something changed. PageDrones is that pipeline as a managed service: you provide the source, the prompt and the schedule.

What can it monitor?

Any public webpage — pricing pages, product pages, careers pages, government sites, availability calendars. You can also give a monitor a topic instead of a URL and it will run a web search on every scan.

Will it alert me every time a pixel moves?

No. Each run compares the fetched content with the previous scan and runs your prompt against it. If the change doesn't match your prompt, it stays quiet. If you'd rather get a report on every scan, switch the monitor to “every run.”

How often does it run?

Anywhere from every 15 minutes to once a week, in your time zone. Custom schedules — say, weekdays at 9am — are a couple of clicks or a cron expression.

Where do alerts go?

Email, Slack, or Microsoft Teams. Add as many destinations as you like per monitor. Every run and delivery is also kept in your log with a snapshot of the page.

Can I see what it will send before I commit?

Yes. Every monitor has a Test button that runs the pipeline once, right now, and shows you the exact alert — before anything is saved or sent.

What does it cost?

PageDrones is free during early access. No credit card required. Paid plans will come with generous free usage.

Stop babysitting cron jobs.

Give us the page and the prompt. We'll run the pipeline.

FREE DURING EARLY ACCESS · NO CREDIT CARD