Get competitor price change alerts in Gmail, filtered by Claude

Kuse, TinyCommand and PageCrawl all check pricing pages daily or faster: Kuse at 8:00 AM New York time, TinyCommand at 9:00 UTC, PageCrawl as often as every 5 minutes. They scrape the page on a schedule and compare it with a saved copy. Kuse's template reads price, discount and stock status from each product URL, compares them with the previous day's snapshot and sends an HTML email through Gmail only if something changed. TinyCommand's recipe runs daily at 9:00 UTC, asks an AI whether a price, plan, feature or trial limit moved, and sends nothing when nothing did.

PageCrawl lets you choose how often it checks, from every 5 minutes to daily, and every alert shows the old and new price. That is a real price tracker. If you need the exact number on a competitor's pricing page within a day, use one of these.

Linkeddit monitors run weekly, with a next run of 2026-11-04 on all five test monitors, and they answer a different question: what did competitors do this week and why does it matter to you. A pricing move arrives as a signal with a category, a priority, a reason and the source link, alongside launches, messaging and partnership moves.

It also covers the half a price scraper cannot see, which is whether answer engines now recommend that competitor to buyers. A price cut that coincides with a competitor appearing in a buying question's answer is a different story from a price cut alone.

Be clear about the limit. Linkeddit does not diff a pricing page and print the old and new number the way PageCrawl does. It reads the competitor's blog, changelog and monitored pages, records a pricing move when one appears, and gives the source link. Open that link before you act on a figure, and treat the email as a prompt to look, not as the price list itself.

In the live run none of the 19 signals was a pricing move, so the email should say so in one line. The week held 12 launches, 5 messaging items and 2 filed as other. The only two high priority items were a Matomo security update and Fathom's AI assistant launch.

TinyCommand's recipe takes the other approach and sends nothing on quiet days. Either works. The prompt above asks for a first-line statement so a missing email is never ambiguous: you can tell a quiet week from a failed run.

In the live run, 2 of 19 signals were high priority and 9 were watch, so an unfiltered email would carry 17 items nobody asked for. TinyCommand's recipe exists because pages change for reasons unrelated to price, so it asks an AI to ignore layout and copy noise. Filtering by category and priority is the same idea applied to signals.

Accuracy is the other worry. A poster in a Reddit ecommerce community asked how others cope when "the price looks fine but it's actually a different variant or the page hasn't been updated properly", and added that blocked pages or failed crawls go unnoticed "until you look at the data later". That is why each signal carries a source link: open it and confirm the figure before you act on the email.

Deduplicate too. Claude can run search_threads on the previous alert and skip anything already sent, and you can ask it to label the thread with label_thread so the alerts collect in one folder.