Watch competitors weekly
Lock 3-5 ASINs and the dimensions that matter, then let a browser-driven collection run scrape what has no API: a daily diff, an alert the moment something changes, not a full re-read of every page.
Copy this prompt to start your first task:
What is this workflow?
"Watch competitors weekly" is the job of locking a small set of ASINs and the dimensions that matter, then letting a browser-driven collection run gather what Amazon has no API for. Each pass produces a diffable snapshot per ASIN — price, Buy Box owner, BSR, rating, review count and velocity, listing changes, coupons, ad placement — feeding a live comparison dashboard and a daily "Competitor Activity Brief" that comes with an actual recommendation attached, not just numbers. A same-day alert breaks that daily rhythm only for the handful of things that can't wait: a competitor's price drop past a threshold, a new entrant breaking into the top ranks, losing the Buy Box, or an abnormal spike in negative reviews.
Collection discipline, not optional: runs 1-2 times a day with randomized delay between actions. Scraping Amazon at high frequency reliably trips anti-bot defenses, and a flagged account produces zero data going forward. If a CAPTCHA or login wall shows up mid-run, it stops and alerts a human rather than trying to force through it. At any real scale, it prefers paid data from a Keepa or Helium10 account you already log into over self-scraping the storefront directly. Self-scraping is the fallback for what those tools don't cover, not the primary source. And it never writes a silently empty field: a failed field extraction triggers a retry and an alert, because a page-structure change on Amazon's end blanks fields exactly this way, and a silent empty write is how that goes unnoticed for weeks.
How to use this workflow
Hire a Researcher: write the bio around locked scope, not "track my competitors"
Open AI Teammates in the sidebar, click Hire AI Teammate, and pick the Researcher template. The default bio is generic web research. Replace it on the Profile step so the teammate starts out already knowing this is a scoped, recurring collection job with a real procedure behind it, not an open-ended watch:
Confirm the name and click Create teammate. They land in their own channel, which is where you'll send everything below. More on hiring →
Connect what actually has a connector: Notion for storage, Slack for alerts
The comparison database and the alert channel are both real native connectors here. Amazon itself and the tools built to track it are not; that gap gets handled two steps down. Connect your tools →
Notion holds the daily snapshot table: the thing every future diff compares against. Slack (or Gmail as a backup channel) is where the daily brief and same-day alerts land. One Slack-specific step matters here: the teammate's Slack app doesn't join channels by itself, so after connecting, invite it into the channel the alerts should land in — otherwise the brief is written but never arrives. Connect Slack → If you keep a paid Keepa or Helium10 account for historical price/BSR charts, that's not a connector either. It gets read the same way as Amazon itself, covered next.
No connector for Amazon, so it collects through a real browser instead
Neither Amazon's storefront nor Keepa/Helium10/Jungle Scout have an API Helio can connect to natively. There's no OAuth button for any of them. What this teammate actually does is install the Stagehand Browser CLI skill (Skills & Plugins → Productivity) and drive an actual browser: open the product page, read the price/BSR/rating/review-count/Buy-Box-seller fields, screenshot the full page as an archive. Login credentials for Seller Central or a paid Keepa/Helium10 account live in Helio's credential vault and get pulled at login time, never typed into a prompt or hardcoded anywhere. Vault access is granted per credential, so add those logins to the Vault and share each one with this teammate now — the first paid-account login blocks until you do. Share credentials →
Build the baseline in Notion, one row per ASIN and one snapshot per day
With one clean collection pass proven out, lock the full tracked set and the schema everything after this diffs against:
Automate the daily diff-and-brief
Open Automations in the sidebar and click Create via Chat. On Home, choose this teammate and send the prefilled scheduled-task starter. When the interview asks what to schedule, use the recurring job below. Once or twice a day is enough, staggered rather than on the hour:
Same-day alerts, separate from the daily brief
Four things don't wait for tomorrow's brief: a competitor drops price past a threshold you set, a new competitor breaks into the top ranks for a tracked keyword, you lose the Buy Box, or a competitor's negative reviews spike abnormally (often a sign of a bad batch or a stockout, sometimes review manipulation):
Nothing here needs your approval by default: every step reads a page — public storefront pages, or a paid Keepa/Helium10/Seller Central account through a vault-shared credential — and writes to Notion or Slack. Nothing is sent externally or is irreversible. See Control for what does pause for sign-off.
Tips for Better Results
| Tip | Description |
|---|---|
| Lock the tracked set before automating | Confirm the exact ASINs and keywords once, in the baseline pass — every future diff compares against that fixed set, not a moving target. |
| Prefer paid data over self-scraping at scale | If you already log into a Keepa or Helium10 account, read from it. Self-scraping Amazon directly is the fallback for what those tools don't cover, not the primary source. |
| Never let a failed field go in blank | A missing price or BSR usually means Amazon changed its page structure, not that there's nothing to report. The teammate retries once, then flags it instead of writing an empty value. |
| Keep same-day alerts separate from the daily brief | Only four things justify an immediate ping — a threshold price drop, a new top-rank entrant, a lost Buy Box, or a review spike. Everything else waits for the brief. |
Common Questions
Research
Back to the research use case overview.
Research a market
Research the whole market, not just named competitors.
Research a market
Give it the questions you need answered about a market. It researches, cross-checks sources, and keeps tracking the space weekly.
Gate-check usage data
Push back on what "active" means before pulling a single number, catch a suspicious zero before it ships as a finding, and never call an automation "running" without a schedule ID and a run record to prove it.