Daily Amazon and Trustpilot review triage to Linear and Slack

Every morning at 8am in my local time zone, run an agent that triages fresh customer reviews of my products from Amazon and Trustpilot using Bright Data and routes them into the right system so nothing actionable slips through. The trigger is a daily cron.

Configuration the agent should accept: a list of Amazon ASINs to monitor, a list of Trustpilot company page URLs to monitor, the Linear team where bugs and feature requests should be filed, the Slack channel for customer-success complaints, and the Slack channel for marketing-friendly compliments. Also accept the channel for the end-of-day summary (default to the customer-success channel).

Step 1: kick off the Bright Data scrapes. For Amazon, use the Bright Data Amazon reviews marketplace dataset; for Trustpilot, use the Trustpilot reviews dataset. For each source, call Trigger Dataset Collection with the configured inputs (ASINs for Amazon, company URLs for Trustpilot). Save the returned snapshot_id values. Then poll Check Snapshot Status for each snapshot until status is ready (or failed). Once ready, call Download Snapshot Results to pull the reviews. If a snapshot fails or is still running after a reasonable wait, log it and continue with whatever is ready so the rest of the workflow does not block.

Step 2: dedupe. The agent should maintain a running list of review IDs it has already processed across runs and skip anything it has seen before. Only act on reviews that are new since the last successful run.

Step 3: classify and route. For each new review, read the star rating and review text and classify it as exactly one of: bug, feature_request, complaint, or compliment. Then route as follows:

- Bugs and feature requests: open a Linear issue in the configured team using Create Issue. The title should be a short, descriptive summary. The description should include the review excerpt, star rating, reviewer name, source URL (Amazon ASIN or Trustpilot review link), and the classification. Set priority based on severity: 1-star or wording that implies broken/unsafe/data loss is High, 2-star or repeated complaints are Medium, everything else is Low. Add a label that distinguishes bugs from feature requests.

- Complaints: post to the customer-success Slack channel using Slack Bot Send a Message. Include the review snippet, star rating, reviewer name, source, and a deep link to the original review so a CS agent can respond fast.

- Compliments: post to the marketing Slack channel using Slack Bot Send a Message, formatted as a quotable testimonial: the verbatim quote, star rating, reviewer name, and product/page it came from.

Step 4: end-of-day summary. After all reviews have been routed, post one summary message to the configured summary Slack channel with: counts per bucket (bugs, feature requests, complaints, compliments), the average star rating across all new reviews for the day, and the three reviews most likely to need a human response (lowest rating, then most emotionally charged language). If there were zero new reviews, post a one-line all-clear message.

Never file the same review twice. If a Bright Data run returns reviews the agent has already processed, skip them silently. If something fails mid-run (Linear is down, Slack rate-limited, a snapshot errors), log what was processed so the next run can pick up cleanly without double-posting.

Additional information

What does this prompt do?

Pulls fresh customer reviews each morning from the Amazon listings and Trustpilot company pages you configure.
Reads every new review and classifies it as a bug, feature request, complaint, or compliment.
Files bugs and feature requests in Linear with the review excerpt, star rating, reviewer name, source link, and a suggested priority based on how severe the issue sounds.
Posts complaints to your customer success Slack channel for fast response, and routes glowing reviews to your marketing channel as a quotable testimonial feed.
Closes the day with one Slack summary: counts per bucket, average star rating, and the three reviews most likely to need a human response.

What do I need to use this?

A Bright Data account with access to the Amazon reviews and Trustpilot reviews datasets.
The Amazon product ASINs and Trustpilot company page URLs you want monitored.
A Linear workspace and the team where bugs and feature requests should land.
A Slack workspace with two channels chosen ahead of time: one for customer success, one for marketing testimonials.
A daily run time (defaults to 8am in your local time zone).

How can I customize it?

Change the run time, run it twice a day, or add a weekend pause.
Add or remove Amazon listings and Trustpilot company pages without rebuilding the workflow.
Pick which Slack channels get complaints versus compliments, and add a third channel for high-severity bugs if you want louder alerts.
Tune how priority is assigned to Linear tickets, for example always make 1-star reviews High and 2-star reviews Medium.
Adjust the end-of-day summary to highlight a different number of reviews or to skip days with no activity.

Frequently asked questions

Will the same review get filed twice if the workflow runs the next morning?

No. The agent keeps track of which reviews it has already processed and skips anything it has seen before, so a review is only ever filed once.

What if a review does not clearly fit any of the four buckets?

The agent picks the closest bucket based on the rating and wording, and if a review is too generic to act on, it is counted in the daily summary but not filed as a ticket or alert.

Can I add more review sources later, like Google Reviews or G2?

Yes. You can extend the workflow to include any other source that Bright Data covers. Just add the new source and tell the agent which products or pages to watch.

What happens if Amazon or Trustpilot is slow to return results in the morning?

The agent waits for the scrape to finish before reading the reviews, so a slow morning just means the digest lands a little later. Nothing is lost.

Do I need a separate setup for every product I sell?

No. You can list as many Amazon ASINs and Trustpilot pages as you want in one workflow, and the agent will pull reviews for all of them in the same morning run.

Will it work without Linear, or with a different ticket tool?

Linear is the default, but the agent can be pointed at another tracker like Jira or Asana with a small change. Slack is used for the digest and alerts in either case.