Support & Account Health · YouTube video · source Sep 25, 2025 · Added
Customer Signals Agent with n8n
Turn buyer messages, refunds and reviews into replies and product fixes
A person still copies each drafted reply and sends it by hand: 'It could be automated fully, but we want to make sure that there is this actual person who who revises that.'
Per source Reported by an agency about its own client work, not independently verified.
Copy-paste prompt
Here are my last 90 days of buyer messages, refund reasons and 1-3 star reviews: [paste]. Group them by product and by issue, show which issues are growing week over week, and flag any that point to a batch or packaging problem. For each issue, suggest one product or packaging fix and one listing change. Then draft a reply template for each common buyer message type that follows my policy: [paste policy].
The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].
What they did
In a Seller Fest session, Michal Špecián of the agency Orange Klik showed an AI agent his team built for an eight-figure Amazon seller. n8n pulls Amazon data into Supabase, and a Lovable dashboard sits on top with three parts: intelligence reports that flag anomalies and product or batch issues across buyer messages, refunds and reviews; an advisor chatbot that answers questions about that data; and tools that classify buyer messages, judge how frustrated the customer is, propose a refund or replacement under the seller's policies with the reasoning, and draft a reply. Another tool scans negative reviews for guideline violations to file removal cases.
How it went
A person still copies each drafted reply and sends it by hand: 'It could be automated fully, but we want to make sure that there is this actual person who who revises that.' He says review removal is 'not 100% effective.' No time or sales figures are given.
We only report what the source says. Results are self-reported unless noted.
Worth knowing
An agency showing a client build to promote its services. He says Claude helped draft the n8n workflows, but 'It's not plug-and-play,' and advises weighing build and maintenance time against the time a workflow saves.