They wanted to automate their customer service support system.
They have a lot of emails coming through in a day asking for support their dehumidifier equipment. They wanted to use AI to categorize emails based on error codes in the email body and send customized replies based on the error categorization.
The Challenge
The volume of Customer Support emails was high and rising as the business grew.
They wanted to keep headcount low, but still there were lots of repetitive and time consuming tasks related to answer customer emails.
The first scenario gets new emails and categorizes the problem as a Product problem, Order followup or others.
If it is a product problem the second scenario categorizes the problem using AI based on the error code. The instructions/rules to determine which error code the email falls under is specified in a document. We create an OpenAI / ChatGPT assistant based on that document.
Once the problem is classified the feedback email is sent to the client.
The Results
AI reduced the time to create drafts to emails from a couple of hours to minutes a day.
A human was involved to check on the answers and respond to non common inquires. 85% of emails were handled semi automatically.
A Notion pointer page is a single, high-signal index that tells humans and AI agents where your databases, SOPs, and dashboards live—without bloating context.
Zapier filter OR conditions let “does not” rules slip through. Learn why negatives still pass—and 3 fixes: sequential filters, Paths, or a boolean flag.