JevMade Sign in
← Back to guides

JevMade field notes / Worked recipe

Sort support messages with four separate questions

This recipe asks Jev which team a message needs, whether it is urgent, how frustrated the customer sounds, and whether they want a refund.

Original by Jeroen Erne / NexibeoClassification

Listen to this guide

JevMade’s plain-English explanation

0:00 /

Our summary

A customer message can need several decisions at once. Jeroen Erne's recipe uses Jev, an AI tool that chooses and rates answers, to identify the right team, urgency, frustration, and a request for money back. Each question addresses a different part of the message.

The program sends all four questions together. Descriptions explain where one team's work ends and another's begins. The program then uses the answer's confidence to send the message to a team, ask a person to check it, or hold it for later review.

The author reports results from a small set of hand-labelled tickets, including ambiguous messages. Those results do not establish accuracy for another company's inbox. Teams need their own examples and rules; a refund flag identifies a request rather than granting money back.

Key takeaways

  1. Describe the boundaries between teams, not only their names.
  2. Keep urgency and frustration as separate questions.
  3. Test the review cutoffs on messages your own team finds ambiguous.

Customer messages are sent to Jev, so check what personal or confidential ticket text may leave your system. The results are the author's small-sample findings.

GitHub cookbook

Read the original guide Opens the author’s site in a new tab.

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.