Prepare the question and policy

Remove names, contact details, account identifiers and any other identifying details from the question. Gather the approved policy excerpt that covers the issue.

ChatGPT describes message drafting and editing in its official overview. Check your team’s rules for using it. The output you want is a draft for review, not a sent reply.

Draft within the policy

  1. Separate the inputs. Label the anonymized question and approved policy so the draft has a clear basis.
  2. Request a bounded reply. Ask for a concise response using only the supplied policy, with gaps listed separately.
  3. Check every claim. Compare the draft with the policy. Remove any invented term, deadline or commitment.
  4. Confirm unanswered points. Ask the appropriate teammate when the policy does not cover the question.
  5. Edit and send yourself. Make the reply fit the customer’s question, then use your existing support channel.

Keep the reply grounded in approved notes

Draft a concise, friendly reply using only the approved policy below.
Do not invent terms, deadlines or commitments.
If the policy does not answer the question, list what a teammate must confirm.
Return a reply draft and a separate review checklist.

Anonymized customer question:
[Paste the question with identifying details removed.]

Approved policy:
[Paste the relevant approved excerpt.]

Check the promise, then the tone

First confirm that the answer follows the policy and addresses the question. Then make the language clear and considerate. A polished sentence can still make a promise your team has not approved.

The hub does not send messages or connect to a support inbox.