prompts.chatprompts.chat
Login
All classifiers
ChoiceCustomer support

Reply resolution check

Read a customer's reply to tell whether the issue is resolved, still open, or a new issue.

F@f

Problem

Tickets reopen on every "thanks!" and close on replies that actually say the fix didn't work. Agents re-read whole threads to find out which.

Goal

Close resolved tickets, reopen unresolved ones with priority, and split new issues into their own tickets.

Input / state

The customer's reply and a one-line summary of the issue.

reply
The customer's latest reply.
issue_summary
One line describing the original issue.

Decision question

ChoiceQuestion key resolution_status

What does the customer's reply say about the original issue?

Options

  • resolvedConfirms the issue is fixed or thanks you without raising anything new.
  • still_unresolvedSays the original issue is still happening or the suggestion didn't work.
  • new_issueThe original issue is fine, but the customer raises a different problem.

Allowed outcomes

Decided in your code from the answer and its confidence. Thresholds are yours to set and test.

  • Close ticketChoice is resolved and confidence meets your threshold.
  • Reopen with priorityChoice is still_unresolved and confidence meets your threshold.
  • Open a new ticketChoice is new_issue and confidence meets your threshold.
  • Agent reads the threadConfidence is below your threshold: probability is spread across options.

Example scenarios

Predefined demonstrations. No model is called; the outcome shown is an illustrative judgment.

Example state

{
  "reply": "That worked, export is fine now. Thanks!",
  "issue_summary": "CSV export times out"
}

Outcome

Illustrative
Close ticket

Choice is resolved and confidence meets your threshold.

Why

Confirms the fix.

Using it well

  • Pass a short summary of the original issue instead of the whole thread; long threads blur which problem is meant.
  • Let customers reopen a ticket with one click whatever the answer was.

Request definition

POST https://api.typesafe.ai/v1/systemone
{
  "model": "jev-latest",
  "state": {
    "reply": "<reply>",
    "issue_summary": "<issue_summary>"
  },
  "questions": {
    "resolution_status": {
      "type": "choice",
      "instructions": "What does the customer's reply say about the original issue?",
      "criteria": {
        "resolved": "Confirms the issue is fixed or thanks you without raising anything new.",
        "still_unresolved": "Says the original issue is still happening or the suggestion didn't work.",
        "new_issue": "The original issue is fine, but the customer raises a different problem."
      }
    }
  }
}

Request shape from the System One API reference; a Kev server accepts the same body. Replace the placeholder state with your data. Requests need your own API key or server.

Version History

v1Current Versionabout 8 hours ago@f

Initial version

Comments (0)