Turn support evidence into ticket and reply drafts

Cursor lets support teams gather logs, compare old cases, and draft a bug report plus customer response for review.

Done with Cursorยท Intermediate

What they did
The support team runs Cursor in Ask Mode with the codebase opened in a multi-root workspace. MCP connections feed in Datadog logs and traces, plus Slack, Notion, and Linear records, while commands and skills help them search telemetry, identify known issues, and assemble drafts. They also use multiple subagents for log review, issue matching, ticket writing, and reply writing, then a human reviews the merged output.
How it went
The team says Cursor now handles over 75% of support interactions. They estimate it gives each support engineer 5โ€“10x more output, but that figure comes from their own judgment rather than a measurement.
Tips
Add clear redaction rules to customer-reply prompts so drafts do not expose internal service names, error codes, or file paths. Keep a person in the loop before any ticket is filed or message is sent.

Try it yourself with Cursor

A customer reported [problem] on [date] for [account]. With my connected [logging system], [past-case source], and [ticket tracker], find the matching logs and any similar earlier cases. Then write a bug ticket and a customer reply, include the evidence you relied on, and wait for my approval before filing, sending, or deleting anything.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on cursor.com โ†—

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.