Agentic use cases · Automation
A daily ticket digest agent that writes your stand-up notes for you
The scenario
Tier 1 and Tier 2 help desk technicians often have a daily stand-up where they report what they worked on yesterday and what's on deck today. Writing those notes after closing 30 tickets is annoying. An automation that pulls your closed-ticket list from a sandbox or your own notes log and writes a tight stand-up paragraph saves you 10 minutes a day and makes you look more organized than you are.
The prompt
You are writing a tight stand-up update for a help desk technician. I will paste a list of yesterday's closed tickets and any open tickets carrying into today.
Output format (exactly):
YESTERDAY:
[one paragraph, 2-4 sentences. What I worked on, framed by themes (password resets, network issues, hardware swaps), not ticket-by-ticket. End with the single most useful thing I learned or noticed.]
TODAY:
[one paragraph, 2-3 sentences. What's open and what I plan to focus on first.]
BLOCKERS:
[one line. If there are no blockers, write "None."]
Rules:
- Total under 100 words.
- No filler ("Worked hard yesterday on a variety of issues..."). Specific over generic.
- Group similar tickets ("five password resets, two of them with locked-out accounts requiring manager confirmation"), do not enumerate.
- If I closed an unusually large or unusually small number of tickets, mention it once neutrally — let the manager interpret.
- If I name a "thing I learned," include it in the YESTERDAY paragraph. This is the part of stand-up that matters.
Yesterday's tickets:
[paste your closed-ticket list with one-line summaries]
Open / today's focus:
[paste any tickets carrying into today]
Blockers (if any):
[paste any blockers]
The output
An example digest from a real day’s tickets.
YESTERDAY:
Closed 28 tickets, mostly password resets (12) and Outlook profile rebuilds (5) following the Tuesday update. Spent 90 minutes diagnosing a recurring printer driver issue in the south-wing office; rolled back to the prior driver as a temporary fix. Noticed that all five Outlook rebuilds had the same calendar add-in installed; will flag in the channel.
TODAY:
Following up on three open tickets: the south-wing printer (waiting on a vendor reply about the new driver), one VPN issue with a remote staff member, and one suspected hardware failure pending bench drop-off. Plan to start with VPN — user has a deadline this afternoon.
BLOCKERS:
None.Notes on the artifact.
The “thing I learned” line in the YESTERDAY paragraph (here: the calendar add-in pattern) is what turns a stand-up from a status report into something a senior technician notices. The agent’s job is to surface it from your raw ticket notes; your job is to actually write the one-line note when you close a ticket so the agent has something to surface.
If you only feed it ticket numbers and titles, you get a status report. If you feed it short notes (“ticket 4421 was a password reset but the user had also been getting MFA prompts at 3am for a week — flagged to security”), the agent has texture to work with. Plant the texture as you go through the day; harvest it once at end of shift.
Where this stops earning its keep.
If your team’s stand-up is a verbal round-table and not a written post, this agent saves you nothing — read the digest yourself before the meeting and use it as your speaking notes. If your team uses a real ticketing system with reports built in (ServiceNow, Freshservice, Jira Service Management), the report is already there; the agent’s job is just to summarize it to a paragraph.
One reasonable answer. Your run may differ. Read it against the scenario before using any of it.
What to watch for
- AI will inflate impact ('successfully resolved a wide range of complex issues'). Strip it. Specific beats grand.
- AI may invent themes that were not actually in your tickets. If your day was 30 password resets and the agent says 'a mix of password and security work,' edit. The truth is more useful than the gloss.
- Sanitize before pasting. Real ticket logs contain real names, real device IDs, real internal hostnames. Replace with [user], [device], [server] before any public AI sees them.
- Stand-ups are a real political artifact in some organizations. If your manager parses each line, do not rely on AI's first draft — read it twice. If your stand-up is informal, the AI's first draft is usually fine after a 30-second skim.
- Do not have the agent auto-post to Slack or Teams. Stand-up notes are something you should read once before posting. Auto-post saves no time worth having.