JIRA × AI
Top 10 JIRA tasks you can replace with AI prompts — and reclaim your sprint.
Ticket descriptions. JQL queries. Sprint summaries. They're the tax you pay for using JIRA. Here are ten JIRA jobs you can hand to ChatGPT or Gemini, with a copy-paste prompt for each.
Estimated time saved: 4–6 hours per sprint.
- 1.
Writing ticket descriptions
The pain: A one-liner ticket that an engineer has to ping you to understand.
Try this prompt"Turn this one-line bug report into a JIRA ticket with: Summary (1 line), Description, Steps to Reproduce, Expected, Actual, Environment, Severity (1–4) with reason. Be concise."
⏱ Saves about 5 min per ticket.
- 2.
Acceptance criteria
The pain: Writing 'Given/When/Then' for the 12th time today.
Try this prompt"Here is a user story. Write 4–6 acceptance criteria in Given/When/Then format covering happy path, validation errors, permissions, and the obvious edge case. Mark any criteria that need design input."
⏱ Saves about 10 min per story.
- 3.
JQL queries
The pain: Remembering whether it's 'status' or 'Status' and which custom field is which.
Try this prompt"Write a JQL query for: tickets in project ACME, assigned to me or my team (members: x, y, z), updated in the last 14 days, not Done, ordered by priority desc. Explain each clause."
⏱ Saves about 10 min per query.
- 4.
Sprint planning estimates
The pain: Staring at a backlog trying to guess what fits in 2 weeks.
Try this prompt"Here are 20 backlog items with rough complexity (S/M/L). My team velocity is 35 points. Propose a sprint with a balanced mix, flag dependencies, and give a 1-line risk note per item over M."
⏱ Saves about 30 min per planning.
- 5.
Standup summary from tickets
The pain: Reading 40 ticket updates to find what actually moved yesterday.
Try this prompt"Here are yesterday's ticket comments and status changes. Summarise for standup: Done, In Progress, Blocked. Group by person. Flag anything that hasn't moved in 3+ days."
⏱ Saves about 15 min per day.
- 6.
Bug triage
The pain: Sorting 30 incoming bugs by what to fix first.
Try this prompt"Here is a list of 30 bug reports. Score each by impact (1–5) and effort (S/M/L). Propose a top-10 fix order with a 1-line reason. Flag anything that smells like a regression."
⏱ Saves about 30 min per triage.
- 7.
Release notes from done tickets
The pain: Translating 40 cryptic ticket titles into customer-facing notes.
Try this prompt"Here are the done tickets for release 2.4. Write public release notes grouped as: New, Improved, Fixed. One bullet per change, customer language, no internal jargon. Add a 2-line summary at the top."
⏱ Saves about 30 min per release.
- 8.
Retrospective themes
The pain: Reading 50 sticky notes and squinting for the pattern.
Try this prompt"Here are notes from our retro under What Went Well / What Didn't / Ideas. Identify the top 3 themes, the strongest signal in each, and propose 1 experiment to try next sprint per theme."
⏱ Saves about 20 min per retro.
- 9.
Splitting epics into stories
The pain: An epic so big it's been 'in progress' for 4 sprints.
Try this prompt"Here is an epic description. Split it into 6–10 user stories that each deliver value independently. For each: title, 1-sentence value, rough size (S/M/L), and any dependency."
⏱ Saves about 45 min per epic.
- 10.
Status update for stakeholders
The pain: Writing the same weekly update with slightly different numbers.
Try this prompt"Here are this week's done tickets, in-progress tickets and blockers. Write a 150-word stakeholder update with: progress, risks, asks, what's next. No jargon. End with 1 question for the reader."
⏱ Saves about 20 min per week.
How to use these prompts
- 1. Open ChatGPT, Gemini or Claude.
- 2. Paste the prompt above.
- 3. Paste your JIRA content (or a sample) underneath.
- 4. Ask for the result in a format you can paste straight back into JIRA.
For sensitive data, use a business/enterprise plan that doesn't train on your inputs, or anonymise names and figures first.
Frequently asked
Is AI better than Atlassian Intelligence?
Atlassian Intelligence is convenient because it lives inside JIRA. For pure quality of summaries, ticket rewrites and JQL, a general model like ChatGPT or Gemini usually wins. Use both.
Can I trust AI with internal ticket data?
Treat tickets like any internal doc. Use a workspace/enterprise plan that doesn't train on inputs, or anonymise names, customer IDs and revenue figures before pasting.
What about Linear, Shortcut, Asana?
Same prompts work — just swap the field names. Ticket descriptions, acceptance criteria, retros and status updates are tool-agnostic. The structure is what AI is good at.
Want one beginner-friendly prompt every morning? One Prompt a Day sends one every day — free.