Disclosure: I'm the developer of this app.
Every release, someone has to turn a pile of Jira tickets into release notes a human would read. On the teams I've been on, that meant opening each card in the version, reading the title and description, and typing up a shortened summary to send to whoever needed it. More recently it means copying those titles and descriptions into a company AI chat and tidying up what comes back. Either way, most of the time goes to getting the ticket content out of Jira — and when the notes are going to stakeholders or customers rather than your own team, the wording (or lack thereof) matters more.
I built AI Release Notes for Jira to do that job inside Jira, in a couple of clicks.
What it does
- Pick a source. A project version (fixVersion) or a sprint from any board. It shows you the exact issues in scope before anything is generated.
- Pick an audience. Internal notes are complete and technical — every ticket, including chores and refactors. Customer-facing notes are selective and benefit-oriented: ticket titles rewritten as plain sentences, bugs described by symptom rather than cause, security fixes announced without describing the vulnerability, and vague tickets left out instead of dressed up.
- Edit before publishing. The title and full markdown are editable in place. Treat it as a first draft — you get the final say, if desired.
- Publish to Confluence. Create a new page or update an existing one in any space you can write to. It remembers each project's usual space. Or grab the markdown from the editor and use it anywhere.
- Sprint reports too. A closed sprint gets an end-of-sprint report with carried-over work stated plainly; an active sprint gets a mid-sprint progress report.
Notes generated from a sprint, with the issues in scope. Internal audience, technical wording.
The same sprint, customer-facing — the XSS fix becomes "a security improvement," issue keys drop, and the vague test task is left out
Publishing Jira release notes to Confluence — new page or update an existing one
Why it's built the way it is
It runs on Atlassian Forge and uses Atlassian-hosted AI for generation, so your ticket content stays inside Atlassian — there's no external AI provider and no third-party server involved. That was a deliberate design choice; I'd guess it's a requirement for many enterprise teams. It also doesn't draw on your team's Rovo, Copilot, or other AI allowances. The app's subscription covers the AI usage — there's no separate token budget, metering, or AI charge on your side.
Works with company-managed and team-managed projects, and with kanban teams via the version source.
Marketplace:
https://marketplace.atlassian.com/apps/3656398471/ai-release-notes-for-jira
Docs:
https://northboundlogic.net/docs-release-notes
Questions and feedback welcome in the comments — I read all of them.
— Northbound Logic
