Blog

Grouped changelogs from raw commits or release notes — feature, fix, improvement.

At a glance

  • Input: raw git commits or existing release notes
  • Output: a grouped changelog — feature, fix, improvement — for release pages or in-app notices
  • Best for: teams publishing release notes without rewriting commit messages by hand
  • Limits: it rewrites what you paste — vague commits produce vague notes

Frequently asked questions

What does Changelog AI do?

It converts raw git commits or release notes into a clean, grouped changelog — feature, fix, improvement — that you can publish to a release page, an in-app notice, or GitHub releases.

Does it follow Keep a Changelog?

The output follows the human-first principle behind Keep a Changelog: grouped, readable changes instead of a raw commit dump. Version labels stay yours to set.

Do I need an OpenAPI spec or a specific commit format?

No. It works from raw commits or existing release notes. Conventional Commits help the grouping, but they are not required.

Can it publish directly to GitHub Releases?

It produces the changelog text. You paste it into wherever you publish releases, after a read-through.

How is this different from dumping the git log?

A commit log is a build artifact. This groups the changes by type and rewrites them in language a user can understand, which is the difference between a changelog and a log.

What are the limits?

It can only work with what you give it. Vague or truncated commit messages produce vague notes, and it does not verify that a change is actually complete.

Deep-dives

HowTo

What is Changelog AI — and when to use it

What a grouped changelog is for, and the signs your release notes need one.

HowTo

How to write release notes in 10 minutes

Paste commits, generate groups, edit for readers, publish.

FAQPage

Common changelog mistakes (and how to avoid them)

Shipping raw commit spam, grouping everything as a fix, and writing for engineers only.

Article

Generated changelogs vs. hand-written release notes

Where automation is enough and where a human pass still matters.