Jul 1, 2026

From Git commits to organizational knowledge

Key Takeaways

Raw commit messages are often cryptic ("fix bug in auth", "wip").

AI can inspect the code diff itself to understand true architectural intent.

Transforming commits into structured markdown creates living team documentation.

In every software organization, there is a recurring disconnect between what engineers ship in code and what customer success, support, and sales teams understand.

Release notes are frequently written days late, rushed before customer calls, or missing critical context. Meanwhile, the actual truth of what changed lives inside the git diff.

GitChangeLog was built as an experiment: what if we feed the raw git commit and code diff to an LLM prompted with domain classification, and have it automatically generate three clean tiers: Code Changes, Feature Updates, and Bug Fixes?

The result is living knowledge that updates directly into Notion or Obsidian, keeping everyone aligned without burdening engineers with manual documentation.