The problem

We ship to production twice a week. Every release meant engineers and GTM teammates had to coordinate branches, cherry-pick last-minute bug fixes, check staging builds, and write release notes and docs on a tight deadline.

Three things kept slowing us down:

  1. Rebuilding branches when fixes came in late: Whenever a bug fix needed to get into an active release, manually cherry-picking commits or rebuilding the release branch from scratch wasted engineering time.
  2. Scattered coordination: QA sign-offs, status checks, and deploy triggers were spread across different tools and threads.
  3. Feature flags leaking into public docs: Some merged PRs were hidden behind feature flags and weren't live for customers yet. If you summarize every merged PR automatically, you risk announcing unfinished features in public release notes.

What we built

I automated the entire release process so everything happens right inside our #release channel in Slack:

1. Release Cut Auto 2x/Week 2. Auto Cherry-Pick Tag PR → Merge Fix 3. Slack Deploy 1-Click Go Live 4. Flag-Safe Docs Checks Code Flags
  1. Automatic release kickoff twice a week: The release starts automatically on schedule, cuts the release branch, and posts a status thread in our Slack release channel.
  2. Tagging a PR to cherry-pick it: If an engineer needs to add a fix to the current release, they just add a tag to their PR. The workflow cherry-picks that PR and merges it into the release branch automatically, so we don't have to rebuild the whole release.
  3. One-click deploy in Slack: Once QA testing is done and the build is stable, anyone on the release team can push the release live to production by clicking a button in Slack.
  4. Checking feature flags in code before updating docs: Right after the release goes live, the agent checks the merged PRs against our feature flag configs in code. Anything still hidden behind a feature flag gets skipped, and only the features that are actually live for users get added to our public docs and release notes.

The result

  • Saves ~2 hours on every release (~4 hours a week): Cut out manual branch work, cherry-pick wrangling, and release-note writing twice a week.
  • Docs stay up to date without leaking hidden features: Because the pipeline checks feature flags in code first, our public docs and changelog update right after every release without ever exposing unfinished work.