Holocaust Archive Project Builds AI-Searchable Corpus With Git-Based Editorial Workflow

aloncarmel · x · 2026-09-11

Alon Carmel describes Holocaust IO: 652 more books are queued for the corpus, and the goal is simple — anyone searching their grandmother's surname should find it and see exactly which page it appears on. Launch is near; they just need money for tokens.

He highlights that the archive lives on git: every edit is a commit on a branch, and every research session closes as a pull request bundling all sources gathered along the way. A human editor reviews and merges before anything publishes, giving full, transparent, revertible history. On material like this, a silent error is the worst possible outcome.

Related event: Holocaust Archive Project Uses AI to Rebuild Memory Books with Human Oversight(3 posts)→

Original post →

More from Apps

Apps channel →