Omissia, the local redaction desk for AI-bound documents

Field notes

The workflow is built around reviewable detections rather than automatic blanket replacement: the user sees each value in its document context and decides what to remove. The example covers eight findings in a legal order, including a tax code, date, IBAN, amount, street address, and two names. The site also says the placeholder mapping remains encrypted on the computer so an AI response can be restored locally.