Right, in terms of redacting. Very concretely speaking, only this year that machine still has humans in turning speech into text. Maybe in a year or two, optical character recognition will also catch up, even on historical everything, documents.