Ability to detect any custom patterns in outgoing messages. From internal to third parties. Should allow either alerting or auto redacting/blocking and capability to add custom patterns and policies. For example configure action according to detection criticality and ability to configure what we consider low/medium/high severity detection.