The Redact agent detects personal and sensitive content in documents and redacts it automatically in the PDF — suitable for GDPR and compliance requirements, for example when sharing with third parties or for long-term archives.
Prepare workspace and documents
Choose the workspace and document types to be anonymized. Check the storage location and permissions — the agent works on the contents of the selected workspace.
Create the Redact agent
Open Document AI → Agents and create a new agent of type Redact. Give the agent a descriptive name so that later batch runs remain clearly attributable.
Choose scenario and detection patterns
Select the redaction scenario that fits your use case and define the detection patterns — for example persons and names, addresses, account and ID numbers. Only active patterns will be detected and redacted later.
Configure the output
Define how redacted areas are rendered in the PDF and in which form the result is produced. These settings apply uniformly to all documents in the run.
Test run with samples
First run the agent on a few documents. Check the samples in Document Edit: are all sensitive areas covered? If a pattern is missing, add it to the agent and repeat the test.
Start the batch and approve
Once the test is clean, start batch processing across the desired documents. Check samples again afterwards, and only then approve or export the redacted documents.
Automate
Combine the Redact agent with document types and storage mounts so new uploads are protected automatically — new documents then pass through redaction without a manual trigger.
