Redact information in documents

The redacting functionality allows you to automatically detect and redact information directly within uploaded documents. This feature helps you protect sensitive data when sharing or processing documents that contain personal information.

Intric employs a hybrid detection approach to identify personal information within your documents. By combining two distinct models, the system ensures both contextual understanding and pattern-based accuracy:

1. Assistant model: Conducts a comprehensive, top-level scan of the entire document to understand context and identify general areas containing sensitive data.
2. Dedicated model: Specifically engineered and trained to recognize intricate formatting patterns associated with personal information (such as identification numbers, phone numbers, or addresses).

Before these models can begin the analysis, the document undergoes a mandatory Optical Character Recognition (OCR) phase. This ensures the system fully understands the text layout and content, allowing the models to scan the file effectively.

When to use PDF redacting

Use PDF redacting when you need to:

• Remove personal information before sharing a document.
• Create redacted versions of documents for compliance purposes.
• Process sensitive documents while protecting individual privacy.
• Prepare documents for broader distribution within your organization.

How to redact a PDF

1. Upload your PDF to Intric

• Upload a PDF file in any conversation.
• Wait for Intric to process and analyze the document (OCR).
• Once processing is complete, click on the PDF to open it in the side panel as an artifact.

2. Configure and apply redacting

• Click the settings icon (cogwheel) next to the Redact button to configure the redacting process.
• Adjust your redacting preferences as needed.
• Click Redact to start the automatic detection and redaction process.
• Wait for the redacting to complete (processing time depends on document length and complexity). You can track the analysis via the progress bar in the viewer.

3. Download your redacted PDF

Once redacting is complete, click the download button to save the redacted PDF to your device. The redacted version will have all detected personal information redacted.

LLM configuration and requirements

To ensure the redacting feature functions correctly, please note the following model requirements within your tenant:

• Assistant model dependencies: The top-level scan is performed by the active assistant’s selected model. Make sure to enforce adequate security classification before use.
• Built-in dedicated model: The pattern-recognition model is an internal Intric service and is enabled by default for all tenants. It does not require any specific third-party AI model toggles to be active and it’s hosted in Sweden by default or in your on-prem environments.
• OCR engine: The initial OCR processing is a core platform feature and is not dependent on your specific AI model settings.

Availability

PDF redacting is available to all users whenever a PDF is uploaded in a conversation. The redacting button will appear automatically once Intric has successfully processed and understood your document.

Best practices

• Review your work: Always review the redacted PDF to ensure all sensitive information has been properly redacted.
• Format matters: Remember that redacting works best on text-based PDFs with clear, readable content.
• Quality counts: For best results, upload high-quality PDFs where text is easily recognizable.
• Proactive protection: Ensure relevant security classification is enforced and access control via Spaces is confirmed before processing.