Local document anonymization in Public Beta
Built for teams that want to find and redact personally identifiable information (PII) locally, with an optional AI assistant for document insight and residual PII review.
Choose your platform. Mac and Linux are marked as coming soon.
How it works
Three simple steps to protect your documents.
Upload
Drag and drop or select your documents.
Detect
Preview findings and tune your rules.
Review and export
Run processing, review the results, and export output.
Upload, detection, and review
Watch the short flow and use the steps as a checklist before sharing output.
- 1Upload documentChoose a DOCX, text-based PDF, or another supported file.
- 2Choose detectorsSelect the personal information types to find and redact.
- 3Process and reviewRun local processing, open the output, and review it before sharing.
Powerful features
Practical tools for local personal information redaction, document insight, and review.
Personal information detection
Pattern- and language-based detection helps identify personal, personally identifiable, and sensitive information before review and export.
Anonymization and pseudonymization
Use rules-based masking, redaction, and deterministic pseudonymization to reduce exposure of personal and sensitive information.
Destructive redaction
Redact identified content in the output and review the result before documents are shared.
Batch processing
Upload folders, monitor progress, and download results easily.
Review-first workflow
Preview findings, adjust rules, and review output before documents are shared or published.
Local AI assistant
Get confidential document insight locally and help identify possible residual PII after rules-based redaction.
See it in action
A quick look at Files, Batch, and Settings pages.
Guides and resources
Short practical guides for local personal information redaction and review-first workflows.
PDF redaction with local review
Text-based PDF, manual review, and Public Beta limitations.
Read guideRemove personal information from documents
Rules-based detection, local processing, and review before sharing.
Read guideDanske guides
Danske ressourcer til redaktion af personoplysninger i dokumenter.
Read guideTest results and accuracy
Controlled tests show the current status for DOCX, text-based PDFs, and scanned PDFs.
These figures measure manifest-based known PII removal rate: how many known personally identifiable information (PII) occurrences are removed from synthetic Danish and English benchmark documents. Results depend on file type, language, layout, scan quality, and document content.
DOCX
Strongest benchmark287 of 300 known PII occurrences were removed in the latest compiled DOCX benchmark outputs. DOCX is currently our strongest and recommended format, but output should still be reviewed before sharing.
PDF text
Manual review250 of 300 known PII occurrences were removed in the latest compiled text-based PDF benchmark outputs. Text-based PDF is supported, but requires manual review because residual PII remained in some outputs.
PDF scan
Work in progressOCR and scan redaction work technically on the test files, but scanned PDF output has not yet been verified with full visual/OCR-based recall. Scanned PDFs should therefore remain marked as experimental and require manual review.
Language testing
English and Danish are the currently tested languages for the published test results.
Additional languages are planned as roadmap candidates and will only be presented as supported after testing is complete.
Tailored for your industry
Anonymizer adapts to regulated and data-heavy workflows.
Compliance and legal
Support review of personally identifiable information for regulatory submissions and legal workflows.
Healthcare records
Prepare medical records and clinical datasets for manual review with privacy risk in mind.
Finance and customer data
Protect documents with account numbers and transaction histories.
Research and academia
De-identify datasets for review before collaboration and publication.
Designed for local processing
Anonymizer is built to process files locally and help teams reduce privacy risk.
Offline-capable anonymization flows and diagnostics.
Desktop-ready architecture with local backend and frontend components.
Optional separate workspace path for local processing, including organization-managed encrypted volumes.
Simple, Local Pricing
Choose the capacity that fits your workflow. New users receive free credits to test before subscribing.
Personal
For individuals and light document workflows
- ✓500 credits / month
- ✓1 credit = 1 page
- ✓Unlimited re-runs of the same file
- ✓Batch limit: 10 files
- ✓1 device license
Professional
BEST VALUEFor power users and small firms
- ✓5,000 credits / month
- ✓1 credit = 1 page
- ✓1 device license
- ✓Full folder-level batch processing
- ✓Advanced local AI chat
Enterprise
For teams and organizations
Contact us for team licensing, offline setups, or deployment questions
- ✓Unlimited pages
- ✓10+ Device Licenses
- ✓Advanced local AI chat
- ✓Air-Gap (Offline) Support
- ✓Priority Integration Support
Frequently asked questions
Can't find the answer? Reach out to us.
How does Anonymizer process my files?
Files are processed locally on your machine.
Which file formats are supported?
TXT, CSV, TSV, XLSX, DOCX, and PDF.
Where are uploaded and generated files stored?
By default, uploaded and generated files are stored locally on your own device in temporary storage. Automatic cleanup is enabled by default with a 30-day retention period, which you can change, and you can also trigger immediate cleanup manually. Anonymizer can also be configured to use a separate local workspace path, for example on a managed encrypted volume.
Do I need internet for the AI assistant?
The optional local AI assistant is available on Professional and Enterprise and runs locally. Professional requires internet for monthly license validation, while Enterprise can be delivered for offline or air-gapped operation.
What system requirements apply to the local AI assistant?
The core redaction, anonymization, and pseudonymization features generally have modest system requirements. The optional local AI assistant, which is available on Professional and Enterprise and can run through Ollama, performs best on newer and more capable hardware. It can still be used on less powerful systems, but response times will typically be slower, especially with larger or more demanding models.
Can I customize detection rules?
Yes. You can create custom patterns and reusable templates.
Is there a batch processing limit?
Batch is designed to scale from one file to large sets.
Should I review output?
Yes. Anonymizer is in Public Beta and primarily uses rules-based redaction and language-based detection with an optional local AI assistant for review. Output may miss sensitive information or redact non-sensitive content and should always be reviewed before sharing or publishing.