Reproducible benchmarks

How promptShield compares to other PII detection tools

No marketing fluff. No hand-waving. We publish the script, the test documents, and the raw CSV output so you can verify everything yourself.

View the benchmark repo

Microsoft Presidio

Available

Open-source Python toolkit from Microsoft. Excellent for developers; we extend the baseline with per-country checksums, entity post-filters, and a finished desktop experience.

See the comparison

Amazon Macie

Available

AWS managed service for discovering sensitive data in S3. Discovery and classification, not redaction.

See the comparison

Google Cloud DLP

Available

Google Cloud's de-identification API, with 200+ infoType detectors and reversible tokenization.

See the comparison

Microsoft Purview

Available

Microsoft's data governance platform: classification, sensitivity labels and policy across Microsoft 365.

See the comparison

Our methodology commitment

Every comparison follows the same rules: competitor's default install, same PDFs, same text extraction, same metrics. If you spot a methodology issue, open an issue — we publish every correction.

  • ✓14 synthetic PDF documents across 7 European languages (en, fr, de, es, it, nl, pt)
  • ✓Same text-extraction library (pypdfium2) for both systems
  • ✓Competitor in default-install configuration — no asymmetric tuning that would skew the comparison
  • ✓Raw CSVs published, dependency versions pinned, measurement dates timestamped

Want to see promptShield compared to another tool? Get in touch or open an issue on GitHub.