FR EN
Join theBetaJoin the Beta

Formats

Enterprise-ready formats

Safe-Doc is built for sensitive documents teams actually use: contracts, reports, legal case files, HR and finance documents. The goal isn't "document parsing" - it's secure AI usage on these files.

What teams really share with AI

PDF and Word cover most sensitive enterprise workflows. Safe-Doc pseudonymizes or anonymizes, reduces risky context, and produces a copy-ready version for secure AI analysis (single doc or Data Room).

PDF (native text) Full support

Typical documents

  • Contracts and appendices
  • Reports, audits, memos
  • Legal case files
  • Finance documents

What Safe-Doc does for AI

  • Anonymization or pseudonymization of sensitive information
  • Metadata removal (author, tools, timestamps)
  • Context precision reduction (dates, amounts, locations)
  • Invisible layer removal
  • Clean document reconstruction
  • Audit summary (depending on level)

Advanced mode

  • Stronger context reduction and generalization
  • Advanced reconstruction (cleaned structure)

Note: Scanned PDFs require text extraction (OCR).

DOCX (Microsoft Word) Full support

Typical documents

  • Contracts and clauses
  • Internal memos
  • HR procedures and case files
  • Legal notes and summaries

What Safe-Doc does for AI

  • Anonymization or pseudonymization of sensitive information
  • Author properties removal
  • Comments removal
  • Revision history removal
  • Internal metadata cleanup

Advanced mode

  • Clean rebuild
  • Deeper structural cleanup

Limitation: Legacy .doc format is not supported at this stage.

OCR (option)

For scanned PDFs (text inside images):

  • Text extraction
  • Then pseudonymization or anonymization and context reduction
  • Same copy-ready output for secure AI analysis

Limitation: Quality depends on the original scan.

90+ entity types detected - multinational coverage

Safe-Doc detects and can pseudonymize or anonymize 90+ categories of sensitive information - a coverage level rarely published openly by competing solutions.

👤 Identity

PERSONNE, NOM_NAISSANCE, PSEUDO, CARTE_PROFESSIONNELLE, NUMERO_TOQUE, RPPS_ADELI, NATIONALITE, ORIGINE, GENRE, SITUATION_FAMILIALE, SIGNE_PHYSIQUE, COULEUR_YEUX, TAILLE_POIDS, DONNEE_BIOMETRIQUE, GROUPE_SANGUIN

🧑‍🏭 Job & role

PROFESSION

📞 Contact details

ADRESSE, CODE_POSTAL, LIEU, LIEU_NAISSANCE, PAYS, COORDONNEES_GPS, REFERENCE_CADASTRALE, TELEPHONE, EMAIL, URL

💳 Banking details

IBAN, BIC, RIB, NUMERO_COMPTE, CARTE_BANCAIRE, CARTE_BANCAIRE_FRAGMENT, CVV, DATE_EXPIRATION_CARTE

💰 Amounts & finance

MONTANT, DONNEE_FINANCIERE, CRYPTO

🪪 Official documents

NUMERO_SECU, CARTE_VITALE, NUMERO_AFFILIATION, CARTE_ID, PASSEPORT, PERMIS_CONDUIRE, AGDREF, INE, NUMERO_REGISTRE_ETRANGER, NUMERO_ELECTEUR, PLAQUE_IMMAT, VIN, CARTE_GRISE, NUMERO_FISCAL, NUMERO_URSSAF, NUMERO_DSN

🔢 Identifiers & references

MATRICULE, NUMERO_BADGE, IDENTIFIANT_UNIQUE, IDENTIFIANT_ADMIN, REFERENCE_UNIQUE, NUMERO_PORTALIS, NUMERO_MARCHE_PUBLIC, NUMERO_AGREMENT, NUMERO_HABILITATION, NUMERO_LICENCE, NUMERO_ASSURANCE, NOM_PROJET

🏥 Health & sensitive data

DONNEE_SANTE, NUMERO_PATIENT, MDPH, DONNEE_JUDICIAIRE, OPINION_POL, RELIGION, ORIENTATION_SEX

🏢 Companies

ORGANISATION, SIRET, SIREN, TVA, RCS, CODE_APE, NUMERO_DUNS, NUMERO_LEI, NUMERO_EORI, CONVENTION_COLLECTIVE

💻 Technical & security

MOT_DE_PASSE, API_KEY, COOKIE_SESSION, IP, ADRESSE_MAC, IMEI, IDENTIFIANT_APPAREIL, IDENTIFIANT_PUBLICITAIRE, USERAGENT, NUMERO_SERIE

📅 Dates

DATE

Countries: France, Germany, Spain, Italy, UK, USA.
Indicative list: actual scope depends on the document.

Enterprise roadmap

Goal: cover real enterprise formats without compromising document security.

XLSX (Excel)

Useful for:

  • Finance exports
  • HR datasets
  • Diligence tables

Excel involves complex structures (formulas, links, macros), requiring a dedicated pipeline for robustness and control.

PPTX (PowerPoint)

Useful for:

  • Committee decks
  • Internal presentations
  • Notes and comments
CSV / Data exports

Useful for:

  • Tool exports
  • Sensitive columns and identifiers
  • Structured data generalization
Batch processing (ZIP)

Useful for:

  • Document batches
  • Consolidated reporting

Current limitations

  • Password-protected files must be unlocked before upload.
  • Embedded scripts are removed.
  • Text contained only in images requires OCR.
  • Safe-Doc does not guarantee absolute anonymity; it significantly reduces exposure and re-identification risk through pseudonymization and de-identification.