Featured on
Upload a PDF, a Word file or a scan: Redactic finds names, tax IDs, IBANs and addresses, then review everything and download a copy with no personal data, ready to share.
Designed for legal, HR, and administrative professionals who handle sensitive information and need a simple, secure way to redact documents.
Redact personal data in client files and contact lists before sharing them or using AI.
Bring AI into the firm without exposing client names.
Redact tenant files, payslips and bank statements under GDPR.
Mask employee names and tax codes in payslips and payroll exports before AI.
Redact psychological and expert witness reports before disclosure.
Redact deed copies, annexes and identity documents before they leave the office.
Redact decisions, access-request files and tender documents before they are published or shared.
Hidden sheets and comments included
Anonymise CVs and candidate lists, and give HR an AI that receives placeholders.
Contracts, payslips and resumes contain personal information. Redactic helps you detect and review sensitive values before applying permanent redaction or reversible masking. Pseudonymised content may remain personal data, and redaction alone does not establish that a document is anonymous.
Redactic hides names, tax codes, and addresses while keeping the rest of the document intact.
This allows you to safely share files or use AI to analyze them without exposing personal information. You can always review what gets masked before proceeding.
Documents are never stored. See Trust & Security for the processing boundary.
Add a PDF, Word, Excel, or image, or paste text. We don't save your files.
We automatically detect sensitive information like names and IDs. You review and select what to mask.
Download the redacted file to share it safely. Or process it by sending the redacted data to an AI model, and your device shows it back to you in the clear.
Choose permanent redaction, reversible masking or a separate optional AI workflow. Uploaded documents are processed by the Redactic service in Europe.
Replace detected sensitive values with placeholders and receive a mapping for restoration in supported workflows. Detection and redaction do not send unredacted content to third-party AI or machine-learning inference services.
Apply permanent redaction to selected sensitive information in supported documents, including scans and images. Review the exported file for missed values, contextual identifiers and format-specific limitations before sharing.
Choose an optional AI task after masking detected sensitive values. Before the document goes to the AI model, you see the masked text with its placeholders and can change what is masked. The provider does not use the text for training. Review it first: missed values and contextual details can still identify individuals.
AI-safe documents →Use an optional chat workflow with a third-party AI provider. Masking helps reduce exposure of detected sensitive values; contextual information and missed identifiers may remain.
Choose the right tool for your specific use case.
Replace sensitive data with placeholders before using AI. Restore the original values in the final output.
Permanently black out data directly on the document. Ideal for securely sharing files with third parties.
A spreadsheet keeps personal data in places a quick look misses, such as a hidden sheet or a comment left by a colleague, and the author's name is saved in the file properties. Redactic reads every sheet, the note cells and the cell comments, then lists what it found, so you check a payroll export or a customer list in the same review screen you use for contracts. One upload covers the whole workbook, and you no longer unhide sheets and open comments one at a time.
When you mask a workbook, each person keeps one label across all sheets, notes and comments, so whoever reads the file, an AI model included, can still tell which rows belong to the same person. Masking is reversible, and the real values can be put back later. Blackout removes the values for good. In both modes the workbook keeps its sheet names, its hidden sheets, its formatting such as bold text and its formulas, and a final check confirms that no detected value is left in any cell before the file is released.
CSV files are handled the same way, whether the columns are separated by commas, semicolons or tabs. That includes the semicolon format that European versions of Excel export. Each cell is masked on its own, the columns stay as they were, accented letters and symbols such as € are preserved, and the downloaded file opens directly in Excel.
Use it when a payroll or HR spreadsheet goes to a consultant, or when you want an AI tool to analyse a customer export with the names and IBANs replaced by labels.
Your collaborators are already pasting contracts, customer emails, and personal details into public chatbots. This shadow AI is one of the hardest risks to govern, because it happens quietly and outside any policy.
Redactic offers a separate, optional chat workflow that masks detected sensitive values before sending prepared content to the selected provider. Automated masking can miss information, and context may still identify individuals. Review sensitive material and follow your organisation's AI policy.
In the AI chat, the AI model receives only masked text, and the provider does not use it for training. Provider availability and retention terms depend on the selected AI workflow and applicable agreement. Consult the Redactic DPA and discuss requirements with us.
Available on FIRM and Enterprise plansEverything you need to redact documents and securely interact with AI.
Temporarily mask sensitive data for AI processing, then restore it in the final output.
Choose an optional AI workflow using content prepared by masking detected sensitive values. Review remains important because contextual information may still identify individuals.
Irreversibly black out sensitive data before sharing. Includes OCR support for scanned PDFs and images.
Download a redaction certificate for every operation, including document hashes and a summary of removed data.
Comments, tracked changes, hidden sheets, document properties and photo location data are cleaned too, and a final check reads the whole file again before you download it.
Review all detected entities and manually decide what to mask or keep before proceeding.
Detects country-specific identifiers like tax codes and VAT numbers, alongside standard contact details.
Process PDFs, Word, Excel, images, and raw text while maintaining the original layout.
Process spreadsheets to redact specific columns or cells while preserving the table structure.
A payroll export or a client list carries personal data in every row, and some of it sits in places nobody opens. Redactic reads every sheet of an Excel workbook, hidden sheets and cell comments included, clears the author fields and gives the same person the same label everywhere. CSV exports work too, whether the columns are separated by commas, semicolons or tabs, and each column stays where it was.
Excel redaction →Automated detection helps identify sensitive information. You decide what gets redacted; detection can miss values and depends on language, document quality and context.
Country formats: FranceCheck-digit detection where supportedNIR, SIREN, SIRET, numéro fiscal, RIB, TVARecognized by format and contextpassport, GermanyCheck-digit detection where supportedSteuer-IdNr, USt-IdNr, Sozialversicherungsnummer, ItalyCheck-digit detection where supportedCodice Fiscale, Partita IVARecognized by format and contextTessera Sanitaria, SpainCheck-digit detection where supportedDNI, NIE, NSS, CIF, PolandCheck-digit detection where supportedPESEL, NIP, REGON, dowód osobisty, CzechiaCheck-digit detection where supportedrodné číslo, IČO, DIČ, číslo účtuRecognized by format and contextobčanský prúkaz, cestovní pas, RomaniaCheck-digit detection where supportedCNP, CUI/CIF, ONRCRecognized by format and contextcarte de identitate, SloveniaCheck-digit detection where supportedEMŠO, davčna številka, matična številka, ID za DDV, SwedenCheck-digit detection where supportedpersonnummer, samordningsnummer, organisationsnummer, DenmarkRecognized by format and contextCPR-nummerCheck-digit detection where supportedCVR-nummer, momsnummer, NorwayCheck-digit detection where supportedfødselsnummer, D-nummer, organisasjonsnummer, FinlandCheck-digit detection where supportedhenkilötunnus, Y-tunnus, ALV-numero, United StatesCheck-digit detection where supportedABA routing number, NPIRecognized by format and contextSSN, ITIN, EIN, Medicare MBI, driver's license, United KingdomCheck-digit detection where supportedNHS number, UTRRecognized by format and contextNational Insurance number, sort code and account number, driving licence, postcode, the Netherlands, Belgium, Switzerland and Austria (when you select the country), Ireland, Canada, India, Singapore and South Africa.
Redactic checks the digits it can verify, so an IBAN, a French NIR, a German tax ID, an Italian codice fiscale, a Spanish DNI or a Polish PESEL is recognised by its check character, and in the United Kingdom it also recognises National Insurance numbers, NHS numbers, UTRs and postcodes.
Working under US privacy rules or UK GDPR? Redactic has pages for both: PII redaction software for US firms and redaction for UK GDPR and subject access requests.
Need a type that is not listed? The model recognizes new categories on request, so coverage can be extended to the documents and identifiers your work depends on.
Technical and legal safeguards to ensure your data remains secure.
Document detection and redaction run on Redactic-controlled infrastructure in Europe. This supports data-minimisation workflows; customers remain responsible for lawful processing.
Documents are never stored and never used to train AI models. When you use AI on a document, the AI model receives only the masked text, and the provider does not use it for training. Account, billing and operation records are handled separately. Consult the Redactic DPA and privacy notice for their retention terms.
Document detection and redaction do not submit unredacted content to external AI inference services. Optional AI provider terms apply to the workflow you select.
Generate certificates with document hashes and redaction summaries for your compliance records.
Reversible masking uses a mapping to restore detected values in supported workflows. Temporary server-side processing can be involved; protect mapping files as sensitive information.
For customer document processing, we act as processor under the Redactic DPA. Customers determine the lawful basis and purpose. Account and billing processing are described in the privacy notice.
Masking and permanent redaction do not consume AI credits. Document upload allowances and per-document limits depend on your plan. Optional AI tasks consume credits based on usage.
Tailored volumes, AI-credit quotas per user and unit, an optional dedicated EU server and on-premise on project.
You only use credits when you run AI tasks within Redactic. Document masking, permanent blackout, and certificate downloads never consume credits.
You always see how many credits you have left and what an operation costs, before you run it.
AI credit usage depends on the selected model and task. Region and retention terms for the applicable workflow are in the Redactic DPA and agreement.
No. Documents are never stored and never used to train AI models. When you use AI on a document, the AI model receives only the masked text, and the provider does not use it for training. Account, billing and operation records are handled separately. Consult the Redactic DPA and privacy notice for their retention terms.
Permanent redaction removes selected sensitive information from the exported document. Reversible masking replaces detected values with placeholders and uses a mapping for restoration in supported workflows. Neither automatically establishes GDPR anonymisation; review the output and remaining context.
No. Detection and redaction work separately from the AI chat and AI tasks. Before a document goes to the AI model, you see the masked text with its placeholders and can change what is masked. Review the masked text and the provider terms first.
When you use AI on a document, the AI model receives only the masked text, and the provider does not use it for training. Retention terms depend on the workflow and applicable agreement. Masked text may still contain personal data through context or missed identifiers.
Yes. It recognizes the text in scans and images, masks it and, if you only want to share, blacks it out directly on the page.
Yes. Redactic cleans footnotes, comments and their authors, tracked changes, headers and footers, hidden sheets and cell notes, document properties, the location data in photos and embedded files. In a PDF it also removes annotations, attachments, bookmarks, links and hidden layers. A final check reads the finished file again and stops the job if a value is still there. You can also give the downloaded file a neutral name.
For standard SaaS, your documents are uploaded to Redactic and detection and redaction run on Redactic-controlled infrastructure in Europe. Unredacted content is not sent to third-party AI/ML inference services for these operations. Optional AI workflows have separate provider terms.
It depends on how long the document is. You can always see how many credits you have left. If they run out, you buy a pack whenever you want, or use your own AI on the masked text without spending credits.
For customer document processing, you determine the purpose and lawful basis as controller, and we act as processor under the Redactic DPA. Account and billing processing are described separately in the privacy notice.
Redactic supports GDPR-aligned data-minimisation workflows. You remain responsible for lawful processing, notices and assessing the output. Automated detection depends on document quality and context and can miss sensitive information.
Yes. On-premise and customer-controlled deployments are available as custom Enterprise projects. Architecture, integrations, security, support and optional AI requirements are scoped individually. This is not a standard self-service plan.
No. Detection runs on our own servers in the EU, and unredacted content is never sent to third-party AI services. A machine learning model recognises names, organisations and addresses, and rules that verify check digits find tax codes, IBANs, identity numbers and other structured identifiers. The AI chat and AI tasks are separate from detection, and the AI model there receives only masked text.
Upload a file, select the data to hide, and safely share it or process it with AI.
Get started →